Memory and Emotion
Series in Affective Science Series Editors Richard J. Davidson Paul Ekman Klaus Scherer The Nature of Emotion: Fundamental Questions Edited by Paul Ekman and Richard J. Davidson Boo! Culture, Experience, and the Startle Reflex By Ronald Simons Emotions in Psychopathology: Theory and Research Edited by William F. Flack Jr. and James D. Laird What the Face Reveals: Basic and Applied Studies of Spontaneous Expression Using the Facial Action Coding System (FACS) Edited by Paul Ekman and Erika Rosenberg Shame: Interpersonal Behavior, Psychopathology, and Culture Edited by Paul Gilbert and Bernice Andrews Affective Neuroscience: The Foundations of Human and Animal Emotions By Jaak Panksepp Extreme Fear, Shyness, and Social Phobia: Origins, Biological Mechanisms, and Clinical Outcomes Edited by Louis A. Schmidt and Jay Schulkin Cognitive Neuroscience of Emotion Edited by Richard D. Lane and Lynn Nadel
The Neuropsychology of Emotion Edited by Joan C. Borod Anxiety, Depression, and Emotion Edited by Richard J. Davidson Persons, Situations, and Emotions: An Ecological Approach Edited by Hermann Brandstätter and Andrzej Eliasz Emotion, Social Relationships, and Health Edited by Carol D. Ryff and Burton Singer Appraisal Processes in Emotion: Theory, Methods, Research Edited by Klaus R. Scherer, Angela Schorr, and Tom Johnstone Music and Emotion: Theory and Research Edited by Patrik N. Juslin and John A. Sloboda Handbook of Affective Sciences Edited by Richard J. Davidson, Klaus Scherer, and H. Hill Goldsmith Nonverbal Behavior in Clinical Settings Edited by Pierre Philippot, Erik J. Coats, and Robert S. Feldman Thinking about Feeling: Contemporary Philosophers on Emotions Edited by Robert C. Solomon Memory and Emotion Edited by Daniel Reisberg and Paula Hertel
Memory and Emotion
Edited by Daniel Reisberg and Paula Hertel
1
2004
3
Oxford New York Auckland Bangkok Buenos Aires Cape Town Chennai Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Sao Paulo Shanghai Taipei Tokyo Toronto
Copyright © 2004 by Oxford University Press, Inc. Published by Oxford University Press, Inc.
198 Madison Avenue, New York, New York 10016
www.oup.com
Oxford is a registered trademark of Oxford University Press
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of Oxford University Press.
Library of Congress Cataloging-in-Publication Data
Memory and emotion / edited by Daniel Reisberg and Paula Hertel.
p. cm.—(Series in affective science)
Includes bibliographical references and index.
ISBN 0-19-515856-3
1. Autobiographical memory. 2. Emotions. 3. Psychophysiology. 4. Psychiatry. I. Reisberg, Daniel. II. Hertel, Paula. III. Series. BF378.A87 M46 2003 152.4—dc21 2003006595
2 4 6 8 9 7 5 3 1 Printed in the United States of America on acid-free paper
T
he study of the interplay between emotion and memory inevitably involves a mixing of perspectives. After all, the study of emotion is an important and sophisticated research enterprise in its own right, and so, too, is the study of memory. But, in addition, each of these topics in turn has been studied from multiple perspectives—the nature of emotion in normal populations and in pathological populations, in children, in adults, and in the elderly; the nature of the biological changes, and then the psychological and subjective changes, that accompany (and perhaps constitute) emotion; the range of emotions common in day-to-day life, and also the range of emotions one experiences only in extreme circumstances. It should be obvious that each of these perspectives on emotion has its own value and asks questions that do not arise within these other contexts. But it should also be obvious that common themes—or points of dissonance—may emerge when we start putting these various perspectives side by side. Unfortunately, though, those common themes often do not come into view: After all, in our field research tends to be published in specialized venues, and so there is less cross-talk among related specialties than one might wish. It is precisely this situation that creates a need for volumes like this one. In this volume, we have brought together some of the leading investigators in the broad area of emotion and memory, deliberately seeking to span as best we could the full range of approaches visible in the current scene. This juxtaposition should provide easy access to any reader, from any perspective, hoping to enter this broad research literature. In addition (and as we’d hoped), the juxtaposition highlighted a number of themes that cross research areas. For example, many authors have proposed that emotion promotes memory, so that materials one has experienced while emotional are better retained than materials experienced while one is calm. Indeed, some evidence suggests that emotional char-
acteristics of an event help to ameliorate memorial deficiencies of one sort or another (in elderly adults, for example, or in individuals diagnosed with schizophrenia). But a broader view of the literature makes it plain that the opposite pattern also appears in some circumstances, with emotion serving to undermine memory—even memory for nonemotional events, particularly in the case of anxiety and depression. How should we think about this contrast? Is one of the effects artifactual? Or (as seems likely) are both effects genuine, with other important variables guiding the direction of the event? And, if the latter, what are those variables? Is the nature of the emotion crucial? The strength of the emotion? The nature of the to-be-remembered materials? The way in which memory is assessed? Each of these proposals arises in one or more chapters in this volume, and each of the proposals has arguments in its favor. Our intention in this preface, therefore, is surely not to settle the question of contrast, or even to suggest that the question should be settled in any straightforward way. Instead, we raise this point simply as a reminder that investigators in this broad domain—despite their very different methods, working assumptions, and forms of data—are often working on closely related questions, so that all may benefit from cross-talk among these various approaches. It is that hope that provides the main impetus for this volume. The volume begins with a contribution by Daniel Reisberg and Friderike Heuer. Reisberg and Heuer celebrate the progress that has been made in our understanding of how people remember emotional events, but also highlight the substantial gaps in our knowledge. A prominent theme in their chapter is that, in a wide range of circumstances, emotion promotes memory for an event’s “central” materials, but seems to have the opposite effect, undermining memory, for details at an event’s “periphery.” Reisberg and Heuer argue, however, that this latter effect may be produced not by emotion itself but by the presence of powerful “attention magnets” found within some (but by no means all) emotional events. From this base, Reisberg and Heuer explore possible limits of this data pattern, with an eye on how different types and different strengths of emotion may influence memory differently. The second chapter, by Tony Buchanan and Ralph Adolphs, explores the neuroanatomy of emotional memory, arguing that the brain mechanisms that encode, consolidate, and retrieve memories may operate differentially in emotional and nonemotional contexts. Specifically, the chapter emphasizes the role of the amygdala in the enhancement of memory for emotional events, largely through the amygdala’s influence over brain areas (including the hippocampus and striatum) during the period of memory consolidation and also during the retrieval of emotional memories. Data are drawn from human studies and from studies of other species, from lesion evidence, and from neuroimaging. The specific effects reported, however, are not uniform, and here, as in other chapters, we see evidence of emotion’s mixed effects on memory. Thus, for example, bilateral amygdala damage seems to eliminate both emotion’s tendency to promote
vi
memory for an event’s gist and also its tendency to undermine memory for an event’s periphery. At the same time, the effects do seem general in other ways, and the chapter argues that the amygdala effects (and, more broadly, emotion’s memory effects) depend on arousal itself, not the emotion’s valence. While these first two chapters acknowledge emotion’s mixed effects on memory, there is no question that these chapters emphasize the ways in which emotional events seem to be remembered better than neutral (but otherwise comparable) events. This emphasis is reversed, however, in the third chapter, by Jessica Payne, Lynn Nadel, Willoughby Britton, and Jake Jacobs. They focus on the biopsychological effects of trauma and how these influence memory. They argue that trauma (and, more precisely, the stress that usually accompanies trauma) has identifiable effects on the hippocampus, impairing both the neuronal structure and the function of this brain region. As a direct consequence, stress (especially uncontrollable stressors) impairs various forms of memory. However, not all memories suffer this effect, and the chapter seeks to explain this point by arguing that the experience of stress (through its impact on the hippocampus) causes stressful events to be recorded in a “fragmented” manner, with the elements of the event not woven into a coherent remembered episode. At the same time, emotion works (via the amygdala) to promote memory for the gist of an event, leading to well-encoded memories for the thematic content of an emotion event, but, again, without the coherent spatio-temporal framework needed to organize the memory (because this framework relies on hippocampal circuits disrupted by stress). In this fashion, the authors seek to explain both the positive effects of emotionality on memory and its negative effects—with an emphasis on differentiable neural structures and, with that, on different types of remembered information. The fourth chapter, by Richard McNally, Susan Clancy, and Heidi Barrett, is also on trauma and, like the Payne et al. chapter, seeks to understand the conditions under which traumatic events are remembered or forgotten. The chapter’s focus, however, is on the frequently espoused belief that all trauma will be forgotten—a belief presumably based on the notion that too much emotion hurts memory. In particular, McNally et al. review the evidence relevant to the debate about the extent to which trauma victims typically repress and then later recover memories of the traumatic event. In examining the reported evidence—from victims of childhood sexual abuse, concentration-camp survivors, war veterans, and alien “abductees”—the authors distinguish among phenomena that have more often been collapsed in discussions of repressed and recovered memories. They invite us to consider that the forgetting of traumatic events can arise from quite “normal” phenomena—such as the absence of rehearsal or the initial lack of attention to some aspects of the event—instead of special mechanisms of motivated repression. Another important feature of the chapter is their review of the laboratory research on forgetting mechanisms employed by people who have suffered through various types of trauma. The pursuit of this line of research
vii
seems certain to shed light on the memory processes that distinguish those who remember from those who forget and, in turn, from those who forget and then remember. Chapter 5, by Colin MacLeod and Andrew Mathews, considers how memory might be influenced by a variety of emotional states and conditions experienced by people with anxiety disorders. This chapter represents a thorough review of research performed with people who describe themselves as generally anxious (without formal diagnosis), as well as with people who have been diagnosed as experiencing generalized anxiety disorder, post-traumatic stress disorder, phobias, obsessive-compulsive disorder, and panic disorder. In the context of research on “mood congruent” memory, one might expect that these individuals will better remember stimuli that “fit” with their anxious thoughts and beliefs; this expectation, however, is clearly challenged by the results that MacLeod and Mathews review. In some cases, anxious people do show evidence of anxietyrelated biases in memory, but the chapter argues that these probably result from special instances of emotional interpretation of events with ambiguous meaning. Under conditions less prone to interpretive ambiguity, anxious people tend not to remember in emotionally special ways. Chapter 6, by Paula Hertel, addresses what is by far the largest research area in the examination of memory in clinical groups—research conducted with depressed or naturally unhappy (but possibly nondiagnosed) people. The chapter theme is the connection between memory phenomena and habits of thought. More so than most anxious people, depressed people ruminate about their troubles. The practiced thought patterns of rumination facilitate memory for emotionally consistent events and interfere with memory for other events. Moreover, as Hertel wants us to understand, these habits take over under conditions of poor cognitive control, the main feature of cognition in depression. Hertel argues that the negative consequences of habitual thinking can be overcome by external control or by the training of new habits of thought. The literature on emotional memory in individuals diagnosed as schizophrenic is much less extensive than the literatures in depression and anxiety, and it is often ignored in collections on cognition and emotion. Recently, however, it has been augmented by the research conducted by Jean-Marie Danion, Caroline Huron, and their colleagues. The chapter here (by Danion, Huron, Lydia Rizzo, and Pierre Vidailhet) first reviews evidence that both memory disturbances and emotional disturbances characterize schizophrenia and then raises questions about their interaction. Danion et al. argue that memory for emotional material operates “normally” in schizophrenic states when emotional aspects of experience are genuinely noted. Of course, often these experiences are not interpreted by a person with schizophrenia in a fashion that accurately reflects the event’s emotionality, with a corresponding memory deficit. But in many cases the emotional characteristics of an event require little controlled attention, and, under these conditions, we should expect to see intact
viii
emotional memory, even in this population disrupted in so many other ways. Finally, in a concluding section of the chapter, Danion and his coauthors make the provocative suggestion that intact emotional processing potentially plays a role in perpetuating delusional cognitions, through constructive memory processes. The eighth chapter continues the emphasis on emotional remembering in special populations, but the population at issue here is not pathological. Instead, Robyn Fivush and Jessica Sales examine emotional memory in children. They also focus on a theme that appears in many other chapters of the volume—the fact that the functional role of emotional remembering often depends on attributes of memories other than their historical accuracy. In particular, Fivush and Sales discuss the ways in which the structure and content of young children’s emotional memories is co-constructed by the children and their parents, a process, they argue, that varies from culture to culture and also depends on the child’s gender. This co-construction is important for many reasons; among them, it is one of the essential means through which children gain a foundation for understanding themselves and their autobiographies. Also crucial here are the ways in which parent-child reminiscing about stressful experiences can guide the child’s understanding of and coping with aversive events. Exploring this latter point demands a comparison of how parents and children reminisce about both aversive and pleasant events and also how they reminisce about both emotionally mild events and emotionally extreme ones, themes that provide another important point of contact between this chapter and other chapters in the volume. In chapter 9, Mara Mather discusses memory at the other end of the lifespan—in the elderly. She first examines age-related changes in emotional processing—in mechanisms of emotional regulation and arousal, for example, considered both psychologically and in terms of their neural mechanisms— and from this base offers an intriguing set of suggestions about how these changes should influence emotional memory. Among other issues, she considers whether the older person’s improved ability to regulate emotion implies that memories should become more emotionally gratifying, as well as whether the emotional qualities of experience might actually protect an individual against the age-related decline in memory. These suggestions are then evaluated in her subsequent review of the current literature on age differences in the effects of emotion on memory. In the end, Mather argues that emotional memory does constitute a “somewhat surprising island of maintained functioning” in a sea of general decline in old age. Chapter 10, by Robin Edelstein, Kristen Alexander, Gail Goodman, and Jeremy Newton, covers topics that have arisen in many other chapters—the memory effects of trauma, emotional remembering by children, the long-term durability of emotional memories, and so on—but covers these themes from the perspective of the legal system, asking how eyewitnesses to crimes remember the events they have observed (or, in many cases, the events in which they were vic-
ix
timized). Of special importance here is the question of whether laboratory findings taken as characterizing emotional memory can be reasonably applied to real-life crime situations. Also prominent in this chapter is the special case of memories for childhood sexual abuse, including memories that are apparently lost and then recovered. The chapter discusses the complexity that arises when one tries to assess these memories and also factors (including a tendency toward dissociation, or various forms of psychopathology) that play a role in determining when a traumatic event will be vividly remembered and when that event will (apparently) be forgotten. Finally, chapter 11, by Robert Kraft, focuses on a particularly important form of emotional memory: memory of the Holocaust by its survivors. Throughout this book, chapters have commented on the memory effects of traumatic or otherwise horrific events; chapters have likewise commented on the process of sharing emotional memories with others—and perhaps sharing again and again. Just as important is the fact of deliberately not sharing memories. How does this sharing (or not sharing) shape the recollection? Archives of Holocaust memories speak powerfully to these issues, and Kraft provides a close qualitative analysis of these memories in order to explore (among other topics) the general characteristics of traumatic memory, how emotion itself is recalled, and how the recall of a memory can lead to the re-experiencing of emotion. Also of interest here is the way in which these emotional memories can shape the survivors’ beliefs and emotions throughout their lives. All of these considerations lead Kraft to proposals about “multiple systems” of memory that in some ways echo, and in other ways may challenge, related multiple-systems proposals offered in several other chapters. We end with the happy task of thanking those who have helped us in assembling this volume. First, we are grateful to Catharine Carlin, our editor at Oxford University Press, who persuaded us to launch this project, and who has been enthusiastically encouraging ever since. We are also grateful to John Rauschenberg, also at OUP, for his labors in putting the volume together. We also thank the Series Editors, Richard Davidson, Klaus Scherer, and Paul Ekman, for their support of this endeavor. Finally, we are grateful to Pat Ullmann, in Trinity University’s Instructional Media Service, for transcribing some figures.
x
Contributors,
xiii
1. Memory for Emotional Events, 3
Daniel Reisberg and Friderike Heuer 2. The Neuroanatomy of Emotional Memory in Humans, Tony W. Buchanan and Ralph Adolphs
42
3. The Biopsychology of Trauma and Memory, 76
Jessica D. Payne, Lynn Nadel, Willoughby B. Britton,
and W. Jake Jacobs
4. Forgetting Trauma? 129
Richard J. McNally, Susan A. Clancy, and Heidi M. Barrett 5. Selective Memory Effects in Anxiety Disorders: An Overview of
Research Findings and Their Implications, 155
Colin MacLeod and Andrew Mathews 6. Memory for Emotional and Nonemotional Events in Depression:
A Question of Habit? 186
Paula Hertel 7. Emotion, Memory, and Conscious Awareness in
Schizophrenia, 217
Jean-Marie Danion, Caroline Huron, Lydia Rizzo,
and Pierre Vidailhet
8. Children’s Memories of Emotional Events, Robyn Fivush and Jessica McDermott Sales
242
xi
9. Aging and Emotional Memory, Mara Mather
272
10. Emotion and Eyewitness Memory, 308
Robin S. Edelstein, Kristen Weede Alexander, Gail S. Goodman, and Jeremy W. Newton 11. Emotional Memory in Survivors of the Holocaust:
A Qualitative Study of Oral Testimony, 347
Robert N. Kraft Index,
xii
391
Ralph Adolphs, Department of Neurology, University of Iowa, Iowa City, Iowa
Kristen Weede Alexander, University of California, Sacramento, California
Heidi M. Barrett, Department of Psychology, Harvard University, Cambridge,
Massachusetts
Willoughby B. Britton, Department of Psychology, University of Arizona, Tucson,
Arizona
Tony W. Buchanan, Department of Neurology, University of Iowa, Iowa City,
Iowa
Susan A. Clancy, Department of Psychology, Harvard University, Cambridge,
Massachusetts
Jean-Marie Danion, Unité INSERM 405, Hopital Universitaire, Strasbourg, France
Robin S. Edelstein, Department of Psychology, University of California, Davis,
California
Robyn Fivush, Department of Psychology, Emory University, Atlanta, Georgia
Gail S. Goodman, Department of Psychology, University of California, Davis,
California
Paula Hertel, Department of Psychology, Trinity University, San Antonio, Texas
Friderike Heuer, Lewis and Clark College, Portland, Oregon
Caroline Huron, INSERM, Paris, France
W. Jake Jacobs, Department of Psychology, University of Arizona, Tucson, Arizona
xiii
Robert N. Kraft, Department of Psychology, Otterbein College, Westerville, Ohio
Colin MacLeod, Department of Psychology, University of Western Australia,
Perth, Australia
Mara Mather, Department of Psychology, University of California, Santa Cruz,
California
Andrew Mathews, MRC Cognition & Brain Sciences Unit, Cambridge, United
Kingdom
Richard J. McNally, Department of Psychology, Harvard University, Cambridge,
Massachusetts
Lynn Nadel, Department of Psychology, University of Arizona, Tucson, Arizona
Jeremy W. Newton, Department of Psychology, University of California, Davis,
California
Jessica D. Payne, Department of Psychology, University of Arizona, Tucson,
Arizona
Daniel Reisberg, Department of Psychology, Reed College, Portland, Oregon
Lydia Rizzo, Laboratoire Universitaire de Psychologie, Metz, France
Jessica McDermott Sales, Department of Psychology, Emory University, Atlanta,
Georgia
Pierre Vidailhet, Unité INSERM 405, Hopital Universitaire, Strasbourg, France
xiv
Memory and Emotion
This page intentionally left blank
1
C
ollege graduation. The death of a pet. A particularly romantic evening. A moment’s reflection will persuade most people that events like these—events that were emotional when they occurred—provide many of their most vivid, most detailed, most compelling autobiographical memories. For many of us, these emotional memories are a basis for engaging and sometimes enjoyable reminiscence. More important, though, these memories also matter for us in important ways. After all, these emotional events are likely to have been consequential for us in one way or another, and are, in most cases, closely linked to issues, goals, or people that we care deeply about. It seems certain, therefore, that the recollection of these crucial events is likely to shape our sense of who we are and also our broad perceptions of the world. This, in turn, will influence many aspects of our actions, perceptions, and beliefs. In addition, many of these emotional events are consequential for other people, and so we may be called on to report the event in one setting or another; when that happens, it is of obvious interest to ask how full and accurate that report will be. These broad issues provide the motivation for our inquiry into the nature of autobiographical recollection. As we will see in this chapter, research in this domain has made considerable progress, but on some key issues, the available data are still rather sparse. Hence, this chapter simultaneously celebrates our advances and highlights the points on which more work is urgently needed. 3
The Accuracy
of Emotional Memories
Setting the Issue: Vividness Versus
Memory Accuracy
There is no question that emotional memories tend to be quite vivid, and, in one early study, we reported a correlation of .71 between participants’ ratings of their memories’ vividness and their ratings of how emotional the original event had been (Reisberg, Heuer, McLean, & O’Shaughnessy, 1988). Interestingly, this relationship was observed no matter what emotion was attached to the event, so the correlation between vividness and emotionality ratings was .89 for sad events, .68 for angry events, .90 for fearful events, and .71 for happy events (Reisberg et al., 1988, Experiment 2). It also didn’t matter whether the event being recalled was a personal one (e.g., death of a parent, graduation from college) or a public one (e.g., first moon landing). The personal events were more vividly recalled and were more emotional than the public events, but the relationship between vividness and strength of affect was the same for both (.661 and .641, respectively). Other studies confirm this pattern, observing extreme memory vividness both for memories of traumatic events and memories for extremely positive events (e.g., Pillemer, Goldsmith, Panter, & White, 1988; Rubin & Kozin, 1984). One recent illustration of this finding is provided in Walker, Vogl, and Thompson (1997), who found that the rated emotionality of an event was consistently a strong predictor of whether participants believed they could remember the event or not, and the same data pattern emerged when participants were recalling extremely pleasant events and when they were recalling extremely unpleasant ones. Likewise, Porter and Birt (2001) invited their participants to describe their most traumatic autobiographical memory and also their most positive emotional experience. In both cases, they found that strong emotion was reliably associated with a high degree of memory vividness. Similarly, Berntsen (2001) examined the highly vivid involuntary (spontaneously arising) memories often called flashbacks; the data showed little difference between involuntary memories for trauma and those for extremely happy events. But are these emotional memories accurate? It would be easy to assume they must be, since, after all, an emotional event is likely to be important to us, virtually guaranteeing that we will pay close attention as the event unfolds. In addition, emotional events are often emotional precisely because they are related to issues we care about and have thought about in other contexts; this will foster the sort of memory connections that we know promote retention and recall. Moreover, we tend to mull over emotional events in the minutes (or
4
hours) following the event, and this is tantamount to memory rehearsal (cf. Bower, 1992). Finally, we are likely to revisit emotional events periodically, in our own thoughts or in conversations with others, and this too should promote retention. Even with these considerations, however, it is certain that at least some of our emotional memories do contain errors, and some may be wholly mistaken. Plainly, therefore, the accuracy of emotional memories must be tested and cannot be taken for granted. Evidence relevant to this point comes from many sources, but most forcefully from the debate over “flashbulb memories,” a debate prominent in the research literature in the 1980s and 1990s. The term “flashbulb memory” was coined by Brown and Kulik (1977), and referred to the exceptionally clear and detailed recollection people seem to have for singular, emotional, and consequential events they have experienced. Brown and Kulik offered the example of people’s memory for John Kennedy’s assassination, an event that was still remembered “as if it were yesterday” many years after the event; more current examples would include the memory that many people have for Princess Diana’s death (in 1997), the O. J. Simpson trial (in 1995), and the destruction of the World Trade Center (in 2001). We will return to the topic of flashbulb memories later in this chapter; for present purposes, we wish only to make a simple point about these memories. Flashbulb recollections tend to be extraordinarily vivid and detailed, and these memories are recalled with enormous confidence that the memory is, in fact, correct. But flashbulb memories can be shown in some circumstances to be wrong, making it plain that we cannot equate memory vividness with memory accuracy or memory confidence with accuracy. One compelling illustration of this point comes from Neisser and Harsch (1992), who interviewed people days after the 1986 space-shuttle disaster, to determine where they were when they heard the news of the disaster, who brought them the news, and so on. Neisser and Harsch then reinterviewed these same people roughly 3 years later (32–34 months after the initial data collection), to see how memories for the event had fared over this time span. They found that people still reported detailed, high-confidence memories in this 3-year follow-up, but many of these memories were simply wrong, sometimes in important ways, completely misrepresenting the original event. We hasten to say that other studies have yielded rather different results, with impressive accuracy in people’s flashbulb recollection (e.g., Conway et al., 1994), and so, as we have noted, we will need to return to this topic before we are done in order to address the question of why some flashbulb memories seem accurate and long-lasting, while others do not. For now, though, we draw from the Neisser and Harsch data only the moral that emotional memories can be extremely vivid, extremely compelling, and yet completely out of step with the historical facts. This demands that we test the accuracy of emotional memories and not assume it.
5
Studying Emotional Memory
in the Laboratory
Studies of flashbulb memory obviously provide insights into how people remember the genuinely emotional events that actually take place in their lives. The disadvantage of these studies, though, is that we often have no way of knowing exactly what happened within the target event and thus no means of assessing the accuracy of memory. For this reason, the study of memory accuracy has often been taken into the laboratory so that we now have full knowledge of, and full control over, the to-be-remembered event. These laboratory studies have employed a variety of to-be-remembered materials. Here our emphasis will be on memory for events, and so we will hold to the side studies of people’s memory for emotional pictures (e.g., Canli, Zhao, Brewer, Gabrieli, & Cahill, 2000), emotion-laden word lists (e.g., Dietrich et al., 2001; Jones, O’Gorman, & Byrne, 1987), and memory for humor (e.g., Schmidt & Williams, 2001). Even with this narrowed focus, however, the evidence derives from a diversity of studies. In some studies, participants have witnessed an event presented via a series of slides, depicting successive moments within a story, with an accompanying (tape-recorded) narrative telling the emotional tale. In a smaller number of cases, the to-be-remembered event has been presented as a video clip (or, in just a few studies, an animation). An even smaller number of studies have employed live events, witnessed by the study’s participants. In virtually all cases, though, the studies compare participants’ memory for emotional materials to their memory for neutral materials, with the assessment of memory typically provided by a fine-grained test in four-alternative forcedchoice (4AFC) format. The experimental materials are usually claimed to be emotional on three bases: the content itself, specifically chosen (or designed) for its emotional themes; self-reported emotionality from participants viewing the content; and physiological measures of arousal (usually heart rate). The neutral materials are then matched as well as possible to the experimental materials in structure and content but, of course, are claimed not to be emotional (usually on the same three grounds). As an illustration, several different laboratories have used the “doctor/mechanic” stimulus set first developed by Heuer (1987). In both versions of this stimulus, participants see slides and hear a narration about a mother and son going to visit father at work. In the emotional version of the stimulus, the father is a surgeon, and the son watches as his father completes a difficult surgical procedure. One slide for this sequence shows a scene of surgery with the patient’s viscera in plain view; another shows the severed and reattached legs of a child. In the neutral version of the stimulus, the father is an automobile mechanic, and the son watches as his father completes a difficult repair. One slide shows a scene of the repair with the engine in view; another shows the supposedly damaged part.
6
The Easterbrook Hypothesis Most of the laboratory studies of emotional remembering have been cast as tests of two intertwined hypotheses. One hypothesis concerns participants’ memory for the gist or “central” materials within the event. As we will see, the weight of the evidence on this point is reasonably clear: in general, emotion seems to promote memory for an event’s gist or center. A second hypothesis derives from claims offered many years ago by Easterbrook (1959). Based on animal studies, Easterbrook proposed that arousal causes a narrowing of attention, so that an aroused organism becomes less sensitive to information at the “periphery” of an event. As a consequence, the organism is likely to become more sensitive to information at the center of the event, perhaps because of diminished distraction from the periphery, or perhaps because the organism’s attentional resources are more “concentrated” on the event’s center. In either case, the hypothesis implies that the memory advantage associated with arousal will be observed only for the event’s center. Arousal should produce a disadvantage in remembering the event’s periphery (relative to memory for otherwise comparable neutral materials). Easterbrook’s own studies manipulated arousal by depriving animals of food for various lengths of time; his measures then assessed the animal’s sensitivity to cues in its immediate environment. There is obviously some extrapolation needed, therefore, to apply his hypotheses to the situation of a human, aroused through anger or fear or joy, remembering a complex event. Even so, that extrapolation is invited by a pattern often observed among victims or witnesses to crimes, a pattern known as the weapon focus effect. This term refers to the fact that the witnesses to crimes often seem to “lock” their attention onto the criminal’s weapon and seem oblivious to much else in the scene. As a result of this attentional pattern, later on the witness will remember the perpetrator’s gun or knife with great clarity but may remember little else about the crime, including such crucial details as what the perpetrator looked like! Weapon focus has often been alleged by those in law enforcement and has also been documented in a variety of laboratory studies (e.g., Loftus, Loftus, & Messo, 1987; Stanny & Johnson, 2000; Steblay, 1992). In obvious ways, this effect parallels the pattern observed by Easterbrook, with good memory for items at the center of the crime (the gun or knife) but poor memory for items at the periphery. Could it be, therefore, that Easterbrook’s claim applies to emotional remembering in general?
Studies of Memory Narrowing Many studies have asked whether participants show the pattern of “memory narrowing” for emotional events that we might expect by extension of the weapon-focus data. Early evidence on this point was mixed, and, indeed, one can
7
easily find in the literature claims that emotion promotes memory, claims that emotion undermines memory, and also claims that emotion has diverse effects. For example, and in keeping with the Easterbrook suggestion, Christianson and Loftus (1991) found improved memory for the central materials within an emotional event but impoverished memory for the peripheral materials, in comparison to memory for a neutral (but otherwise similar) event. However, both Clifford and Scott (1978) and Loftus and Burns (1982) reported evidence suggesting that emotionality impaired memory, and, using a somewhat different paradigm, Heuer and Reisberg (1990) found that emotion seemed uniformly to improve participants’ memory, a benefit that emerged for both central and peripheral materials (in comparison to memory for a matched neutral event). Although confusing when they were first published, these contrasting data patterns are in retrospect easily explained. If emotion does have opposite effects on central and peripheral materials, it is essential that we define these categories with care. To the extent that items on the memory test are miscategorized, this could easily obscure the results pattern. Similarly, our ability to make comparisons across studies is obviously compromised if different investigators use different categorization schemes for central and peripheral materials; that was, in fact, a problem in this early literature. Christianson and Loftus (1991), for example, defined their categories largely in spatial or perceptual terms; peripheral details were those that were truly in the background. Heuer and Reisberg (1990), on the other hand, used definitions that were more conceptual and counted as central any bits of information directly relevant to the plot, or in any way important for how the story unfolded. Peripheral information, in contrast, was information that could be changed without in any material way changing the story. Which of these categorization schemes “carves nature at the joints”? Burke, Heuer, and Reisberg (1992) decided to treat this as an empirical question and therefore used a very fine-grained categorization scheme in order to ask, in a data-driven fashion, which of these distinctions were relevant to emotion’s memory effects and which were not. In their data analysis, Burke et al. separated items that were relevant to how the story unfolded from items that were not (thus replicating Heuer and Reisberg’s relevance-based distinction). The plot-relevant items were then subdivided into “gist” items, items that essentially defined the story, and “basic level visual information,” items that, in the broadest terms, described what each of the slides in the sequence showed. Likewise, the plotirrelevant items were also subdivided, into those (irrelevant) details that happened to be spatially associated with plot-relevant actors or objections and those truly in the background. Finally, all of these categories were subdivided once again, but this time temporally, with questions divided according to whether they probed memory for materials that happened before, during, or after the parts of the story that were, in fact, arousing. Figure 1.1 depicts the Burke et al. categorization scheme, and, in a rough fashion, describes their results. As can be seen, the difference between plot-relevant
8
Figure 1.1. A schematic description of the Burke et al. design and findings.
and plot-irrelevant materials does matter, inasmuch as the pattern in the top half of the figure is different from the pattern in the bottom half. Likewise, the spatial/perceptual distinction also matters, evident in the contrast between the figure’s third and fourth rows. Finally, the temporal dimension also matters, as shown in the contrast between the table’s middle column and its two outer columns. (For related data, showing a similar effect of the temporal dimension, see, for example, Bornstein, Liebel, & Scarberry, 1998.) How to summarize this pattern? First, these data do confirm the memorynarrowing pattern, with emotion improving memory for materials central to the event and hurting memory for more peripheral materials. Second, the definition of the event’s center is complex—with materials favored by emotion if they are in any fashion tied to the “action” in the story—either conceptually or spatio-temporally. Several other studies have since confirmed this pattern, albeit not at the same fine grain. Safer, Christianson, Autry, and Österlund (1998), for example, showed participants a sequence in which a woman was shown either gathering flowers in a park (neutral version) or stabbed in the throat and lying on the ground, bleeding (emotional version). The participants’ memory was then tested with various photographs differing only in how much of a “close-up” they were; par-
9
ticipants were asked to select the exact photo they had seen in the earlier sequence. We know from other studies (e.g., Intraub & Richardson, 1989) that people often remember photographs as being less zoomed-in than they actually were and correspondingly remember the photo as including more of the background than it actually did, a pattern known as boundary extension. Safer et al., however, found precisely the opposite pattern, with the emotional photos remembered as more zoomed in. Apparently, the participants’ memories excluded the peripheral information and also excluded the fact that there even was peripheral information. Similarly Wessel and Merckelbach (1997, 1998) invited spider phobics to the laboratory and, in one procedure, showed them a large, live spider (contained within a glass jar) and, in another procedure, showed them pictures of spiders mounted on a bulletin board. In both cases, these were particularly arousing stimuli for these participants but not for control subjects. In a subsequent memory test, the more aroused (phobic) participants showed the expected pattern of narrowing, with better memory for the event’s center (the spider) and worse memory for the event’s periphery than control participants. It should be acknowledged, though, that not all studies confirm this broad picture, and two studies, one by Libkuman, Nichols-Whitehead, Griffith, and Thomas (1999) and one by Wessel, van der Kooy, and Merckelbach (2000), have failed to replicate the memory narrowing result. (We return to these nonreplications later.) Even so, the memory-narrowing pattern associated with emotional events has been replicated often enough to be regarded as wellestablished—especially when it is joined with the separate but substantial body of research directly examining the closely related weapon effect. Overall, then, where does this leave us? We began this section by asking whether emotional events are remembered not just vividly but also accurately. The answer depends on what aspects of the emotional event we are considering. If we focus on central materials within an event—the gist of the event and details that are spatially central and also associated with the gist—the data seem reasonably uniform, with emotion seeming rather reliably to improve memory; we will consider even more data on this point (and some complications) later in the chapter. If we look instead at more peripheral materials (plot-irrelevant details that are spatially removed from the ‘action’), the evidence is less clear, but, overall, the data seem to indicate that memory is impaired by emotion. Thus, emotional events seem to be remembered accurately but incompletely.
Could Memory Narrowing
Be Artifactual?
Emotional events are distinctive in many ways. Quite obviously, they are accompanied by the feelings and bodily changes that we call “emotion” and, for this reason alone, may be remembered differently from neutral events. But, as we
10
have already mentioned, several other factors may also be crucial here. Emotional events typically receive closer scrutiny than neutral events, and it may be this scrutiny, not the emotion itself, that explains the data we have so far reviewed. Emotional events are also likely to be perceived as worth thinking about after the fact and so probably are rehearsed to a greater extent than neutral events are. Perhaps it is this that shapes how these events are remembered, rather than the emotionality per se. In addition, emotional events are likely to be somewhat unusual (if they were more familiar, it seems likely that they would lose their emotional power), and this, too, may influence how they are remembered. Finally, emotional events are usually related to important themes or goals in our lives; this is presumably what makes them emotional in the first place. Perhaps it is this relation to important themes that is crucial for memory and, again, not the emotionality itself. These factors are difficult to control in any study of emotional remembering, but, even so, at least some evidence suggests that it is the emotionality that matters, over and above the undeniable contribution of these other points. For example, it is important that, in the Wessel and Merckelbach (1997, 1998) studies of spider phobics, it is the same event that is being remembered by the aroused and less aroused (nonphobic) subjects. This allows us to set aside a range of concerns that focus on some aspect of the emotional events (their familiarity, their coherence, their plausibility, their intelligibility) other than their emotionality. Moreover, a number of studies have directly tackled some of these extraneous factors. Christianson and Loftus (1991), for example, compared participants’ memory for three types of stimuli: a neutral stimulus, an emotional stimulus, and a stimulus designed to be unusual and attention-grabbing but unemotional. Specifically, the neutral stimulus told a story about a woman riding her bicycle; the arousal story told of a woman injured while riding her bicycle; the unusual story showed the woman carrying her bicycle on her shoulder. The results showed similar patterns for the emotional and unusual story with regard to the peripheral aspects of the story (a car seen in the background), but, for central information (the color of the woman’s clothing), performance was markedly better for the emotional event than for the unusual event. In short, then, the emotional and unusual stories led to different memory patterns, suggesting that the effects of emotion must be different from those of sheer distinctiveness or unfamiliarity. In a similar spirit, Heuer and Reisberg (1990) compared memory under four different circumstances. One group of subjects viewed an emotional sequence, and one a neutral sequence. (The doctor/mechanic series, already described, was used in this study.) A third group viewed the neutral sequence but was specifically urged to memorize the story as best they could; this condition seemed likely to elicit extra rehearsal of the story and so could illuminate rehearsal’s effects. A fourth group was urged to scrutinize the (neutral) story closely and was told specifically that the stimulus they were about to see paralleled a recent event in
11
the news; their task was to discern what that event was. (In truth, there was no such deliberate parallel, but participants had no way to know this.) This group was included to provide a comparison between emotion’s effects and the effects of close scrutiny (without emotion) and postevent contemplation. The comparison of these conditions is important if we are to isolate and identify emotion’s memory effects. It is unfortunate, therefore, that this study, done some years ago, relied on a relatively crude measure of memory, one resting on a categorization of central and peripheral materials that the investigators subsequently abandoned (cf. Burke et al., 1992). Nonetheless, it is notable that the memory data for the emotion group were easily distinguishable from the data for any of the other three groups. Thus, though some caution is needed here, this study does suggest that emotion’s memory effects cannot be reduced to these other factors. Roughly the same conclusion follows from a number of studies that have sought to examine emotion’s biological impact on memory. For example, several investigators have suggested that the amygdala plays a crucial role in emotional memory, a claim supported, for example, by studies of patients with UrbachWeithe disease, a disease that damages the amygdala without harming other brain structures. Adolphs, Cahill, Schull, and Babinsky (1997; also chapter 2 in this volume) presented two such patients with the stimuli developed by Heuer and Reisberg (1990). The patients found the emotional slides initially arousing, but showed no memory benefit from this arousal, in comparison to their memory for neutral materials. These data are interesting for what they say about the biological mechanisms underlying emotion’s memory effects but are also important for our purposes in this section. Presumably, Urbach-Weithe disease makes these stimuli no less unusual and no less coherent; if it is these factors that made the stimuli memorable, then the patients with this disease would remember the stimuli just as ordinary participants do. The data, of course, suggest otherwise. A parallel argument can be made for a study by Cahill, Prins, Weber, and McGaugh (1994). They showed (their adaptation of) the doctor/mechanic stimuli to participants but, prior to the presentation, injected half of the participants with propanolol, a beta-adrenergic blocker chosen to diminish emotion’s bodily (arousal) effects, and injected the remainder with a placebo. If the memory effects of the emotional stimulus truly depend on emotion (and, in particular, on the arousal that accompanies emotion), then the beta-blocker would be expected to reduce or eliminate these memory effects. If, on the other hand, the memory effects depend on factors like rehearsal or the unusual nature of the story, then it is not obvious why the beta-blocker should have an effect. Cahill et al.’s results indicate that beta-blockers do reduce (and may even eliminate) the memory differences between neutral and arousal stories, buttressing the claim that it is indeed emotion that matters for memory, and not some other attributes of these stimuli. Once again, though, caution is needed here,
12
because Cahill et al.’s study did not distinguish between central and peripheral elements of the story and so may not provide a complete portrait of emotion’s effects. In short, the evidence on these points is strongly suggestive, but not conclusive. The weight of the evidence suggests that emotion itself has an impact on memory, with the effect plausibly depending on emotion’s arousing effects (which are specifically diminished by the administration of beta-blockers). For reasons we have mentioned, however, this is certainly a point in need of further research scrutiny.
Does All Emotion Have
the Same Effects on Memory?
Does the Source of
the Emotion Matter?
We have so far painted a relatively straightforward picture of emotion’s effects. A large number of studies indicate that emotion improves memory for an event’s center; a somewhat smaller, but still substantial, group of studies indicate that emotion disrupts memory for an event’s periphery. There are, however, two conspicuous problems with this claim. One is the studies, already mentioned, that have explicitly tried, but failed, to replicate the memory-narrowing pattern (Libkuman et al., 1999; Wessel et al., 2000). A second, and perhaps broader, problem lies with the generalizability of the studies cited so far. Emotion’s capacity to promote memory for an event’s gist or center has been demonstrated in many settings, both inside the laboratory and out (e.g., Bohannon, 1988; Brewer, 1988; Linton & Melin, 1982; Pillemer, 1984). To mention just a few of the more recent studies, Bluck and Li (2001) showed that the strength of participants’ emotional feelings about the O. J. Simpson verdict was predictive of how fully the event was recalled 8 months later. Bornstein et al. (1998) reported that the emotional center of an R-rated commercially produced movie was better recalled than a comparable but neutral episode. Peterson and Whalen (2001) reported that high stress levels facilitated young children’s recall of an emergency trip to the hospital 5 years earlier. Davidson, Luo, and Burden (2001) reported that children are better able to remember emotional behaviors than unemotional ones. And so on. But what of the narrowing effect and, specifically, the tendency for emotion to diminish memory for an event’s periphery? Here the diversity of evidence is actually rather limited, because virtually all of the studies relevant to this effect have induced emotion in roughly the same way: by showing participants something gruesome, or gory, or anxiety-provoking. Examples include pictures of disfigured faces, a picture of a boy shot in the eye, a picture of a woman with her
13
throat slit, a picture of the severed legs of a child, or the sight of a dangerouslooking spider. All of these manipulations produce strong reactions, evident both in heart rates and self-report, but one might worry that these studies do not represent emotion as it naturally occurs outside the laboratory. In particular, one might argue that the emotion experienced outside the laboratory is more often induced, not by a particular visual stimulus but instead by involvement and empathy with an unfolding event. We become emotional, in other words, when we encounter issues and information that are pertinent to our lives, goals, and values (or, perhaps, the lives and values of people we care about). We refer to this more common kind of emotion as thematically induced, in contrast to the visually induced reactions involved in most previous studies, produced by the presentation of a specifically defined emotional visual stimulus. This distinction between thematically and visually induced emotion raises two concerns about the extant evidence. First, can we generalize from the prior studies, or does the evidence instead allow us only to characterize one type of emotional event, with a different profile needed for other events? Second, and more troubling, this distinction raises the specific possibility that the “narrowing” of memory, observed in many studies, may be artifactual. Recall that the Easterbrook claim attributes the pattern of memory narrowing to arousal, with the implication that this pattern would not be observed if the arousal were somehow avoided or diminished. An alternative possibility, however, is that this “narrowing” has nothing to do with arousal but is observed simply because the experimental stimuli provide a highly salient stimulus to focus on, a strong “attention magnet” that seizes participants’ concentration during the event and therefore dominates their subsequent recollection of that event. We note that essentially the same concerns have been raised by investigators examining the phenomenon of weapon focus, which we described in an earlier section. Many authors have attributed this pattern to the emotional arousal experienced by the eyewitness; on this view, weapon focus is just another manifestation of the effect described by Easterbrook. Other authors have pointed out, however, that the eyewitness may focus on the weapon simply because it is by far the most important and interesting aspect of the visual input. After all, what could be more important to a crime witness than to know whether he or she is in immediate danger or not, and, to this end, nothing in the scene is more important than knowing whether the weapon is cocked and pointed at him or her, or whether the criminal’s finger is on the trigger? On this logic, even an entirely calm witness might still show the weapon focus pattern, zooming in on the weapon because looking toward the weapon provides crucial information! As it turns out, the available evidence suggests that both of these mechanisms—one hinging on arousal, and one hinging on the weapon’s visual importance—may play a role in producing weapon focus. In several studies, for example, weapon focus has been observed even in the absence of emotion, indi-
14
cating the importance of the weapon’s potency as an attention magnet (Kramer, Buckhout, & Eugenio, 1990; Loftus et al., 1987; Maass & Köhnken, 1989). Other studies indicate that it may be the unusualness of the weapon, and not the threat, that produces weapon focus (Pickel, 1998). Still other studies, however, suggest that the strength of the weapon focus effect increases as arousal increases, indicating that arousal also plays a role (e.g., Peters, 1988). To the best of our knowledge, though, this issue has rarely been raised in the study of emotional memory (i.e., in the absence of a weapon). In the next section, therefore, we describe four studies designed to explore these points.
Thematically Induced Versus
Visually Induced Emotion
As a first step toward addressing the issues just raised, it seems sensible to ask about the nature of the emotional events that actually fill our lives. Is the emotion in these events typically visually induced (in which case the laboratory studies might be representative of emotionality in our day-to-day lives), or is the emotion typically thematically induced? In a pair of studies, Laney, Heuer, and Reisberg (in press) asked participants simply to list a series of emotional events from their lives; in their first study, the participants were all college undergraduates at a prestigious institution; in their second study, the participants were much more diverse in ages, professions, and educational backgrounds. In both cases, the participants were then interviewed about the events they had listed, and, based on these interviews, the events were coded as either thematically or visually arousing. These two studies yielded virtually identical data. Despite a coding scheme set up to bias things toward counting events as visually induced, the huge majority of events reported by participants were in fact thematically induced— 82% in the study with undergraduates and 71% with the broader population. These results strongly suggest that (as Laney et al. [in press] put it) human emotional memory is not like a blockbuster movie with great special effects; it is instead like a docudrama with complex characters and an emotionally engaging plot. Thus, most laboratory studies of emotional remembering are considering a form of emotion that is neither typical nor representative, and that of course invites the next question. Can we explore in the laboratory how participants remember thematically arousing events? In two unpublished studies, Laney et al. pursued this issue, asking how participants remember events that are arousing for thematic, not visual, reasons. The first of these studies evoked arousal by appealing to an issue clearly emotional for the college-age participants, namely, date rape. As in previous research, the stimuli involved a slide sequence plus tape-recorded narration. The sequence showed a man and woman on their first date, and the neutral and arousal sequences were visually identical except for one slide, late in the series. Hence, the
15
arousal manipulation was not contained within the slides themselves, but within the narration, ensuring that the arousal was not induced by a specific visual target. In the neutral version, subjects heard that the woman was relaxed and happy about the date, and the man was polite and friendly. In the arousal version, subjects heard that the woman was growing increasingly apprehensive as the date progressed, and these fears turn out to be well founded, as the man attacks the woman late in the sequence and has to be forcibly pushed away. The memory data in this study showed a robust effect of story, with participants having more complete and more accurate memories for the arousal story. This confirms the positive effect of emotion on memory for central materials, even with story materials that are thematically, not visually, arousing. On this point, Laney et al.’s study (unpublished) confirms the generality of prior findings. However (and crucially), there was no hint of an impairment in memory for peripheral aspects of the emotional story; instead, these were remembered better than peripheral aspects of the neutral story. Said differently, emotion seemed in this case to improve memory for all aspects of the story, and thus there was no indication at all, in these data, of memory narrowing. Another study confirmed these findings. In this case, participants viewed a series of 33 photographs conveying a story about a college student named Megan. In the arousal version, subjects learned early on that Megan is doing badly in her classes and may lose the financial support she’s been getting from her parents. In addition, Megan’s boyfriend just dumped her—on her birthday. As the story unfolds, Megan discusses the possibility of suicide and gets quite specific about it, contemplating the combination of pills and alcohol that is, in fact, one of the most common paths to suicide among college students. We emphasize, though, that all of this upsetting information was conveyed in the narrative that accompanied these slides; there was nothing in the visuals that was at all upsetting. Indeed, neutral subjects saw the exact same visuals and heard a story with it that paralleled the arousal story, but with some essential differences. They heard that Megan was doing well in her classes and that she and her boyfriend were getting along fine. They also heard about Megan reaching for a pill bottle, but, this time, in response to a hangover caused by her birthday celebration. Again, the memory data showed a strong effect of story, with better memory overall for the arousal story. But, as in the previous study, there was no indication in the data of memory narrowing; the positive effect that emotion had on memory was as reliable for peripheral details as it was for central materials. These two studies obviously suggest that memory narrowing is not inevitably produced by emotionality, in clear contrast to Easterbrook’s suggestion, many years ago, that it is arousal per se that leads to narrowed attention. Instead, these studies suggest that we need separate hypotheses to account for emotion’s positive effects on memory for central materials and its negative effects on memory for peripheral materials. The former is surely an effect of emotion and the arousal that accompanies it; this is suggested by the beta-blocker results and other find-
16
ings. But the latter effect seems not to be produced by emotion. Instead, it is the result of an event containing a powerful attention magnet that summons attention and, correspondingly, draws attention away from other aspects of the event. We hasten to say, however, that these claims about thematic arousal must be tentative, and this is a point on which further data are surely needed. A number of studies have documented memory narrowing for emotional events, but few studies have specifically examined thematic arousal, and, until the data base is larger, we urge caution on this theme. Indeed, one of our own early studies indicated that thematically arousing events do produce memory narrowing (Heuer, Reisberg, & Rios, 1997). We now believe that this early result, with its relatively crude stimuli and test materials, should be set aside in favor of more recent data collected with improved procedures. Even so, we must not pretend the data are univocal on the memory effects of thematic arousal, and replication studies are plainly required. Let us for the moment, however, assume that Laney et al.’s (unpublished) third and fourth studies can be taken at face value. Concretely, this is a claim that emotion improves memory for all aspects of a story, not just the center, and that the negative memory effects observed in prior studies are not a consequence of emotion but are, instead, produced by powerful attention magnets embedded within the emotional story. Does this mean investigators interested in emotion should henceforth ignore the memory-narrowing pattern, realizing that it is not a pattern produced by emotion? We believe that this question should get a firm no for an answer, for three reasons. First, it simply is the case that many emotional events do have a visual focus, do have an attention magnet embedded within the event. Laney et al.’s data tell us that such events are atypical, far outnumbered by thematically arousing events, but even so at least some emotional events do contain visible wounds, horrifying scenes, threatening weapons, and the like. If we are to understand how these events are remembered, we must consider the role of these salient visual targets in shaping memory. Second, even if the narrowing pattern is not produced directly by emotion, it may be potentiated by emotion. After all, what makes a visual stimulus a powerful magnet for our attention? Arguably, the key lies (at least in part) in the emotional importance of that stimulus—whether as a source of threat (e.g., a weapon), a focus of horror (e.g., a gaping wound), or a source of joy (e.g., the sight of a long-absent friend’s face). Thus, as emotion grows, the visual salience of these stimuli grows (cf. Öhman, Flykt, & Esteves, 2001), and it is then the visual salience that produces memory narrowing. Evidence consistent with this conjecture comes from a study to which we alluded earlier, by Peters (1988), examining adults’ memory for an occasion on which they received a rubella injection. The data showed a weapon focus effect, with the sight of the hypodermic needle apparently seizing the adult’s attention,
17
leading to an inability, later on, to identify the nurse who had given the injection. Importantly, participants’ ability to make this identification was negatively correlated with arousal (measured by heart rate), suggesting that arousal served to magnify the “threatening” quality of the “weapon” and thus to increase the “weapon’s” power to seize attention, thus undermining memory for other aspects of the event. Third, and more immediately, this separation of mechanisms may allow us to untangle some of the apparent contradictions in the empirical literature. We noted earlier that two studies, by Libkuman et al. (1999) and Wessel et al. (2000), have failed to obtain the “standard” memory-narrowing effect, despite their use of stimuli and procedures that seemed fully appropriate. If we assume that their stimuli were arousing, and if we assume that memory narrowing is (as Easterbrook suggested) a direct consequence of arousal, then it is odd indeed that these studies did not replicate the narrowing pattern. However, the framework we are developing here offers a path toward explaining this nonreplication. Memory narrowing, we are proposing, is not a direct consequence of arousal; instead, it is a consequence of how witnesses direct their attention during an emotional event. This allocation of attention is to some extent under strategic control; it is likely to be influenced by instructions and taskset; it is likely to differ somewhat from one subject population to the next. For these reasons, our current hypothesis (unlike Easterbrook’s claim) leads us to expect some degree of unevenness in the data pattern, with some studies demonstrating the narrowing pattern and some not. To be sure, this line of argument contains a substantial promissory note, because we have provided no details about any of these factors guiding attention and so no firm predictions about when narrowing will or will not be observed. Even so, an account cast in terms of attention offers flexibility that an account cast in terms of arousal does not, and this by itself seems noteworthy, given the mixed pattern of evidence.
Emotion Intrinsic to the Event,
Emotion Extraneous to the Event
We have now argued that the source of emotion may matter in determining how an emotional event is remembered. Does the arousal arise from the themes and meaning inherent in the emotional event or from some salient visual stimulus within the scene, such as the sight of a wound or a weapon? A related question also concerns the source of the emotion. Sometimes we are emotional during an event because the event itself is emotional—uplifting, perhaps, or frightening, or anger-provoking. But sometimes we are emotional for reasons external to the event we are participating in. We might still be emotionally aroused from some earlier event, now done. Or we might be aroused for some reason other than emotion—exercise, perhaps, or an overlarge dose of caffeine, or perhaps even an experimenter’s injection. Does this contrast matter? Are the memory effects of
18
emotion intrinsic to an event the same as the effects of emotion that merely accompanies the event? This issue is of interest for several reasons, including a question of how we should conceptualize emotion’s impact on memory. Many accounts describe this impact in terms of arousal mechanisms—increased norepinephrine levels in the bloodstream, for example, and increased serum glucose levels, and their effect on brain mechanisms serving memory (cf. chapters 2 and 3). If these are the keys in producing emotion’s memory effects, then the nature of the arousal (intrinsic to the event or extraneous) should not matter. But if, on the other hand, the content and meaning of an emotional event influence memory (as plausibly they would), then the contrast between intrinsic and extraneous arousal may be crucial. Over the last few years, a number of studies have pursued this issue. Christianson and Mjörndal (1985; also see Christianson, Nilsson, Mjörndal, Perris, & Tjellden, 1986) asked how injections of adrenaline influenced memory for pictures. They reported no effects of the injection on memory for neutral pictures and a difference between how participants remembered emotional pictures (after a saline injection) and how they remembered neutral pictures after the injection of a stimulant. This result suggests that the source of the arousal does matter and that only arousal intrinsic to the to-be-remembered materials has memory effects. (For a related, but somewhat more complex study, see Clark, Milberg, & Ross, 1983.) In contrast, though, we know that ingestion of glucose does improve memory in some circumstances. Hall, Gonder-Frederick, Chewning, Silveira, and Gold (1989) and Manning, Hall and Gold (1990) reported that ingestion of glucose (dissolved in a lemon-flavored drink) improved performance on a number of memory measures, including recall of an earlier heard narrative. In this case, mimicking one of arousal’s effects in a fashion entirely external to the to-beremembered material did improve memory. It is not obvious what to make of these mixed results, although we suspect that part of the problem lies with the arousal construct itself. There has been considerable dispute over whether arousal can be understood as a single, unified construct or whether, instead, we need to distinguish various types and multiple dimensions of arousal (for some “classic” statements on this issue, see Anderson, 1990; Neiss, 1990). With this point unsettled, it is difficult to make any comparisons across procedures. In addition, and perhaps more important for our purposes, what about memory for events, rather than memory for word lists, picture series, or verbal narrative? To address this question, Libkuman et al. (1999) showed half of their participants an arousing story and half a neutral story (using Heuer and Reisberg’s doctor/mechanic stimulus, already described). Within each group, half of the participants viewed the slides after a minute of sitting quietly, and half viewed the slides after spending a minute energetically running in place, to produce a
19
baseline of elevated arousal. Their results showed little or no effect of exerciseproduced arousal on memory. A second study replicated this finding, using a cued recall measure in place of a recognition test. A third study yielded similar data but this time with the exercise maintained during the slide presentation. (Participants pedaled on an exercise bicycle while viewing the slides.) Where does all of this leave us? We know that arousal does play an important role in mediating emotion’s memory effects. This is evident, for example, in Cahill et al.’s (1994, 2000) studies, showing that an interruption of arousal (via betablockers) disrupts emotion’s impact on memory. But, as we have now seen, at least some evidence suggests that this arousal must come from the “right” source and must somehow be tied to the to-be-remembered event; otherwise, the arousal seems to have no effect on memory. How should we put these clues together? One obvious conjecture is that physiological arousal may be necessary for promoting event memory but is not by itself sufficient for producing this memory effect. Also necessary is an emotional content to the event, something that engages participants’ attention, processing, and subsequent rehearsal. In essence, this cognitive work could provide the specific steps needed to establish an event memory and to link this memory to other bits of knowledge (making the target memory accessible after on). The physiological arousal could serve to facilitate these steps or to consolidate the memory once it is established through these steps. This position is fully compatible with the claim, already in the literature (e.g., McGaugh, 2000), that physiological arousal has its effect on memory chiefly through its impact on postevent memory-consolidation processes. In this view, arousal might play little role in establishing the content of emotional memories; that would be determined by other factors (such as the nature and meaning of the emotional episode itself). Arousal would instead play its role by cementing the content (whatever it is) in place, so that the resulting memory would be more complete and more long-lasting. From this perspective, neither the cognitive work evoked by an emotional memory nor the arousal by itself would be sufficient to produce emotion’s full memory effect; instead, both acting in concert would produce the effects we have been describing.
Does the Valence of
the Emotion Matter?
We have now argued that the source of emotional arousal matters in shaping emotion’s memory effects. Thematically induced emotion seems to produce different memory effects than visually induced, and emotional arousal intrinsic to the to-be-remembered event seems in several studies to produce different effects than extraneous arousal. But, of course, emotionality also varies in other ways, including the emotion’s valence. We saw earlier in the chapter that valence seems largely irrelevant to the relationship between emotion and memory vividness, so that sad memories are recalled just as vividly as happy memories, and traumatic
20
memories are no more vivid than memories for intensely pleasant events. But what about memory accuracy or completeness? Relatively few studies have pursued this issue. We do know that humor promotes memory—both for verbally presented jokes and for cartoons (e.g., Schmidt & Williams, 2001). But what about more complex materials? There is some suggestion in the literature that emotion’s memory effects are mediated by the release of stress hormones (Gold, 1989, 1992; McGaugh et al., 1993); one might draw from this the claim that only aversive, stressful events will show the memory benefits of these hormones. But, of course, there are many similarities between the biological effects of emotionally positive experiences and those of aversive experiences (e.g., Hamann & Ely, 1999), so, on this view, it may be the arousal itself that promotes memory, independent of the emotion’s valence. A recent study from our laboratory (Moyer, 2002) was designed to explore these issues, albeit using an unusual stimulus: the form of animation known as anime. This choice of stimulus was motivated by the fact that it is often difficult to evoke an emotional response in the laboratory, especially with a relatively brief stimulus designed to experimental specifications. It is helpful, therefore, to rely on a genre and a medium that is already engaging to students, and anime, a style of Japanese animation immensely popular with many college students, offered these advantages. We created three stimuli, each an edited version of commercially available animes, one depicting an emotionally neutral story, one depicting an emotionally negative story, and one a positive story. The emotionality of the stimuli was confirmed by pilot testing, as was the comparability of the stimuli on several other dimensions (quality, complexity, duration, number of scene changes). The positive event was drawn from the Magic User’s Club (Takahashi & Asari, 1996) and begins with four student members of the club, discussing flying on their brooms. One of the students attempts to launch her broom, and her efforts are both comical and erotic. Later, the anime shows the friends flying along, and one of the students zooms past them, unable to control her broom. She is forced to let go of the broom, and her rescue (by another of the students) is again both funny and sexy. Participants reliably rated this video as making them “amused” and, perhaps more impressive, many laughed out loud while viewing the video. How did they remember the video? Questions in the memory test had been classified as central or peripheral by a panel of judges, using criteria derived from the Burke et al. (1992) data. Items were central if they were either relevant to how the story unfolded or visually prominent within the anime (e.g., tied to a plot-relevant character or object). Items were peripheral if they were both irrelevant to the story and distant, in the judge’s assessment, from the ‘attention centers’ of the video. In the recognition test (Moyer, 2002), the participants remembered 65% of the central details in the comical anime, compared to 51% in the emotionally neutral stimulus. Thus, once again we see the broad advantage for emotional materials relative to neutral ones. In addition, participants remembered 34%
21
of the details considered peripheral by the judges, compared to 46% for the neutral stimulus. These data therefore replicate the memory-narrowing pattern (with improved memory for central materials, impaired memory for peripheral materials, in the emotional condition) but show this effect for the first time, as far as we know, with an emotionally positive stimulus. Moyer’s (2002) study also included a negatively tinged anime, drawn from Nosaka, Takahata, and Sato’s (1988) Graveyard of the Fireflies. The clip begins with a view of American fighter planes and a voice yelling “air raid.” The scene then shifts to a child (about 15 years old) and his sister, leaving their house for the air raid shelter. The anime shows them running among falling bombs and debris and eventually shows them at a school that has been converted to a hospital. A neighbor takes the young boy to see his mother, who is covered in bandages and obviously dying. Although we did not intend this, the negative anime lacks any specific visual target that can be construed as a focus for viewers’ attention or as an attention magnet (to use the term we introduced earlier). Instead, the affect, demonstrated clearly in pilot testing, involves an overarching feeling of sadness over these two children trying to survive and ultimately losing their mother. Therefore, inadvertently, this stimulus is akin to the thematically arousing stimuli studied by Laney et al. (in press), and this is reflected in the data: Once again, the emotion improved memory, with 65% of the central details remembered for this negative stimulus, in comparison to 51% for the neutral stimulus. However, in keeping with Laney et al.’s results, this (thematically arousing) anime showed no evidence of memory narrowing, that is, no evidence of an emotion-produced impairment for peripheral materials. Concretely, participants remembered 48% of the peripheral details of this anime, in comparison to 46% for the neutral stimulus. This study provides an initial hint, therefore, that emotionally positive events are remembered in much the same way that emotionally negative events are. Positive emotion conveys a broad memory advantage for the story’s gist and central materials, but positive emotion can also be accompanied by its own version of memory narrowing, its own version of weapon focus. A recently published study, however, seems at first pass to be inconsistent with Moyer’s (2002) data. Berntsen (2002) asked participants to think of the most traumatic experience they had ever experienced, and the happiest experience they had ever experienced, and then to record as many details of these events as they could. These details were then categorized as either central or peripheral. (In one experiment, external judges did this categorization; in another, the participants themselves did this.) Berntsen concludes from this study that “tunnel memories” (a pattern of memory narrowing) “are due to the combined effects of high arousal and negative valence, whereas high positive arousal . . . is not able to create the effect” (p. 1018, emphasis added). This seems to contradict Moyer’s finding, which did show memory narrowing with positively valenced events.
22
However, this data contrast is difficult to interpret. As one concern, Berntsen’s study (2002) does not assess memory accuracy, because there was no way to confirm (or disconfirm) her respondents’ recollections. Second, with participants able to select their own events for recollection, there is no way to know if the positive and negative events in Bernsten’s study were truly comparable. In particular, it is hard to know whether visually evoked and thematically evoked events were equivalently represented in the positive and negative recalls, and, as we have discussed, this may well have shaped the pattern of memory narrowing. Third, and most important, Berntsen’s study (2002) does not include data for neutral events, and this makes her results, as valuable as they are for her own purposes, impossible to interpret for our purposes. Concretely, what her data actually show is that the proportion of central details (relative to peripheral details) is greater for “shocking” memories than it is for positive memories. However, this leaves open the possibility that the positive memories may nonetheless have contained a greater proportion of central details (and, more to the point, a smaller number of correctly recalled peripheral details) than a set of neutral memories would have. With no way to evaluate this possibility (and with the other concerns just mentioned), there is no way to know from these data whether the positive events produced a pattern of memory narrowing or not. If we hold Bernsten’s study (2002) aside, then, Moyer’s (2002) data suggest that it is emotionality per se that matters for memory, not the emotional valence. A different line of data, however, raises questions about this proposal and suggests overall that we may need a more fine-grained account of how different qualities of emotion influence memory. Levine and Burgess (1997) have argued that the valence of emotion does matter for memory; they argue that the specific emotion will influence not just how much within an event is remembered but also what within the event is remembered. In their view, emotions enhance memory for information that is “functionally relevant” to the emotional state, and what information that is will vary from one emotion to another. Thus, a broad categorization of an episode’s elements as central and peripheral may be too crude, because information central for one emotional state may be less so for another state. To explore this claim, Levine and Burgess (1997) manipulated their participants’ initial mood by telling them they had received either an “A” or a “D” on a surprise quiz. Participants then took part in what they believed to be an unrelated study during which they heard a narrative that they subsequently had to recall. The results from this memory test were complex, with the content of participants’ recall influenced (as predicted) by the participants’ exact emotional state. Thus, for example, participants who were angry tended to have better recall for information that concerned goals of the individuals described in the narrative; there was no relation between degree of anger and recall of other types of information. Sadness, on the other hand, was associated with enhanced recall
23
of information about outcomes for individuals described in the narrative but not with other aspects of recall. In our view, Levine and Burgess’s (1997) argument merits close consideration. It is not yet clear whether the details of their proposal (with regard to “goals” and “outcomes” and so on) are warranted by data, and it is also worrisome from our point of view that the arousal at stake in their experiment was produced by a source external to the to-be-remembered event. (That is, it was the report of a grade that produced the arousal, but this grade had nothing to do with the narrative the participants subsequently had to remember.) Nonetheless, Levine and Burgess’s more general proposal seems plausible and is surely consistent with themes we have developed in earlier sections of this chapter. For example, we have suggested that neither arousal nor emotion is enough itself to produce the pattern of memory narrowing; instead, this narrowing will be observed only if the emotional event contains a salient stimulus, drawing participants’ attention. Moreover, we have suggested that in many cases a stimulus will gain its salience, that is, will gain its power to draw attention, from its meaning within the emotional event. Let us now add that this emotional meaning may well depend on the particular emotion being experienced, so that a stimulus salient for someone who is afraid may well be less salient for someone who is angry. With this additional step, the position we have sketched blends smoothly into that presented by Levine and Burgess. With this said, however, the fact remains that the available data are rather sparse for exploring the memory effects of different qualities of emotion. In our view, this provides powerful reason for investigators to broaden the focus of their work, and, until that is done, any claims regarding these issues need to be couched with caution. In the meantime, though, we do have at least some indications that the memory pattern for emotionally positive materials will resemble the pattern for emotionally negative materials, with an advantage overall (primarily deriving from information at the event’s center) but also with a disadvantage for peripheral materials if the to-be-remembered event contains a suitable visual focus. Even so, we are mindful of Levine and Burgess’s (1997) potentially crucial point that different emotions will favor memory for different aspects of the target event, and this is a proposal that cries out for further research.
Does the Intensity of
the Emotion Matter?
We have now considered difference sources of emotion and different qualities of emotion; what about different intensities of emotion? For years, the answer to this question has been cast in terms of the Yerkes and Dodson (1908) inverted-U curve, with the proposal that there is some ‘optimal’ level of emotion for promoting memory and that poorer and poorer memory is observed as one moves further away (in either direction) from this optimum. This pro-
24
posal seems intuitive enough and finds support in the fact that many studies have shown a positive relationship between memory and level of arousal, while other studies have shown a negative relationship. (Studies in the latter group are common, for example, in the research literature on stress and are also reported in the literature on trauma; for discussion of both, see chapter 3, this volume.) Plausibly, studies in the first group are on the uphill side of the YerkesDodson function, so that increasing the level of arousal moves us toward the optimum; studies in the second group could then be understood as on the downhill side of the function, so that increasing the level of arousal moves us away from this optimum. However, it may well be time to retire the venerable Yerkes-Dodson function, at least as applied to emotional memory. Other chapters in this volume will pursue the issue of how people remember extremely emotional events, but, for now, we will simply note that these events should not be remembered on most construals of the Yerkes-Dodson function but that they often are remembered—for quite some time and in considerable detail (e.g., chapters 2, 4, and 11). To be sure, there are some complications here (see chapters 3 and 8), but in any case the data cannot be read as indicating the Yerkes-Dodson, no-memory-for-extreme-emotion pattern. In addition, and more directly, Christianson (1992) argued persuasively that there is very little direct evidence in favor of the Yerkes-Dodson pattern in emotional memory. Indeed, the Yerkes-Dodson claim may actually be untestable, since it can explain virtually any result. (Note, for example, how readily this claim accommodates both positive effects of emotion on memory and their opposite.) With what should we replace the Yerkes-Dodson function? At the least, we should acknowledge that multiple mechanisms are likely to contribute to emotion’s memory effects, and it is plausible that each will have its own ‘optimal’ operating circumstances (cf. chapter 3). If so, the function linking arousal level and memory may not be unimodal, as the Yerkes-Dodson function is, but more complex. Consistent with this suggestion, consider a result reported by White (1991). Using an animal model, White examined the rate of learning as a function of the sugar dose delivered to the animal via a posttraining injection. His data do not reveal anything like the inverted-U function suggested by Yerkes and Dodson. Instead, his data plots have two peaks, with the clear suggestion that multiple mechanisms are in play, each contributing to the overall performance but with its own optimal level. This result implies that the relation between arousal and memory will be more complex than Yerkes and Dodson envisioned but still a relation that can be understood in terms of a single predictor variable. Figure 1.2 offers an appreciably more complex model, initially offered as a proposal for describing sports performance by Fazey and Hardy (1988) and then applied to memory by Deffenbacher (1994). This model implies that the relation between arousal and performance may depend on other variables, such as the person’s level of anxiety. At low anxiety levels, Fazey
25
Figure 1.2. The Yerkes-Dodson inverted-U function is often offered as a description of how memory performance will change as a function of increasing arousal. This figure, suggested by Fazey and Hardy (1988), provides a more complex and perhaps more realistic model. Whether the details of this model turn out to be correct or not, Fazey and Hardy remind us that the relationship between memory and arousal is likely to interact with other variables. They suggest that a key third is “cognitive anxiety.”
and Hardy forecast a pattern similar to the Yerkes-Dodson inverted-U. As anxiety increases, however, the pattern changes, with a sharp discontinuity in the curve and the level of performance depending on whether arousal is gradually being increased or decreased. Not enough data are available to evaluate this model (or variants on it), but it does remind us that the “optimal arousal level” for memory (if one can be defined) is very likely to depend on other factors, much as figure 1.2 suggests. If so, the Yerkes-Dodson function is not just unproven and probably untestable, it may be on the wrong track altogether, conceptualizing the data pattern in a (unidimensional) fashion that is far too simple. Once again, though, we call attention to the fact that this theme is plainly underresearched.
“Lost Memories”? Before moving on, we need to address one further topic clearly related to the themes of the last two sections. What happens when the emotion associated with an event is extremely intense and also strongly negative in its valence, perhaps with the valence marked by anxiety, perhaps by a sense of betrayal? Many authors have suggested that, under these circumstances, the relation between emotion and memory changes dramatically. No longer does emotion promote
26
memory, but, instead, in these circumstances, emotion will be associated with amnesia for the painful events. This is not the place to delve deeply into this difficult and contentious issue, but we do wish to touch a few points related to themes elsewhere in this chapter. (For more on this broad issue, see chapters 3, 4, and 11.) In doing so, we believe it important to separate the empirical claims from the interpretations often offered for the data; we begin, therefore, by considering the facts and (as we view it) the considerable ambiguity attached to those facts. Can one experience a highly emotional event but not remember it? The answer is yes, for several reasons. First, some emotional events discussed in the literature have been experienced by very young children (e.g., Williams, 1995) and so are open to the widespread pattern of childhood amnesia (cf. Shobe & Kihlstrom, 1997), a pattern that applies to all events in the first years of life, emotional or not. Second, emotional memories, as strong as they tend to be, are still vulnerable to the same sorts of forgetting mechanisms as other memories, including interference from other remembered events and retrieval failure. This, too, can lead to apparent amnesia for emotional events, and there is at least some indication that the rate of forgetting is similar for emotional and nonemotional memories (Read & Lindsay, 2000). Third, there is reason to believe that extreme levels of emotion can, in some circumstances, have a catastrophic effect on memory, causing relatively complete amnesia for horrific events that happened just hours or days earlier (Arrigo & Pezdek, 1997). It seems likely in these cases that the amnesia is caused by the bodily changes associated with extreme stress, changes that could disrupt the biological processes of memory consolidation needed to establish a memory in the first place. On this view, the high levels of stress have washed away the seeds of memory, so to speak, before they had a chance to take root. What about cases in which people report that they have long been amnesic about an emotional event but have now recovered the relevant memories? There is no doubt that these reports of memory discovery do occur, and here too there are many mechanisms that might contribute to such reports. First, we believe that some of these cases involve memories that have never been truly lost; the “recovery” merely refers to a new-found willingness to talk about this always remembered but never discussed episode. This willingness might emerge in the safety of a therapy setting; it might be encouraged by a social climate in which reports of previous traumas are supported and perhaps even encouraged. Second, we believe that some of these cases involve “recategorizations” of the relevant memory. “I always knew that he did X, but until recently I didn’t realize there was something blameworthy in what he did, and it’s only now, when I realize that the action is culpable, that I’m talking about it.” Third, there are some cases in which, remarkably, someone is convinced he was amnesic about an event at some prior time but, in truth, did remember the event all along. In other words, this is an example of a memory error, but what is being misremembered is the
27
fact of amnesia itself! Several such cases have now been documented, cases that are, in a sense, the inverse of someone falsely claiming, “I knew it all along.” Hence, these cases are sometimes referred to as revealing the “I forgot it all along” effect (Schooler, Bendriksen, & Amadar, 1997; also Padilla-Walker & Poole, 2002). Fourth, we have noted that at least some emotional memories will be forgotten for routine reasons such as retrieval failure; these cases of forgetting are likely to be reversed if suitable retrieval cues should present themselves. Fifth, and finally, it does seem sadly plausible that at least some recovered memories may turn out to be false memories, recalling events that never took place or events that unfolded differently from the way they are remembered. In light of all these comments, we believe it is unsurprising that some emotional memories do seem genuinely to be forgotten, and also unsurprising that there are many reports of emotional memories allegedly forgotten and then rediscovered. Moreover, we have suggested that many different causal sequences could lead to these facts, and, notably, in most of these causal sequences, there is no reason at all to challenge the veracity of the allegedly recovered memory. Why, then, have recovered memories been so controversial? The debate, we believe, does not revolve around the facts themselves; those, we have just suggested, are relatively uncontentious but also open to multiple interpretations. Instead, the debate hinges on a particular claim about the facts, namely, that emotional memories are lost because of a special mechanism of “repression” or “dissociation,” a mechanism that involves mental processes markedly different from those involved in ordinary emotional remembering. Our view is that the facts do not warrant these latter claims, because, as just discussed, the available facts are easily accommodated with no need for new and distinctive mechanisms. We stress, though, that this is in no way intended as a statement of blanket skepticism about these lost or recovered memories. Instead, it is simply an assertion that there are multiple non-exotic steps through which emotional memories might well be apparently lost and then rediscovered.
Alternative Approaches Beyond Accuracy:
Memory Detailedness and Coherence
We began this chapter by asking how accurately emotional events are remembered. The answer, it seems, is “it depends.” Overall, many studies show that emotion promotes memory for an event’s center and that at least part of this promotion depends on bodily arousal. But many studies also show that emotion impedes accurate memory if the to-be-remembered event contains a salient visual target, summoning attention; in this case, memory for the event’s center will be enhanced but at the cost of poorer memory for the event’s periphery. In addi-
28
tion, we have at least some indications that this pattern can be observed for emotionally positive events (e.g., an event that is comical and somewhat erotic) as well as emotionally negative events. Beyond this, though, at least two complications are needed in this account. First, the pattern of what specific content is remembered and what is not may depend on the nature of the emotion (after Levine & Burgess, 1997), and this issue needs further exploration. Second, the relation between memory and emotional intensity remains to be fully explicated, and it does seem likely that there will be circumstances in which emotion (perhaps only extreme emotion) has none of the effects we are describing but, instead, undercuts memory (see, for example, chapter 8). It should be acknowledged, though, that this account views emotional memory from a distinct perspective—one highlighting a memory’s accuracy and completeness as its most important attributes. To be sure, this concern with memory accuracy is easy to justify, especially in light of psychology’s increasing concern over the accuracy of eyewitness memory in the courts. But other perspectives on emotional memory should also be explored and have their own value. As just one example, consider a study by Williams et al. (1996). Their study begins with the fact that hopelessness is a key factor in leading a depressed person to contemplate suicide, and so Williams et al. ask: Where does hopelessness come from? They explore the proposal that one becomes hopeless when one cannot imagine future events in any detail, and this impairment in imagination, in turn, becomes likely when one does not recall past events in detail. Their study confirms these claims, with both a correlational design (using hospitalized patients who had, in fact, recently attempted suicide) and an experimental design (to explore causal relations). The Williams et al. article is interesting for several reasons, including the fact that memory accuracy plays no role in their account. Instead, what matters is memory detailedness and, arguably, memory vividness, and, more, these attributes matter for functions of considerable interest—the use of memory in problem solving; the use of memory in gauging the likelihood of future events; and, prominently, the use of memory in generating hope for the future. Thus, this article provides a compelling reminder that these other dimensions of memory are well worth our scrutiny. What do we know about emotion’s influence on these other dimensions? As we mentioned at the very outset, memory vividness is strongly correlated with how emotional an event is recalled to have been (Reisberg et al., 1988), but there are two concerns about this result (and related findings in the literature). First, note that the predictor variable here is retrospective assessment of an event’s emotionality, and, as we will see in a moment, we may not be able to take this retrospection at face value. Hence, we cannot be certain that there is a relation between how emotional an event was at the time of its occurrence and subsequent memory vividness. Second, the Reisberg et al. data (and, again, more recent, related findings) are in any case correlational, leaving questions about
29
causal relationships. Indeed, Reisberg et al. also reported a strong correlation between memory vividness and how consequential the event was judged to have been (and also a correlation between rated emotionality and rated consequentiality), and this obviously clouds causal interpretation. If we can set these cautions aside, though, it does seem that emotional events will be recalled more vividly than other events and, hence, following Williams et al. (1996), will be more influential and more compelling in gauging the future than other events. In addition, the content of emotional memories (whether accurate or not) may be somewhat different from that for neutral memories, and this, too, may matter for the function of these memories in problem solving, predicting the future, and the like. This difference in content is suggested by a result reported by Heuer and Reisberg (1990). They found that the number of intrusion errors made by participants was the same for emotional and neutral events, but the type of errors was different. Participants who had viewed the neutral stimulus (about the boy visiting his mechanic father) tended to recall plot elements that had, in fact, not been mentioned. Participants who had viewed the emotional stimulus (with the boy visiting his surgeon father) tended to recall ‘psychological elements’—about how upset the boy was, for example, or how angry the mother was that her son had seen such gruesome sights. All of this leads us, then, to a by-now familiar message. We know a lot about memory for emotional events, but it is not difficult to find gaps in our knowledge, including our knowledge of the relationship between the emotionality of an event and how detailed, vivid, compelling, or coherent the memory for the event will be later on. In this setting, it is interesting to set psychologists’ inquiries into memory (and autobiographical memory in particular) side by side with the literary study of the genre of autobiography. In the study of this genre, there is ample discussion of how a recollection shapes someone’s values or character and discussion of how the qualities of a remembered event can compel someone to action, or fail to. Notably absent from this study, though, is much concern about the historical correctness of these memories; it is not the correctness that makes the memories formative or compelling. Perhaps psychologists would be well-advised to learn from this tradition—certainly not abandoning our research on memory accuracy but supplementing it with a rich and (perhaps) equally important set of questions about how emotion shapes memories in a way that can, in turn, mold who we are, how we will act, and what we will believe in the future. (For an important psychological foray into these issues, see Pillemer, 1998.)
Remembering the Emotion Itself Our concern about alternative approaches to emotional memory also leads us to another issue. Not only do we remember the content of an emotional event, we also remember our assessment of, and reactions to, the event. In particular,
30
we remember how emotional the event was or was not for us, and this memory, in turn, can also be more or less accurate. For many people, introspection suggests that recall of prior feelings is immediate, long-lasting, and quite compelling. One may not remember much about a particular speaker’s lecture, but one remembers that the talk was boring; one may not recall the plot of a movie, but one remembers that it was quite funny; and so on. Despite these intuitions, however, evidence suggests that retrospective reports of emotionality are often reconstructions, influenced heavily by current assessments, and, in some cases, inaccurate. Of course, some memory for past feelings is rather accurate. For example, Safer, Bonanno, and Field (2001) surveyed participants whose spouse had died 6 months prior to the study. The participants reported their level of grief at the time of the survey and, then, roughly 4.5 years later, tried to recall their level of grief at the time of the survey. Safer et al. report that this recall was impressively accurate, although participants whose grief diminished relatively little over time did tend to overestimate their prior grief. It is important, though, that participants’ contemporary assessment of prior emotional events does change, so that the sting of past mishaps gradually fades, as does the glow of past triumphs. This pattern was documented by Walker et al. (1997), who asked participants to rate how they currently felt about previously experienced emotional events; these ratings were collected 3 months, 1 year, and 4.5 years after the target event. Walker et al. report that all the ratings became less extreme as time went on, and, interestingly, the unpleasantness of past humiliations and defeats faded more quickly than the pleasantness of past joys and celebrations. How do these changes in current assessment of an event influence the memory for the event? Some insight is provided by Levine’s (1997) study of how supporters of Ross Perot’s presidential campaign (in 1992) recalled their feelings upon hearing that Perot had withdrawn from the race and also their feelings upon hearing about Perot’s subsequent decision to rejoin the race. Levine reports that her participants showed frequent errors in how they recalled their past emotions but not (as some have suggested) a general tendency to overestimate past emotion. Instead, she found systematic distortions in emotion recall, bringing emotions-as-remembered into closer proximity to current appraisals. In short, it appears that her participants were to some extent reconstructing what their past emotions were likely to have been and were basing this reconstruction on their current emotions. A follow-up study found similar results in students’ recollection of their emotional reactions when they first heard that O. J. Simpson had been acquitted of accusations that he had murdered his wife (Levine, Prohaska, Burgess, Rise, & Laulhere, 2001; also Levine & Safer, 2002). As time passed after the acquittal, Levine et al. found systematic changes in how people remembered their own emotional reactions, with these changes tending quite strongly toward bring-
31
ing past emotions (as remembered) into alignment with current assessments. Thus, participants who had gradually grown angrier about the acquittal recalled their initial reaction as being angrier than it was; participants who had grown less upset recalled their initial reaction accordingly. A different form of reconstruction of emotion, based on other cues, can also be observed in a shorter time scale. In one laboratory study, participants viewed photographs of faces while listening to sentences spoken in either a positive or negative tone of voice (Ochsner, Schacter, & Edwards, 1997). The faces themselves conveyed either positive or negative affect, albeit in a subtle way. In a subsequent memory test, participants were shown the faces once again and asked to recall the tone of voice in which the pictured person had spoken. The participants tended to recall the voice as having the same affect as the picture—a memory illusion in which remembered-affect is apparently reconstructed from other information, both perceptually given and drawn from memory. Other results show similar patterns, with affect associated with one aspect of a stimulus altering (and in some cases distorting) how one remembers other aspects of the stimulus. For example, in a study by Hertel and Narvaez (1986), participants watched videotapes of conversations; memory was then tested (via a recognition test in one procedure, recall in another) for the specific words uttered in the conversation. Memory was clearly influenced by the emotional valence of the nonverbal gestures and facial expressions visible in the videotape (with the actual script held constant) in a fashion that suggested that the (nonverbal) affect had influenced how participants remembered the gist of the conversation. (For a related result, see Nygaard & Lunders, 2002.) Surely, then, our memory for prior emotion (either experienced or observed) is based to some extent on reconstruction, and our memory for prior emotion can, in turn, shape how we recall other aspects of the event. These facts must be part of our broader account of how emotional events are remembered, but note that these facts also have methodological implications. We mentioned in the previous section that the data linking emotion to memory vividness rest on retrospective assessments of emotion; we now see good reason to be cautious about those assessments. Similarly, we will, in a later section, return to the topic of flashbulb memories, and one of the issues there will be the role of emotion in forming such memories; many of the relevant studies, however, have relied on retrospective assessments of this emotion. For all of these reasons, we would be well served by an improved understanding of how emotion itself is recalled. Finally, one other complication should also be mentioned. Many events are likely not to be uniformly emotional; instead, they will have a mix of intense moments and less intense ones, moments of strong feelings interwoven with moments of relative calm. How will this changing profile be reflected in memory? One might think that the subsequent memory will reflect some sort of ‘average’ of the emotion felt over the course of an event, but several studies suggest that this is not the case. Instead, people seem to retain just a few ‘snapshots’ of the overall
32
event, and memory for the whole is dominated by the contents of these few snapshots. One snapshot seems to record the emotional peak of the target event— not surprisingly, since that peak is likely to be salient both in perception and in memory. Another snapshot seems to record the emotional ending of the event— again, not surprisingly, given what we know about recency effects in memory (e.g., Reisberg, 2001). These two snapshots—the peak and the end—are then weighted heavily in subsequent recall of the event’s emotional meaning, so much so that the subsequent evaluation can be predicted almost entirely from evaluations of these two moments (Fredrickson & Kahneman, 1993; Kahneman, 2000; Schreiber & Kahneman, 2000). Certainly, though, further work is needed to explore how one derives a summary evaluation from these snapshots and then how one’s broader recollection of the target event is shaped by this summary evaluation.
Working Backward
From an Emotional Memory
One last perspective on emotional memories invites our attention, but it is a perspective we have already touched in passing. In thinking about memory for emotional events, it is tempting to begin with the event itself and ask: If someone experiences an event that makes him angry, or happy, or afraid, how will he remember the event later on? But we could plausibly reverse this logic and begin instead with a bit of emotional recall. In this case, we might ask: If someone remembers an emotional event, then what can we conclude from this recollection? This shift in perspective is important for a simple reason. Some instances of emotional remembering may be false, recording events that unfolded rather differently from the way they are recalled, or perhaps recording events that never happened at all. In other words, it is true that, if an emotional event occurs, there will likely be a memory corresponding to that event later on, and this memory is likely to be detailed and reasonably accurate. (This is a consequence of the positive effects of emotion on memory, already discussed.) But the converse is not true. If an emotional event is recalled in vivid detail, it does not follow that there is likely to have been a prior event corresponding to the memory. Or, to put this succinctly, if there is an emotional event, then there will probably be an emotional memory for it; but if there is an emotional memory, there may (or may not) have been an emotional event. As illustrations, we have already mentioned that the intensity of emotion in an event may well be different from the intensity as it is recalled (Levine, 1997; Levine et al., 2001). We have also mentioned the Neisser and Harsch (1992) data, in which students’ very emotional recollections of the space-shuttle explosion turned out to be wrong not just in detail but in major elements as well. Presumably, Neisser and Harsch’s participants were the victims of their own errors in reconstructing the earlier episode, and, ironically, emotional recollec-
33
tion may be particularly vulnerable to this sort of error. To see why, imagine an event that was not considered noteworthy when it occurred, but which was later deemed consequential. One example might be a first (and quite unremarkable) encounter between two people that later grew into an intense romance. A different example might be a political event that seemed small at the time, but which marked the beginning of a gradual slide (visible only later) toward some triumph or some tragedy. In such cases, there is little reason to have encoded the initial event with great care, no reason to have paid special attention, no reason to have rehearsed the event, no arousal to promote memory consolidation. Yet, after the fact, there is ample reason to want to recall the event, and this will spur extra effort toward reconstruction. Of course, without much initial encoding, information actually in memory will provide slim support for this reconstruction, leaving a large potential for schema-based inferences (which may or may not be correct) and intrusions from other, related, episodes. To explore these points, one would want a design in which two groups of individuals are exposed to a minor and emotionally bland episode and then led to recall the episode as best they can later on. Prior to the recall, however, one group would be exposed to information persuading them that the episode was, in fact, emotionally important in some fashion; this would tell us how, and how fully, perceived emotion influences memory reconstruction (its vividness, its completeness, its accuracy). To the best of our knowledge, however, no study of this sort has been conducted, despite the relevance of such a study to the ferocious debate over (potentially) false memories. The literature does contain hints on this topic—the different types of intrusion errors for emotional and neutral memories (e.g., Heuer & Reisberg, 1990) or the impact of different story outcomes on how a story is recalled (e.g., Spiro, 1980)—but, unmistakably, this is another topic crying out for empirical pursuit.
Flashbulb Memories Revisited We began this chapter with a mention of flashbulb memories, and it will be useful to return to this topic in light of the wide range of issues we have now considered. We know that these memories are vivid, detailed, and long-lasting, reasonably accurate in many circumstances (e.g., Conway et al., 1994; McCloskey, Wible, & Cohen, 1988) but also sometimes error-filled (e.g., Neisser & Harsch, 1992). What produces this pattern? Why are some flashbulb events remembered accurately, while others are not? As a number of authors have noted, several factors are relevant (cf. Conway, 1995; Finkenauer et al., 1998). One is the timing of the memory tests—that is, when the initial recording of the event occurs and then when the follow-up testing takes place (e.g., Winningham, Hyman, & Dinnel, 2000). It may also matter whether, how, and how often the flashbulb event is discussed—for example, as part of the routine “sharing” of special memories that people often do (e.g., Philippot & Rime, 1998; Rime, Philippot,
34
Boca, & Mesquita, 1992). Another, and crucial, factor is the consequentiality of the remembered event, with consequential flashbulb events (e.g., Margaret Thatcher’s resignation, for many in England, or the 1989 Santa Clara earthquake, for those living near the epicenter) remembered more accurately than less consequential events such as Thatcher’s resignation for those living in the United States or the earthquake for those living far from the epicenter (Conway et al., 1994; Neisser, 1996). This is perhaps an unsurprising result. Consequentiality will motivate close attention and will spur rehearsal; consequentiality is also likely to be associated with stronger emotion and thus will invite emotion’s overall positive contribution to remembering (see Finkenauer et al., 1998, for a more extensive discussion of emotion’s role in flashbulb formation). In short, then, flashbulb memories—like other memories—tend to be accurate but are not always accurate. Flashbulb memories, like other memories, fade with the passage of time, are promoted by rehearsal, are enhanced if the eventas-experienced is consequential. In these (and other) regards, flashbulb memories seem qualitatively similar to other emotional memories and so, despite early claims to the contrary, are almost surely not in a class by themselves.
Conclusions The main points of this chapter are relatively easy to describe. In general, emotion seems to have a positive effect on memory, increasing memory vividness, accuracy, completeness, and longevity. But emotion’s effects are not uniformly positive. Many emotional events contain a prominent visual stimulus, and, if so, emotion seems to promote a focus on this stimulus in a fashion that impairs memory for the event’s periphery. Emotion assigned to an event after the fact may also spur memory reconstruction based on too little information, and this may foster reconstructive error. And, finally, extremely intense emotion may work against memory, perhaps by interrupting the biological processes needed for memory consolidation. This summary of the evidence reflects considerable progress over the last 50 years, progress that has included the accumulation of a large quantity of data and has also included questions that have arisen in the literature and then been resolved to the satisfaction of many investigators (e.g., the “special status” of flashbulb memories). This progress has also included evidence that directly challenges some long-held and often-quoted claims, including the Yerkes-Dodson function and the Easterbrook notion that arousal directly leads to memory narrowing. At the same time, though, the progress on these various fronts has helped highlight the gaps in what we know about memory for emotional events. In this chapter, we have considered the distinction between events that are visually arousing and those that are thematically arousing, but in truth the data pertinent to this distinction are few. We have also considered how people remember
35
events that are emotionally positive or how the exact content of an emotional memory might be influenced by the particular emotion in place during an event. On these points, too, we know far less than we might wish. Other issues are also largely untouched, including the nature of intrusion errors in memory for emotional events, emotion’s influence on our efforts toward reconstructing a poorly remembered past event, and emotion’s role in shaping the sorts of memory detailedness that seem to matter for problem solving and the generation of hope. Finally, also in need of elaboration is the exact role that arousal plays in shaping emotional memories. We have considered evidence that arousal may be necessary but not sufficient for producing emotion’s memory effects, and the mechanisms behind this pattern need to be specified. We close, therefore, simultaneously celebrating what we know about emotional remembering and showcasing the work still to be done. We have come a long way in the last 50 years, and the knowledge we have gained provides a rich base for now tackling the questions currently before us.
References Adolphs, R., Cahill, L., Schul, R., & Babinsky, R. (1997). Impaired declarative memory for emotional material following bilateral amygdala damage in humans. Learning & Memory, 4, 291–300. Anderson, K. J. (1990). Arousal and the inverted-U hypothesis: A critique of Neiss’s “Reconceptualizing arousal.” Psychological Bulletin, 107, 96–100. Arrigo, J. M., & Pezdek, K. (1997). Lessons from the study of psychogenic amnesia. Current Directions in Psychological Science, 6, 148–152. Berntsen, D. (2002). Tunnel memories for autobiographical events: Central details are remembered more frequently from shocking than happy experiences. Memory & Cognition, 30, 1010–1020. Bluck, S., & Li, K. Z. H. (2001). Predicting memory completeness and accuracy: Emotion and exposure in repeated autobiographical recall. Applied Cognitive Psychology, 15, 145–158. Bohannon, J. N. (1988). Flashbulb memories of the space shuttle disaster: A tale of two theories. Cognition, 29, 179–196. Bornstein, B. H., Liebel, L. M., & Scarberry, N. C. (1998). Repeated testing in eyewitness memory: A means to improve recall of a negative emotional event. Applied Cognitive Psychology, 12, 119–132. Bower, G. H. (1992). How might emotions affect learning? In S.-Å. Christianson (Ed.), The handbook of emotion and memory: Research and theory (pp. 3–32). Hillsdale, NJ: Erlbaum. Brewer, W. F. (1988). Memory for randomly sampled autobiographical events. In U. Neisser & E. W. Winograd (Eds.), Remembering reconsidered (pp. 21–90). Cambridge: Cambridge University Press. Brown, R., & Kulik, J. (1977). Flashbulb memories. Cognition, 5, 73–99. Burke, A., Heuer, F., & Reisberg, D. (1992). Remembering emotional events. Memory & Cognition, 20, 277–290.
36
Cahill, L., Prins, B., Weber, M., & McGaugh, J. (1994). ß-Adrenergic activation and memory for emotional events. Nature, 371, 702–704. Canli, T., Zhao, Z., Brewer, J., Gabrieli, J. D. E., & Cahill, L. (2000). Event-related activation in the human amygdala associates with later memory for individual emotional response. Journal of Neuroscience, 20, RC99. Christianson, S.-Å. (1992). Emotional stress and eyewitness memory: A critical review. Psychological Bulletin, 112, 284–309. Christianson, S.-Å., & Loftus, E. (1991). Remembering emotional events: The fate of detailed information. Cognition & Emotion, 5, 81–108. Christianson, S.-Å., & Mjörndal, T. (1985). Adrenalin, emotional arousal, and memory. Scandinavian Journal of Psychology, 26, 237–248. Christianson, S.-Å., Nilsson, L.-G., Mjörndal, T., Perris, C., & Tjellden, G. (1986). Psychological versus physiological determinants of emotional arousal and its relation to laboratory induced amnesia. Scandinavian Journal of Psychology, 27, 300–310. Clark, M. S., Milberg, S., & Ross, J. (1983). Arousal cues arousal-related material in memory: Implications for understanding effects of mood on memory. Journal of Verbal Learning and Verbal Behavior, 11, 671–684. Clifford, B., & Scott, J. (1978). Individual and situational factors in eyewitness testimony. Journal of Applied Psychology, 63, 352–359. Conway, M., Anderson, S., Larsen, S., Donnelly, C., McDaniel, M., McClelland, A. G. R., et al. (1994). The formation of flashbulb memories. Memory & Cognition, 22, 326–343. Conway, M. A. (1995). Flashbulb memories. Hillsdale, NJ: Erlbaum. Davidson, D., Luo, Z., & Burden, M. J. (2001). Children’s recall of emotional behaviours, emotional labels, and nonemotional behaviours: Does emotion enhance memory? Cognition & Emotion, 15, 1–26. Deffenbacher, K. (1994). Effects of arousal on everyday memory. Human Performance, 7, 141–161. Dietrich, D. E., Waller, C., Johannes, S., Wieringa, B. M., Emrich, H. M., & Muente, T. F. (2001). Differential effects of emotional content on event-related potentials in word recognition memory. Neuropsychobiology, 43, 96–101. Easterbrook, J. A. (1959). The effect of emotion on cue utilization and the organization of behavior. Psychological Review, 66, 183–201. Fazey, J. A., & Hardy, L. (1988). The inverted-U hypothesis: A catastrophe for sport psychology. British Association for Sports Sciences Monograph No. 1. Leeds, England: National Coaching Foundation. Finkenauer, C., Luminet, O., Gisle, L., El-Ahmadi, A., van der Linden, M., & Philippot, P. (1998). Flashbulb memories and the underlying mechanisms of their formation: Toward an emotional-integrative model. Memory & Cognition, 26, 516–531. Fredrickson, B., & Kahneman, D. (1993). Duration neglect in retrospective evaluations of affective episodes. Journal of Personality and Social Psychology, 65, 45–55. Gold, P. E. (1989). Neurobiological features common to memory modulation by many treatments. Animal Learning & Behavior, 17, 94–100. Gold, P. E. (1992). A proposed neurological basis for regulating memory storage for significant events. In E. Winograd & U. Neisser (Eds.), Affect and accuracy in re-
37
call: Studies of ‘flashbulb’ memories (pp. 141–161). New York: Cambridge University Press. Hall, J., Gonder-Frederick, L., Chewning, W., Silveira, J., & Gold, P. (1989). Glucose enhancement of memory in young and aged humans. Neuropsychologia, 27, 1129–1138. Hamann, S. B., & Ely, T. D. (1999). Amygdala activity related to enhanced memory for pleasant and aversive stimuli. Nature Neuroscience, 2, 289–293. Hertel, P. T., & Narvaez, A. (1986). Confusing memories for verbal and nonverbal communication. Journal of Personality and Social Psychology, 50(3), 474– 481. Heuer, F. (1987). Remembering detail: The role of emotion in long-term memory. Unpublished doctoral dissertation, New School for Social Research. Heuer, F., & Reisberg, D. (1990). Vivid memories of emotional events: The accuracy of remembered minutiae. Memory & Cognition, 18, 496–506. Heuer, F., Reisberg, D., & Rios, C. (1997). The memory effects of thematically induced emotion. In D. G. Payne & F. Conrad (Eds.), Intersections in basic and applied memory research (pp. 113–132). Hillsdale, NJ: Erlbaum. Intraub, H., & Richardson, M. (1989). Wide-angle memories of close-up scenes. Journal of Experimental Psychology: Learning, Memory & Cognition, 15, 179–187. Jones, E. B., O’Gorman, J. G., & Byrne, B. (1987). Forgetting of word associates as a function of recall interval. British Journal of Psychology, 78, 79–89. Kahneman, D. (2000). Evaluation by moments: Past and future. In D. Kahneman & A. Tversky, A. (Eds.), Choices, values and frames (pp. 693–708). New York: Cambridge University Press. Kramer, T., Buckhout, R., & Eugenio, P. (1990). Weapon focus, arousal and eyewitness memory: Attention must be paid. Law and Human Behavior, 14, 167–184. Laney, C., Heuer, F., & Reisberg, D. (in press). Exploring the contrast between thematically-induced and visually-induced emotion. Applied Cognitive Psychology. Levine, L. J. (1997). Reconstructing memory for emotions. Journal of Experimental Psychology: General, 126, 165–177. Levine, L. J., & Burgess, S. L. (1997). Beyond general arousal: Effects of specific emotions on memory. Social Cognition, 15, 157–181. Levine, L. J., Prohaska, V., Burgess, S. L., Rice, J. A., & Laulhere, T. M. (2001). Remembering past emotions: The role of current appraisals. Cognition & Emotion, 15(4), 393–417. Levine, L. J., & Safer, M. A. (2002). Sources of bias in memory for emotions. Current Directions in Psychological Science, 11, 169–173. Libkuman, T. M., Nichols-Whitenead, P., Griffith, J., & Thomas, R. (1999). Source of arousal and memory for detail. Memory & Cognition, 27, 166–190. Linton, S. J., & Melin, L. (1982). The accuracy of remembering chronic pain. Pain, 13, 281–285. Loftus, E., & Burns, T. (1982). Mental shock can reproduce retrograde amnesia. Memory & Cognition, 10, 318–323. Loftus, E., Loftus, G., & Messo, J. (1987). Some facts about “weapon focus.” Law and Human Behavior, 11, 55–62.
38
Maass, A., & Köhnken, G. (1989). Eyewitness identification. Law and Human Behavior, 13, 397–408. Manning, C., Hall, J., & Gold, P. (1990). Glucose effects on memory and other neuropsychological tests in elderly humans. Psychological Science, 1, 307–311. McCloskey, M., Wible, C. G., & Cohen, N. J. (1988). Is there a special flashbulbmemory mechanism? Journal of Experimental Psychology: General, 117, 171– 181. McGaugh, J. L. (2000). Memory—A century of consolidation. Science, 287, 248–251. McGaugh, J. L., Intoini-Collision, I. B., Cahill, L. F., Castellenano, C., Dalmaz, C., Parent, M. B., et al. (1993). Neuromodulatory systems and memory storage: Role of the amygdala. Behavioral Brain Research, 58, 81–90. Moyer, M. (2002). Polyanna revisited: Do we remember equally the good and the bad? Unpublished B.A. thesis, Reed College, Portland, Oregon. Neiss, R. (1990). Ending arousal’s reign of error: A reply to Anderson. Psychological Bulletin, 107, 101–105. Neisser, U. (1996). Remembering the earthquake: Direct experience vs. hearing the news. Memory, 4, 337–358. Neisser, U., & Harsch, N. (1992). Phantom flashbulbs: False recollections of hearing the news about Challenger. In E. Winograd & U. Neisser (Eds.), Affect and accuracy in recall: Studies of ‘flashbulb’ memories (pp. 9–31). Cambridge: Cambridge University Press. Nosaka, A. (Writer), Takahata, I. (Director), & Sato, R. (Producer). (1988). Graveyard of the fireflies [DVD]. Central Park Media Corporation, USA. Nygaard, L. & Lunders, E. (2002). Resolution of lexical ambiguity by emotional tone of voice. Memory & Cognition, 30, 583–593. Ochsner, K. N., Schacter, D. L., & Edwards, K. (1997). Illusory recall of vocal affect. Memory, 5, 433–455. Öhman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention: Detecting the snake in the grass. Journal of Experimental Psychology: General, 130, 466–478. Padilla-Walker, L. & Poole, D. (2002). Memory for previous recall: A comparison of free and cued recall. Applied Cognitive Psychology, 16, 515–524. Peters, D. (1988). Eyewitness memory and arousal in a natural setting. In M. Gruneberg, P. Morris, & R. Sykes (Eds.), Practical aspects of memory: Current research and issues (pp. 89–94). New York: John Wiley. Peterson, C., & Whalen, N. (2001). Five years later: Children’s memory for medical emergencies. Applied Cognitive Psychology, 15(7), S7–S24. Philippot, P., & Rime, B. (1998). Social and cognitive processing in emotion: A heuristic for psychopathology. In W. F. Flack, Jr. & J. D. Laird (Eds.), Emotions in psychopathology: Theory and research. Series in affective science (pp. 114–129). New York: Oxford University Press. Pickel, K. L. (1998). Unusualness and threat as possible causes of “weapon focus.” Memory, 6, 277–295. Pillemer, D. B. (1984). Flashbulb memories of the assassination attempt on President Reagan. Cognition, 16, 63–80. Pillemer, D. B. (1998). Momentous events, vivid memories. Cambridge, MA: Harvard University Press.
39
Pillemer, D., Goldsmith, L. Panter, A., & White, S. (1988). Very long-term memories of the first year in college. Journal of Experimental Psychology: Learning, Memory and Cognition, 14, 709–715. Porter, S., & Birt, A. (2001). Is traumatic memory special? A comparison of traumatic memory characteristics with memory for other emotional life experiences. Applied Cognitive Psychology, 15, S101–S117. Read, J. D., & Lindsay, D. S. (2000). “Amnesia” for summer camps and high school graduation: Memory work increases reports of prior periods of remembering less. Journal of Traumatic Stress, 13, 129–147. Reisberg, D. (2001). Cognition: Exploring the science of the mind. 2nd ed. New York: W. W. Norton. Reisberg, D., Heuer, F., McLean, J., & O’Shaughnessy, M. (1988). The quantity, not the quality, of affect predicts memory vividness. Bulletin of the Psychonomic Society, 26, 100–103. Rime, B., Philippot, P., Boca, S., & Mesquita, B. (1992). Long-lasting cognitive and social consequences of emotion: Social sharing and rumination. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology, Vol. 3 (pp. 225–258). New York: John Wiley. Rubin, D. C., & Kozin, M. (1984). Vivid memories. Cognition, 16, 81–95. Safer, M. A., Bonanno, G. A., & Field, N. P. (2001). “It was never that bad”: Biased recall of grief and long-term adjustment to the death of a spouse. Memory, 9, 195–204. Safer, M. A., Christianson, S.-Å., Autry, M. W., & Österlund, K. (1998). Tunnel memory for traumatic events. Applied Cognitive Psychology, 12, 99–118. Schmidt, S. R., & Williams, A. R. (2001). Memory for humorous cartoons. Memory & Cognition, 29, 305–311. Schooler, J., Bendriksen, M., & Amadar, Z. (1997). Recovered memories and false memories. Oxford: Oxford University Press. Schreiber, C., & Kahneman, D. (2000). Determinants of the remembered utility of aversive sounds. Journal of Experimental Psychology: General, 129, 27–42. Shobe, K. K., & Kihlstrom, J. F. (1997). Is traumatic memory special? Current Directions in Psychological Science, 6, 70–74. Spiro, R. J. (1980). Accommodative reconstruction in prose recall. Journal of Verbal Learning and Verbal Behavior, 19, 84–95. Stanny, C. J., & Johnson, T. C. (2000). Effects of stress induced by a simulated shooting on recall by police and citizen witnesses. American Journal of Psychology, 113, 359–386. Steblay, N. J. (1992). A meta-analytic review of the weapon focus effect. Law and Human Behavior, 16, 413–424. Takahashi, H. (Director), & Asari, S. W. (Producer). (1996). Magic User’s Club: I’ll follow you [DVD]. Bandai Visual Corporation, Inc. Walker, W. R., Vogl, R. J., & Thompson, C. P. (1997). Autobiographical memory: Unpleasantness fades faster than pleasantness. Applied Cognitive Psychology, 11, 399–414. Wessel, I., & Merckelbach, H. (1997). The impact of anxiety on memory for details in spider phobics. Applied Cognitive Psychology, 11, 223–232.
40
Wessel, I., & Merckelbach, H. (1998). Memory for threat-relevant and threatirrelevant cues in spider phobics. Cognition & Emotion, 12, 93–104. Wessel, I., van der Kooy, P., & Merckelbach, H. (2000). Differential recall of central and peripheral details of emotional slides is not a stable phenomenon. Memory, 8, 95–109. White, N. M. (1991). Peripheral and central memory-enhancing actions of glucose. In R. C. A. Frederickson, J. L. McGaugh, & D. L. Felten (Eds.), Peripheral signalling of the brain: Neural, immune and cognitive function (pp. 421–441). Toronto: Hogrefe and Huber. Williams, J. M. G., Ellis, N. C., Tyers, C., Healy, H., Rose, G., & MacLeod, A. (1996). The specificity of autobiographical memory and imageability of the future. Memory & Cognition, 24, 116–125. Williams, L. M. (1995). Recovered memories of abuse in women with documented child sexual victimization histories. Journal of Traumatic Stress, 8, 649–673. Winningham, R. G., Hyman, I. E., Jr., & Dinnel, D. L. (2000). Flashbulb memories? The effects of when the initial memory report was obtained. Memory, 8, 209– 216. Yerkes, R., & Dodson, J. (1908). The relation of strength of stimulus to rapidity of habit-formation. Journal of Comparative Neurology of Psychology, 18, 459–482.
41
2
.
A
re there special neural mechanisms to account for our memories for emotional events? In the service of a species’ survival, has evolution equipped organisms with a specialized set of mechanisms that encode, consolidate, and retrieve memories in a domain-specific manner, operating differentially in emotional and in nonemotional contexts? Indeed, evidence from cognitive psychology and neuroscience suggests that such distinct emotional memory mechanisms exist and depend on specific neural structures that we will review. The uncommon vividness and (presumed) accuracy of flashbulb memories led Brown and Kulik (1977) to suggest that autobiographical memories surrounding the learning of emotionally arousing information relied on a unique memory process. These authors speculated that encountering emotionally salient events resulted in enhanced activity in the brainstem’s reticular formation, leading to a “now-print” mechanism that permanently fixed memories of these events—in addition to autobiographical ephemera accompanying them— in memory (Brown & Kulick, 1977; Schooler & Eich, 2000). More recent research has revised this proposal somewhat and indicates that while such memories can indeed be vivid, their accuracy is modest at best and often distorted (see Schmolck, Buffalo, & Squire, 2000). Though “flashbulb memories” are no longer held to be a qualitatively different kind of memory, the broader notion of emotional memories complements earlier ideas that such memories are typically experienced with special vividness and personal meaning and that there is evidence for specialized processes that subserve them. The current taxonomy of memory systems describes distinct neural mechanisms for declarative and nondeclarative memory processing (Squire & Kandel, 2000). The medial temporal lobe memory system—including the hippocampal formation as well as the overlying cortex—is known to be integral to the forma-
42
42
tion and consolidation of declarative memories (or relational memories; see Eichenbaum & Cohen, 2001), whereas nondeclarative memories depend on regions outside the medial temporal lobe. This classic taxonomy distinguishes memory systems on the basis of their mode of operation—in what way they are acquired and in what way they are accessible to guide behavior. Of the other ways of distinguishing memory systems, two are important to operationalizing the term “emotional memory”: (1) systems directly involved in memory versus those having a modulatory role and (2) systems differentially involved depending on the nature of the material processed (domain-specific). Emotional memory concerns a specific domain of declarative memory, namely, memory for events or stimuli that are themselves emotional or that occurred in an emotional context. Emotional memory encompasses the enhancement of memory (assessed by, for example, a recognition test) for stimuli that are emotionally arousing (e.g., pictures of mutilated faces) compared to memory for stimuli that are emotionally neutral (e.g., pictures of neutral faces); it also includes the enhancement of memory for neutral faces that were encoded in an emotionally arousing context compared to a neutral context; and it even includes the possible repression of memory for faces associated with extremely traumatic events (a controversial possibility). Our review focuses on the facilitative effects of emotion on declarative memory rather than its possible suppressive effects. As such, memory for the mutilated faces may be a particularly potent declarative memory, relying on the medial temporal lobe system just as in memory for neutral faces. This effect, though well described, needs to be distinguished from the equally well-described effects of chronic stress and severely traumatic context on memory; in those cases, declarative memory is also modulated but in a qualitatively different way. We thus envision a nonmonotonic relationship of emotional arousal to memory accuracy. Mild to moderate, and transient, emotional arousal generally enhances memory; severe or chronic emotional arousal instead suppresses and distorts memory. Though emotional arousal can be mapped on a continuum, its effects on declarative memory are not linear [see chapters 1, 3, and 8 for more discussion of this topic—Eds.]. Another important distinction concerns the effects of emotional arousal triggered by properties that are intrinsic to the stimuli being encoded (e.g., memory for pictures of mutilations) versus effects provided by a context (e.g., memory for neutral pictures encoded in a highly arousing context). Again, without going into detail, our view is that the two engage equivalent mechanisms and lead to similar effects when the emotional arousal relates to the stimulus. They may lead to different effects when the emotional arousal is irrelevant to the stimulus (see chapter 1). As a final note, and to avoid any confusion, we emphasize that we use “emotional memory” to refer to declarative memory for emotional stimuli (where “emotional stimuli” means stimuli that induce an emotion in those who perceive
43
the stimuli—either as a result of intrinsic properties of the stimuli themselves or as a result of an emotional context related to the stimuli). This usage differs from another usage of the term, especially in animal studies, as a form of nondeclarative memory (e.g., fear conditioning). Thus, classical fear conditioning has been occasionally referred to as emotional memory, and in a sense, it is that. But we do not use the term that way here. What neural structures are involved in making emotionally arousing stimuli more memorable than neutral stimuli? Are these structures in some way specialized to process emotional memory? And what mechanisms do such structures implement—in particular, do they modulate declarative memory at the level of encoding, consolidation, or retrieval? These questions will be the focus of this chapter. We will emphasize the role of the human amygdala in the enhancement of memory for emotional events through its influence over medial temporal lobe structures and other neural regions; we will also survey several other structures that may participate in emotional memory.
Emotional Memory in
Nonhuman Animals
Much of the human research presented in this chapter is predicated on animal research, specifically work in rats and nonhuman primates. In this section, we review some of the studies in animals, but the breadth of this work is beyond the scope of this chapter. We encourage the reader to look elsewhere for a comprehensive review of the literature on the topic (Baxter & Murray, 2000; Eichenbaum & Cohen, 2001; LeDoux, 2000; McGaugh, 2000). By virtue of its location in the brain, for many years the amygdala has been the focus of much interest as a possible site of learning and memory (Squire, 1987). It was originally thought to play an integral role in the formation of declarative memories, together with other proximal structures in the medial temporal lobe. More careful study has revealed the role of the amygdala to be separate from the traditional ‘medial temporal lobe memory system’ that includes the hippocampus and adjacent cortex (Murray, 1992; Zola-Morgan, Squire, Alvarez-Royo, & Clower, 1991; Zola-Morgan, Squire, & Amaral, 1986). Following the initial characterization of the memory deficits of the famous patient H. M. (Scoville & Milner, 1957), considerable work went into developing a nonhuman primate model of amnesia through lesions of the medial temporal lobe (see Eichenbaum & Cohen, 2001, for a review of these initial studies). Typically, these early studies used a surgical approach similar to that used in patient H. M. Namely, the entire medial temporal lobe of the monkeys was removed, including the amygdala, hippocampus, and surrounding cortical regions: the entorhinal, parahippocampal, and perirhinal cortices. After the successful development of a nonhuman primate model of
44
amnesia, work began on delineating the role of the individual anatomical components of the medial temporal lobe in the amnesic animal. This work led to the finding that circumscribed lesions of the amygdala (sparing the surrounding cortical areas) do not produce the amnesic effects produced by damage to other medial temporal structures, such as the hippocampus and entorhinal cortex (Zola-Morgan et al., 1991; Zola-Morgan, Squire, & Amaral, 1989). This is not to say that these lesions have no behavioral or mnemonic effects. Many of these animals displayed altered emotional behavior reminiscent of a subset of the symptoms described by Klüver and Bucy in the 1930s (1937; see also Brown & Schafer, 1888, and Weiskrantz, 1956), such as increased visual and oral inspection of potentially threatening objects and increased tameness. These studies by Zola-Morgan and colleagues showed that reduced declarative memory performance after damage to the medial temporal lobe cannot be due to amygdala damage per se. Whereas studies in nonhuman primates showed that the amygdala is not necessary for the formation of new declarative memories, earlier work in rats had begun to elucidate the modulatory role that the amygdala plays in memory formation. For the purposes of this review, we will focus on the role of the amygdala in the modulation of long-term declarative memories and refer the reader to other reviews (Davis, 1997; LeDoux, 2000) for coverage of the amygdala’s role in nondeclarative forms of emotional memory, such as Pavlovian fear conditioning. Perhaps the first indication of the amygdala’s modulatory role in memory came from studies of the effects of electrical stimulation of the rodent amygdala. Goddard (1964) demonstrated that amygdala stimulation following aversive training resulted in reduced memory for the training period, pointing to a role in the modulation of consolidation rather than a direct role in memory formation. Subsequent work showed that amygdala stimulation could either reduce or enhance the consolidation of previously learned materials depending on the intensity of the stimulation (Gold, Hankins, Edwards, Chester, & McGaugh, 1975; McGaugh, 2000), possibly analogous to the nonmonotonic relation between emotional arousal and human declarative memory that we briefly alluded to in the introduction. The neurobiology of this phenomenon has subsequently been the focus of much research. Based on this work, we now know that key components through which the amygdala modulates declarative memory include the stress hormones epinephrine and corticosterone, as well as noradrenergic, GABAergic, glutamatergic, and peptidergic neurotransmission within the amygdala (McGaugh, 2000; Roozendaal, 2000; Tomaz et al., 1993). Hormonal modulation of memory storage has been a major focus of study in physiological psychology (McGaugh, 1983). Many of the behavioral paradigms used to study learning and memory in the rat are sufficiently stressful to result in the release of the adrenal hormones epinephrine and corticosterone. Gold and van Buskirk (1975) showed that the administration of epinephrine after a learning episode enhanced memory for that episode. Numerous studies have replicated this finding using both endogenous and exogenous manipulation of epinephrine.
45
Similarly, corticosterone manipulation has shown that this hormone plays a modulatory role in memory (Roozendaal, 2000). The mechanism of action of these hormones on memory function appears to be at the level of the amygdala (McGaugh, 2000). Manipulations such as amygdala lesions or blockade of either the ß-adrenergic receptor (Liang, Juler, & McGaugh, 1986) or the glucocorticoid receptor (Oitzl & de Kloet, 1992) in the amygdala block the memory-enhancing effects of emotional arousal. The final common pathway of the effects of these hormones on memory appears to be noradrenergic neurotransmission within and originating from the amygdala (McGaugh, 2000). Blockade of ß-adrenergic activity within the amygdala blocks the memory-enhancing effects of both epinephrine (Liang et al., 1986) and corticosterone (Roozendaal, Williams, & McGaugh, 1999). The relationships among epinephrine, the amygdala, and memory are somewhat complicated, since peripheral epinephrine does not cross into the brain, but instead exerts its effects via the release of glucose (Gold & van Buskirk, 1975) or via neural transmission of body-state information through the vagus nerve (Clark, Krahl, Smith, & Jensen, 1995; Clark, Naritoku, Smith, Browning, & Jensen, 1999). The amygdala is not involved in all aspects of the memory but only influences other regions, such as the hippocampus and striatum (Packard & Teather, 1998), during emotional arousal and during a specific time window in consolidation (Bianchin, Mello e Souza, Medina, & Izquierdo, 1999). Inactivation of the amygdala does not block the memory for an emotionally arousing event once established (Packard, Cahill, & McGaugh, 1994), suggesting that the memory trace per se does not depend on the amygdala. As we said in the introduction, an important issue in relation to the effects of emotion on long-term memory is the well-documented deleterious effects of stress on memory. (A full account of the effects of stress on cognitive function is beyond the scope of this chapter; a brief explanation follows.) We think of emotion as a transient state (on the order of msec to seconds) during which physiological response systems are activated in preparation for action; stress, on the other hand, is a more prolonged physiological perturbation due to a challenge to homeostasis (see Lovallo, 1997; McEwen, 2000). The onset of a stressful experience likely involves the experience of negative emotion, often characterized by a lack of control (Lovallo, 1997). The more prolonged nature of stress may result in a completely different pattern of effects on physiology and behavior. One might think of this relationship as analogous to the Yerkes-Dodson law (1908), which is characterized by an inverted-U-shaped effect wherein midrange levels of emotion are beneficial to memory whereas higher, more prolonged emotional experience may result in stress and subsequently impaired memory performance. Considerable research has documented just such a relationship between stress and memory performance in animals and humans (see Lupien & Lepage, 2001; McEwen & Sapolsky, 1995). This inverted-U effect likely does not operate within the realm of emotional memory that we refer to in this chapter (see chapter 1 for a discussion of this topic); instead, the descending limb of the inverted-U—at which point
46
memory would be deleteriously affected—corresponds to the onset of stress effects on memory. The relationships among emotion, stress, and memory are quite complicated and the focus of much research. Throughout this chapter, we focus primarily on the effects of emotion on memory. We refer the reader to one of many recent reviews on the topic of stress and memory for full coverage of this fascinating topic (see de Kloet, Oitzl, & Joels, 1999; Kim & Diamond, 2002; Lupien & Lepage, 2001; see also chapter 3 in this volume). Results from these studies suggest that a special neurobiological system— relying on the amygdala and stress hormones—is active during arousing learning situations to enhance memory for these events. The role of the amygdala, then, is not in the storage of emotional memories but in the modulation of other neural structures that directly implement such storage. During emotional arousal, the amygdala increases its modulatory effect and, together with the influence of peripheral stress hormones, modulates other neural structures such as the hippocampus and striatum to enhance the memory trace for emotional events. This animal research has served as a basis for much of the work in the neuropsychology of human emotional memory. Are the same neural mechanisms outlined in nonhuman primates and rodents applicable to the study of human emotional memory in the laboratory? The answer to this question will be the focus of the remainder of this chapter.
The Human Amygdala
and Emotional Memory
Lesion Studies For most brain-behavior relationships, our initial source of information on the role of the brain in human behavior, as well as on the relationship between the amygdala and emotional declarative memories, has come from patients with specific brain lesions. Though selective bilateral lesions of the amygdala are rare, the few reported cases are illustrative. In addition to bilateral temporal lobectomies—which are no longer performed due to the profound amnesia produced in patient H. M. (Scoville & Milner, 1957)—two disorders have been described that result in the bilateral destruction of the amygdala: Urbach-Wiethe disease and herpes simplex encephalitis. Urbach-Wiethe disease (also known as lipoid proteinosis) is a rare hereditary disorder characterized by the deposition of hyaline material in the skin and mouth areas and is associated with bilateral mineralization of medial temporal lobe structures in about half of the cases, with specific mineralization of the amygdala in some cases. Herpes simplex encephalitis is an inflammation of neural structures following viral infection. Even though this disease may result in widespread pathology throughout the nervous system, its earliest pathology almost invariably includes the amygdalae. Additionally,
47
unilateral temporal lobectomy is commonly employed in the surgical treatment of intractable epilepsy. Here, we report on studies of long-term declarative memory of emotional events from subjects with both bilateral and unilateral amygdala damage. Bilateral Amygdala Damage One well-characterized case of bilateral amygdala damage was originally reported by Tranel and Hyman (1990). The patient (SM046) had been diagnosed with Urbach-Wiethe disease. Computerized tomography (CT) and magnetic resonance imaging (MRI) confirmed the mineralization of both amygdalae, as well as minimal damage to anterior entorhinal cortices, but no other structural abnormality, an anatomical picture confirmed by functional imaging. Neuropsychological evaluation of this patient revealed normal general intellect, language, and verbal memory function. These results are in stark contrast to the profound declarative memory impairment following hippocampal damage (Squire, 1987). Follow-up research with patient SM046 and others with amygdala damage has highlighted the modulatory role in memory played by the amygdala. Markowitsch and colleagues examined emotional memory formation in two patients with Urbach-Wiethe disease, including bilateral mineralization of the amygdala (Babinsky et al., 1993; Markowitsch et al., 1994). Neuropsychological evaluation revealed that though neither patient was amnesic, they showed reduced performance on several standard neuropsychological tests of memory (e.g., Auditory Verbal Learning Test). Memory for emotional material was similarly depressed, with one patient (C. P.) showing more impairment than the other (B. P.). Specifically, in a word stem completion task of previously presented emotional or neutral words, C. P. remembered the neutral words better than the emotional words, whereas B. P. showed roughly equivalent performance for neutral and emotional stimuli, as compared to control subjects who tended to remember the emotional stimuli better. Similarly, in a recognition test of previously presented emotional pictures, C. P. recognized neutral pictures better than emotional pictures in contrast to normal controls who showed enhanced recognition performance of the emotional materials. Testing one of the same patients previously mentioned (B. P.), Cahill, Babinsky, Markowitsch, and McGaugh (1995) showed that this patient did not show the normal enhancement of memory for an emotionally arousing slide show. In the task used in this study, originally described in Cahill and McGaugh (1995; see also Heuer & Reisberg, 1990), all participants were exposed to a slide show that included both neutral and emotionally arousing slides. Age-matched control subjects recalled the emotionally arousing stimuli much better than the neutral stimuli, but patient B. P. showed equivalent memory for both types of stimuli. Patient B. P. rated his emotional reaction to the slide show similarly to control participants (B. P.’s rating: 8, control’s rating: 7.25 ± 1.2, on a scale of 0 to 10). Even though B. P. rated his emotional experience the same as
48
that of control subjects, this emotional experience did not translate into enhanced memory. Using the same task, Adolphs, Cahil, Schul, and Babinsky (1997) also tested the performance of an additional patient with bilateral amygdala damage (SM046). This study featured analysis of memory for each individual slide for both SM046 and for B. P. Whereas normal participants remembered emotionally arousing slides significantly better than the neutral slides, neither amygdaladamaged patient showed this pattern (see fig. 2.1). Specifically, the slide remembered best by the control participants (a slide showing the surgically reattached legs of a car crash victim) was the one on which the two patients deviated most from the controls’ scores (see fig. 2.1A). Like B. P., SM046 also endorsed normal ratings of subjective emotional arousal for the story. The pattern of impaired facilitation of memory for emotionally arousing material in the face of apparently normal memory performance for neutral material in subjects with bilateral amygdala damage contrasts sharply with the performances of amnesic subjects. Subjects with hippocampal or diencephalic amnesia are impaired in their overall memory performance, regardless of the nature of the material, but show a normal enhancement (albeit of smaller magnitude) when the subject matter is emotionally arousing (Hamann, Cahill, & Squire, 1997). Unilateral Amygdala Damage Using the same task, Adolphs, Tranel, and Denburg (2000) examined the pattern of emotional memory performance following unilateral amygdala damage. Eight subjects with unilateral amygdala damage consequent to temporal lobectomy (6 left; 2 right), 9 brain-damaged controls with no damage to the anterior temporal lobe, and 7 normal controls participated in the study. In this experiment, each slide in the story was rated on scales of emotional valence, arousal, unusualness, and complexity. There were no group differences in slide ratings. As in the previous study, both normal controls and brain-damaged controls showed enhanced memory for phase 2 of the slide/narrative story, specifically for the most highly arousing slide. By contrast, the group with left amygdala damage failed to show enhanced memory for this slide, showing the same pattern as previously reported for subjects with bilateral amygdala damage. (The two subjects with right amygdala damage appeared to perform normally, but these findings are inconclusive due to the small sample size of this group.) These findings point to a role for the left amygdala in the consolidation of declarative memory for emotionally arousing stimuli. Further studies have addressed the role of unilateral amygdala damage in the formation of memory for emotional words. Phelps, LaBar, and Spencer (1997) examined emotional memory formation in 26 subjects following unilateral temporal lobectomy. In this study, subjects were presented with a list of 27 words (9 positive, 9 negative, and 9 neutral) while skin conductance responses (SCRs) were recorded. A surprise recall test was administered 1 minute after the presentation of the word list. Results illustrated that the left temporal lobectomy
49
Figure 2.1. A. Plots of standard deviations from normal of patient SM046 across all 15 picture stimuli. B. Raw data for patient SM046 and normal controls. (Reprinted from “Impaired declarative memory for emotional material following bilateral amygdala damage in humans,” Adolphs, Cahill, Schul, & Babinsky, 1997, Learning & Memory, 4, 291–300. Copyright © 1997 by Cold Spring Harbor Laboratory Press.)
group had the worst recall of the word list, but this effect was not statistically significant. Each group (controls and both temporal lobectomy groups) recalled the negative and positive words better than the neutral words, but there was no difference among the groups in terms of the pattern of word recall. The authors noted that the words used in the negative and positive categories (e.g., victim, comedy) were perhaps not salient enough to produce the emotional arousal necessary to show any group differences in emotional memory performance. In
50
fact, psychophysiological data from SCR, a measure of autonomic arousal, illustrated that the neutral word category elicited a greater response than did either emotion word category. In light of these findings, these investigators reported an additional study designed specifically to address the role of the arousal dimension in memory consolidation (LaBar & Phelps, 1998). In this follow-up study, 22 temporal lobectomy patients (10 left, 12 right) were presented with a list of 40 words (20 arousing, 20 neutral). The arousing words chosen in this study consisted of “profanities, sexually explicit words, and words depicting social taboos.” These words were successful in producing increased SCRs over neutral words and were rated as significantly more arousing than neutral words in controls, as well as in both right and left temporal lobectomy groups. Free recall for the words was assessed both immediately and at 1–hour postencoding. The results were assessed in terms of forgetting rates between the immediate and delayed free recall tests. Only the control group showed a differential forgetting rate for the arousing versus neutral words, showing increased recall for the arousing words at the delayed recall test. Both the right and the left temporal lobectomy groups showed decreased memory for the arousing words at delayed recall. Consistent with the two previously reported studies, in this study the left temporal lobectomy group showed the poorest levels of recall; their performance was significantly worse than that of both controls and the right temporal lobectomy group. This effect was not, however, specific to emotionally arousing material, as the left temporal lobectomy group showed reduced overall performance but not a specific impairment on memory for emotional stimuli. Although these results are consistent with previous research showing a left-hemisphere dominance for verbal memory, they do not address a specific role of the right or left amygdala in memory for verbal or visual emotional stimuli. A recent study in our lab has attempted to address this issue (Buchanan, Denburg, Tranel, & Adolphs, 2001). Participants consisted of 20 subjects with unilateral amygdala damage following temporal lobectomy for treatment of epilepsy (11 left, 9 right) and 25 brain-damaged controls with unilateral lesions outside the temporal lobe. Additionally, 35 normal control volunteers were recruited for participation. Participants were tested on 2 days; on the first session 15 pictures differing in emotional salience (5 pleasant, 5 unpleasant, 5 neutral) were presented along with a one-sentence verbal narrative description (e.g., accompanying a picture of two parents with their new twin babies was the narrative: “After the babies were born, both parents were very happy, although a bit exhausted”). Subjects were told to watch attentively while the emotional responses to the stimuli were recorded; no mention of a follow-up memory test was made. Twenty-four hours after the first session, subjects’ memory for the slides was assessed with free recall, multiple choice, and four-alternative forcedchoice recognition tests. Within both free recall and multiple choice tests, memory for narrative and picture information was assessed. The recognition
51
test assessed only recognition of visual detail. Results illustrated that the group with left amygdala damage was specifically impaired on memory for emotional narratives relative to memory for neutral narratives. Interestingly, this group was not impaired on memory for emotional picture information. The right amygdala group, on the other hand, was impaired on visual recognition memory; however, this impairment was not specific to emotional pictures (see fig. 2.2). These findings support a material-specific role of the left amygdala in the processing of verbal emotional stimuli and a role of the right amygdala in processing visual emotional stimuli, corroborating previous work describing the separable language versus visuospatial processing roles for the left and right hemispheres, respectively (Dobbins, Kroll, Tulving, Knight, & Gazzaniga, 1998). These data replicate previous work illustrating a deficit in verbal emotional memory in individuals with left amygdala damage (LaBar & Phelps, 1998; Phelps et al., 1997) while illustrating a lateralized pattern of the amygdala’s influence on emotional memory.
The Human Amygdala and Memory
for Gist Versus Detail
Memory for gist as well as for peripheral details has been a major topic in the study of the effects of emotion on memory (Christianson & Loftus, 1991; Heuer & Reisberg, 1990). Despite discrepancies in the findings, several studies have shown that emotional arousal enhances memory for gist, but not memory for detail (Burke, Heuer, & Reisberg, 1992; Reisberg & Heuer, 1992; see also chapter 1 here). More work is needed in operationalizing these constructs; however, the basic idea is that gist or central information pertains to the most salient, relevant aspects of a stimulus (the aspects that one would focus on when describing it to someone else and whose alteration would change the meaning of the stimulus), whereas detail or peripheral information is everything else. Clearly, these categories do not form a strict dichotomy, but some features of a stimulus are almost unanimously deemed gist, whereas others are described as detail. Recent work in our lab has addressed the role of the amygdala in memory for the gist/central information versus the peripheral details of emotional stimuli. Using a task similar to that already described (Buchanan et al., 2001), Adolphs, Denburg, and Tranel (2001) assessed memory for the gist and details of emotionally negative and neutral slides and narratives from patient SM046 as well as from patients with unilateral temporal lobectomy (12 left, 8 right), braindamaged controls (n = 15), and age-matched controls (n = 47). Memory for gist was operationalized as salient, general information of the stimulus sufficient to distinguish that particular stimulus from all the other stimuli and that did not depend on remembering details of the scene; it was assessed with a fouralternative forced-choice written questionnaire. Detail memory was defined as
52
Figure 2.2. Free recall for narratives and pictures across valence categories. A. Mean (± standard error) of correctly recalled narratives across all subjects from each valence category. B. Mean (± standard error) of correctly recalled pictures across all subjects from each valence category. (Reprinted from “Verbal and nonverbal emotional memory following unilateral amygdala damage,” Buchanan, Denburg, Tranel, & Adolphs, 2001, Learning & Memory, 8, 326–335. Copyright © 2001 by Cold Spring Harbor Laboratory Press.)
53
information that could be accessed only from a detailed memory of the visual image and was assessed using a four-alternative forced-choice visual recognition task showing the original visual stimulus and three computer-manipulated foils that differed only in their details. The study found that all groups (including those with unilateral temporal lobectomies) showed enhanced memory for the gist of negative rather than neutral stimuli, whereas patient SM046 showed the opposite pattern, with greater memory for the gist of the neutral than the negative stimuli. Performance on detail memory, on the other hand, showed a different pattern. Both control groups (age-matched and brain-damaged controls), as well as the left temporal lobectomy group, showed greater detail memory for the neutral than the negative stimuli, whereas SM046 and the right temporal lobectomy group remembered the details of both the negative and neutral stimuli equivalently. The general pattern of memory performance on this task in the control subjects was that emotionally negative stimuli enhanced gist memory but reduced detail memory, compared to neutral stimuli. Bilateral amygdala damage, on the other hand, interfered with both of these effects: SM046 remembered gist for unpleasant stimuli relatively worse than controls, and her detail memory performance for negative stimuli was better than that of controls (see fig. 2.3). This study had an inherent confound precluding an unambiguous interpretation of the data; gist and detail memory were assessed with different methods (questionnaire and visual recognition, respectively). Current studies under way in our laboratory are providing data confirming that the effects reported here can really be attributed to the amygdala’s role in processing gist and detail information. These findings provide evidence that the human amygdala modulates declarative memory for emotionally arousing stimuli through differential effects on memory for gist and for visual detail. The data are consistent with the idea that the amygdala acts as a filter in the encoding of relevant information from emotional stimuli. Whereas the healthy amygdalae are able to enhance the memory processing of the gist of these stimuli and disregard the irrelevant details, damage to this structure impairs this ability. This pattern is consistent with recent work showing that the amygdala is involved in the enhancement of perception of emotionally salient events (Anderson & Phelps, 2001; Öhman & Mineka, 2001), suggesting that both perceptual and mnemonic processing of unpleasant events critically involve the amygdala. Findings from these studies indicate that the emotional arousal, not the pleasantness of the stimuli, most determines whether a stimulus will be remembered (see Bradley, Greenwald, Petry, & Lang, 1992). This arousal-mediated memory enhancement is the feature that appears to be most affected following amygdala damage (it is also arousal that is most predictive of amygdala activity in functional imaging studies, as we discuss in the next section). These findings accord with recent neuroimaging work showing amygdala activity while viewing both negative and positive emotional stimuli, whether pictures (Hamann,
54
Figure 2.3. A. Memory for gist. Mean (± standard error) number of questions (out of a maximum of two) answered correctly per stimulus, for the two emotion categories. All subjects except SM046 showed the same pattern: superior memory for emotionally aversive stimuli compared with neutral stimuli. B. Memory for visual detail. Subjects’ mean percentage correct on a four-alternative forced choice recognition memory task. The four stimuli consisted of the original stimulus and three computer manipulated foils that were identical in gist but differed in detail. Whereas controls remembered the details of aversive stimuli less well than those of positive stimuli, subjects with amygdala damage remembered details about both types of stimuli equally well. (Reprinted from “The amygdala’s role in long-term declarative memory for gist and detail,” Adolphs, Denburg, & Tranel, 2001, Behavioral Neuroscience, 15, 983–992. Copyright © 2001 by the American Psychological Association, Inc.)
55
Ely, Hoffman, & Kilts, 2002) or words (Hamann & Mao, 2002). Together, these lines of work suggest that the primary variable that determines memory for emotional material is arousal and also that this variable is the most salient instigator of amygdala activity. The principal difficulty here has been in constructing stimuli of pleasant valence that can be as arousing as those of negative valence.
Neuroimaging of the Amygdala
and Emotional Memory
Functional neuroimaging has provided another tool with which to examine the role of the human amygdala in the formation of emotional memories. Several recent studies have been able to test specific hypotheses derived from both animal research and studies in humans with amygdala lesions using positron emission tomography (PET) and functional magnetic resonance imaging (fMRI). The first study to examine the role of the human amygdala in the formation of emotional memories using functional neuroimaging was conducted by Cahill et al. (1996). Healthy participants viewed emotionally arousing and neutral videos during PET scanning. Three weeks later, participants were asked to recall all the information that they could remember from each video. As expected, participants recalled significantly more information from the emotionally arousing video than from the neutral video. Correlation analyses revealed a significant positive correlation between the glucose metabolic rate of the right amygdala and the number of emotional film clips recalled (r = 0.93; see section on gender differences for discussion of a lateralization of amygdala activity in a similar task). Further analyses showed no such association with recall of neutral film clips. These findings suggest that the amygdala is activated during the encoding of emotionally arousing events and is involved in the translation of these events into long-term memory. A follow-up study corroborated the finding that amygdala activity is not involved in the formation of declarative memory for nonemotional material (Alkire, Haier, Fallon, & Cahill, 1998). This study did, however, document an association between hippocampal activity and the formation of memory for a nonemotional word list. These two studies further illustrate the dissociation of memory functions between the hippocampus and amygdala and highlight a specific role of the amygdala in the formation of emotional memories. Subsequent studies have extended these findings to show an association between bilateral amygdala activity during encoding and memory for both emotionally pleasant and unpleasant stimuli using both PET (Hamann, Ely, Grafton, & Kilts, 1999) and fMRI (Canli, Zhao, Desmond, Glover, & Gabrieli, 1999). The study by Hamann et al. additionally tested the emotional specificity of the amygdala’s influence on memory by including a stimulus category of interesting and unusual pictures. These interesting pictures included a chrome rhinoceros and a scene from a surrealist painting. Presumably, if the amygdala is involved in general memory enhancement, then amygdala activity should be
56
associated with memory for unusual yet nonemotional stimuli. Results from this study illustrate that while these unusual pictures were better remembered than neutral (e.g., a book) and pleasant pictures (e.g., opposite-sex nudes) 4 weeks after encoding, this enhancement was unrelated to amygdala activity, which was instead specifically related only to memory for aversive and pleasant stimuli (a finding also consistent with studies in monkeys, which have failed to find any effect of amygdala lesions on the von Restorff effect—enhanced memory for especially unusual or distinctive stimuli in a set; Parker, Wilding, & Akerman, 1998). These results highlight the specific role of the amygdala in the enhancement of memories for emotionally significant material regardless of valence (see fig. 2.4). A recent event-related fMRI study has shown that stimuli rated as emotionally intense are associated with increased amygdala activity and increased memory performance (Canli, Zhao, Brewer, Gabrieli, & Cahill, 2000). In this study, 10 female volunteers were exposed to a selection of neutral and emotionally negative pictures (e.g., scenes of mutilation) while the fMRI response was recorded for each picture. Immediately after viewing each picture, subjects rated their emotional response on a scale from 0 (not emotionally intense at all) to 3 (extremely emotionally intense). Analysis of the fMRI response illustrated that bilateral amygdala activity was correlated with increased ratings of emotional intensity, such that the greater the emotional intensity, the greater the amygdala response. Three weeks later, subjects returned for a surprise memory test in which they were asked to report whether they were certain that they remembered a slide, whether the slide seemed familiar, or whether they did not remember the slide (in a remember/know recognition paradigm). Performance data illustrated that those slides rated as extremely emotionally intense were remembered significantly better than those rated as less intense. Additionally, the degree of left amygdala activation during picture encoding was correlated with subsequent memory for the pictures (see section on gender differences for discussion of lateralization of amygdala activity related to memory performance). Those pictures that tended to produce the greatest response in the left amygdala were also remembered most often.
Autobiographical Memories and
the Amygdala
In addition to the evidence reviewed, that the amygdala plays a role in the encoding and consolidation of emotional memories, there is evidence that the retrieval of autobiographical memories may depend, in part, on the amygdala. Likely, the majority of our distant autobiographical memories are associated with an emotional response, suggesting that perhaps the amygdala plays a role both in the encoding and retrieval of these memories by virtue of their emotionally arousing nature. Mori et al. (1997) assessed memory for autobiographical events
57
Figure 2.4. Top. Brain activity correlated with memory enhancement. Maps of pixels in which individual subject rCBF was significantly correlated with individualsubject episodic memory enhancement superimposed on an axial MRI image. Left: correlation map for pleasant stimuli at z = -10.5. Right: correlation map for aversive stimuli at z = -16.5. Note that in this figure the right hemisphere is shown on the right. Bottom. Relationship between pleasant-picture memory and brain activity for individual subjects. Correlation scatterplots for the pleasant picture memory with rCBF. a. Left amygdala. b. Right amygdala. c. Left hippocampus. d. Right hippocampus. (Reprinted with permission from “Amygdala activity related to enhanced memory for pleasant and aversive stimuli,” Hamann, Ely, Grafton, & Kilts, 1999, Nature Neuroscience, 2, 289–294. Copyright © 1999 by Nature America Inc.)
58
surrounding the 1995 Kobe earthquake in patients with probable Alzheimer’s disease. Neuropathological studies have shown that the characteristic plaques and neurofibrillary tangles associated with Alzheimer’s disease are often located in the amygdala, and atrophy of this structure has been reported in both autopsied patients (Scott, DeKosky, & Scheff, 1991) and in vivo in patients with early signs of the disease (Cuenod et al., 1993). Mori et al. examined the preexisting individual differences in the amygdala volume of a group of patients with putative Alzheimer’s disease (diagnosed prior to the earthquake) to determine whether changes in the size of this structure influenced emotional autobiographical memory. Results from this study illustrate that the volume of the amygdala was positively correlated with patients’ memories for their experiences during and after the earthquake (see fig. 2.5). Hippocampal volume was also correlated with emotional memory but not as strongly as the relationship between amygdala volume and emotional memory. This association between amygdala volume and emotional memory was significant even when controlling for factors such as age, education, whole brain volume, and ratings of dementia. There was no relationship between the size of the amygdala and memory for general (nonemotional) knowledge about the events surrounding the earthquake. These findings extend work from laboratory research focusing on the amygdala and emotional memory
8
Total Emotional Memory Score
Total Emotional Memory Score
8
6
4
2
0 600
1200 1600 2000 2400 Amygdalar Volume (mm3)
6
4
2
0 1600 2000 2400 2800 3200 3600 Hippocampal Volume (mm3)
Figure 2.5. Scatterplots of amygdala and hippocampal volumes with total emotional memory scores of 36 patients with Alzheimer’s disease. Pearson correlation analysis showed significant correlations between amygdala volume and total emotional memory score and between hippocampal volume and total emotional memory score. After controlling for the effects of age, sex, education, whole brain volume, and disease severity, the former correlation remained significant, while the latter was no longer significant. (Reprinted with permission from “Amygalar volume and emotional memory in Alzheimer’s disease,” Mori, Ikeda, Hirono, Kitagaki, Imamura, & Shimomura, 1999, American Journal of Psychiatry, 156, 216–222. Copyright © 1999 by the American Psychiatric Association, Inc.).
59
and suggest that memory for real-life emotional situations depends on the integrity of this structure. A hypothesis has been proposed suggesting that the temporofrontal cortical areas including the amygdala and uncinate fasciculus fiber bundle connecting these areas are specifically involved in the retrieval of autobiographical memories (Kroll, Markowitsch, Knight, & von Cramon, 1997; Markowitsch et al., 2000). More specifically, a right-sided temporofrontal network purportedly retrieves autobiographical memories, while a left-sided temporofrontal network encodes this information into memory (Tulving, Kapur, Craik, Moscovitch, & Houle, 1994; Tulving et al., 1994). This hypothesis has found support from studies of patients with lesions of these areas (Kroll et al., 1997; Levine et al., 1998), as well as functional neuroimaging studies of healthy participants (Fink et al., 1996; Markowitsch et al., 2000). Fink et al. showed that when subjects listened to transcripts of their own autobiographical memories, activity in areas including the right amygdala, right hippocampus, and right prefrontal cortex increased significantly. The authors of several of these studies suggest that the role of these areas in the retrieval of memory is, in part, due to the emotional nature of autobiographical memories (Markowitsch et al., 2000). Memories from our lives that we are able to retrieve presumably have some affective character, thus making these memories salient for recollection. These studies, in combination with previous work on emotional memory encoding, suggest that the amygdala may be involved not only in encoding emotional information but also in its retrieval. One note of caution regarding the studies reviewed in this last section is that they have not pinpointed the observed effects solely to the amygdala but may reflect structures in the immediate vicinity of the amygdala including the temporal pole.
Gender Differences in
Amygdala Activity
Findings from several of these studies have suggested the possibility of gender differences related to lateralized amygdala activity in memory for emotional visual stimuli. In studies on men, results have shown a predominantly right-sided activation of the amygdala (Cahill et al., 1996; Hamann et al., 1999), whereas two studies documented more left lateralized activation in women (Canli et al., 1999, 2000). These gender differences have recently been investigated directly in a study including both men and women (Cahill et al., 2001) and using experimental conditions identical to those previously used with men only (described in the section on neuroimaging; Cahill et al., 1996). Results from this study illustrated the same gender-specific lateralized pattern of activation previously documented across separate studies, with enhanced emotional memory performance correlating with right amygdala activity in men but left amygdala activity in women (Cahill et al., 2001). Another interesting facet of this work is the finding that women show better memory for emotionally arousing material than do men
60
(Seidlitz & Diener, 1998). Canli, Desmond, Zhao, and Gabrieli (2002) examined the possible neural mechanisms of women’s enhanced memory for emotional material. In a functional MRI study similar in methodology to previous experiments using emotional picture stimuli (Cahill et al., 2001; Canli et al., 2000), the authors replicated the finding of women’s enhanced memory for emotional stimuli. Additionally, they found that women showed a greater network of neural activity related to emotional memory (including the left amygdala, hippocampus, and frontal cortex) than men (whose activity was primarily located in the right amygdala). Subjective ratings of emotional arousal were equivalent between men and women, so this finding suggests that gender differences in emotional responsiveness alone are not responsible for this cognitive difference. The authors of these studies suggest that these gender differences may reflect different cognitive strategies between men and women in the processing of these stimuli, perhaps related to differential attention to gist and to peripheral details, a cognitive phenomenon that also shows a gender difference (see Cahill & van Stegeren, 2003). These intriguing findings clearly warrant the inclusion of gender and laterality as factors in future work on the relationship between the amygdala and emotional memory. [For another perspective on gender differences in emotional memory, see chapter 8—Eds.]
The Human Hippocampus
and Emotional Memory
The role of the hippocampus in the enhancement of memory by emotion has been assessed in several studies. Two studies by Hamann, Cahill, McGaugh, and Squire (1997) and Hamann, Cahill, and Squire (1997b) assessed memory for emotionally arousing materials in amnesic patients with damage to either the hippocampus only, damage to the hippocampus and amygdala, or damage to the diencephalon. Results from these studies illustrated that damage to the hippocampus alone or damage to the diencephalon leaves the enhancement of declarative memory by emotional arousal intact, presumably by relying on the intact amygdala in the subjects included in these studies. Two patients with combined damage to the hippocampus and amygdala (including overlying cortices) showed no discernible memory for the emotionally arousing or neutral stimuli. A neuroimaging study (mentioned earlier in the section on the role of the amygdala; Alkire et al., 1998) has demonstrated that activity of the hippocampus, but not the amygdala, correlates with memory for nonemotional materials. In another neuroimaging study, Hamann et al. (1999) addressed the relationship between the amygdala and hippocampus in the enhancement of memory by emotion. Bilateral activity in both the amygdala and hippocampus was associated with memory enhancement for both pleasant and aversive stimuli in this study. Results from the studies with amnesics illustrate that the
61
hippocampus, while necessary for the formation of new declarative memories, is not itself responsible for the modulation of memory by emotional arousal. The neuroimaging work shows that correlated activity between the amygdala and hippocampus may be a necessary mechanism for the enhancement of memory for emotionally arousing events. This work further demonstrates the modulatory role of the amygdala, which in combination with the hippocampus lays down a stronger memory trace for emotional situations than for nonemotional situations.
The Prefrontal Cortex
and Emotional Memory
The prefrontal cortex, notably its dorsolateral regions, is known to be involved in both the encoding and retrieval of memories (Buckner & Wheeler, 2001; Tulving, Kapur, Craik, et al., 1994) and is thought to provide “executive” strategies for recollecting, comparing, and ordering memories. Work from numerous investigators has also suggested a role of the orbitofrontal region in emotional processing (Bechara, Damasio, & Damasio, 2000; Rolls, 2000b). As dorsolateral and orbital sectors of the prefrontal cortex are intimately connected (Rolls, 2000a), and as orbitofrontal cortex is also massively connected with the amygdala, it is natural to hypothesize that networks within the prefrontal cortex participate in structuring emotional memories as well. Whether and how the prefrontal cortex may be involved in the modulation of memory by emotion has been less well studied than the case of the amygdala, but several studies have begun to provide some hints. Numerous studies have documented the effects of orbitofrontal damage on decision making (Bechara et al., 2000; Bechara, Damasio, Damasio, & Lee, 1999; Bechara, Damasio, Tranel, & Damasio, 1997; Bechara, Tranel, Damasio, & Damasio, 1996). This work has outlined a model whereby the orbitofrontal cortex processes emotional signals from the body that affect decision-making processes (Damasio, 1994). One study has sought to distinguish this decision-making bias from the enhancement of memory by emotion (Bechara et al., 2000). Patients with orbitofrontal damage were presented with a series of neutral and emotionally arousing picture stimuli in four sets. One set of pictures was presented only once, another set was presented twice, a third set was presented three times, and a fourth set was presented eight times. Memory for these stimuli was assessed immediately after all stimuli had been presented. Results from this study illustrated that while the orbitofrontal patients showed a reduced overall level of memory, these patients still showed enhanced memory for the emotionally arousing pictures rather than for the neutral pictures, and both groups showed improved memory for those stimuli that were repeated most often. This study suggests that the biasing of decision making by emotion that depends on the orbitofrontal cortex is separable from the influence of emotion on memory, which appears to rely less on the orbitofrontal cortex.
62
Recent neuroimaging studies have documented a role of the prefrontal cortex in the memory for neutral material encoded in an emotionally arousing context (Maratos, Dolan, Morris, Henson, & Rugg, 2001; Maratos & Rugg, 2001). For example, these studies give the example of the word “corn” used in a negative context: “The farmer was shredded when he fell into the corn grinder,” and used in a positive context: “The farmer was overjoyed with his bountiful crop of corn.” Using electrophysiological recordings, the authors in these two studies showed that the areas that are normally activated during the retrieval of old neutral words (including the right prefrontal cortex) exhibited enhanced activity when these words were presented in a negative context (Maratos & Rugg, 2001). In an event-related fMRI study using a similar experimental design (with the addition of a positive emotional context), Maratos et al. (2001) demonstrated that the re-presentation of neutral words that had been presented in a positive emotional context resulted in heightened activity in the orbitofrontal cortex, the dorsolateral prefrontal cortex, and the medial temporal lobe areas such as the amygdala and hippocampus. Results from these studies suggest (1) that activity within areas normally involved in the retrieval of neutral information (including the prefrontal cortex) is enhanced by an emotionally arousing context and (2) that the same areas activated during encoding of emotionally arousing materials (including the amygdala) are active during the retrieval of these materials. These studies from both lesion and neuroimaging paradigms illustrate the complex relationship between the prefrontal cortex and the enhanced memory for emotional material. Though the role of the PFC in emotional memory has not been well documented in lesion patients, the neuroimaging data suggest a subtle role for this region in the encoding and recognition of emotional stimuli. It is worth emphasizing the close connectivity between the amygdala, orbitofrontal cortex, basal forebrain, and the dorsal prefrontal cortex. Studies in monkeys have demonstrated that the amygdala and orbitofrontal cortex are two essential components of a system for processing the emotional properties of stimuli; disconnection of the two structures can result in impairments on certain tasks of associative emotional memory that are as severe as lesions in either structure (Gaffan, Murray, & Fabre-Thorpe, 1993). It seems plausible to suggest the same systems-level architecture for declarative emotional memory.
Neurodegeneration
and Emotional Memory
A final section to our review concerns the insights gleaned not from focal lesion or activation studies but from the more diffuse types of damage incurred by neurodegenerative diseases (see chapter 9 here for a discussion of these conditions in relation to aging). Alzheimer’s disease (AD) is initially characterized by memory
63
complaints, often concomitant with alteration in emotional behavior. The plaques and neurofibrillary tangles—the neuropathological hallmarks of AD—are often first noted in medial temporal lobe structures including the amygdala, entorhinal cortex, and hippocampus (Hyman, Van Hoesen, & Damasio, 1990; Van Hoesen, Augustinack, & Redman, 1999). These findings have led several investigators recently to examine emotional memory in AD. Results from two studies have shown impairments in the enhancement of memory by emotion in patients with probable AD (Abrisqueta-Gomez, Bueno, Oliveira, & Bertolucci, 2002; Hamann, Monarch, & Goldstein, 2000), yet two other studies have shown no such impairment (Kazui et al., 2000; Moayeri, Cahill, Jin, & Potkin, 2000). The major difference between the studies showing impairment and those showing no impairment is that the latter studies used a slide show paradigm similar to that described in Cahill, Prins, Weber, and McGaugh (1994), whereas the former studies showing an impairment used individual emotional pictures devoid of a narrative. This difference in methodology deserves attention in future research in emotional memory. Perhaps a cohesive narrative structure is necessary to keep such patients’ attention, whereas the presentation of unrelated emotional pictures may not be salient enough (see discussion of the distinction between visually arousing and thematically arousing influcences on memory in chapter 1 for a fuller explication of this possibility). The aforementioned study by Mori et al. (1999) that demonstrated an association between amygdala volume and emotional memories for the Kobe earthquake in AD patients suggests that, at least in some of these patients, the volumetric loss of the amygdala plays a role in the ability to remember emotional events (see previous section on autobiographical memories and the amygdala for fuller description of this study). The same neural structures (amygdala and hippocampus) that have been described as playing a role in emotional memory in neurodegenerative diseases are also known to be affected in normal aging. Specifically, the volumes of both the amygdala (Jack et al., 1997; Mu, Xie, Wen, Weng, & Shuyan, 1999) and hippocampus (Golomb et al., 1993; Jernigan et al., 2001) diminish with age. In line with this reduction in brain volume are the well-documented memory impairments that accompany aging (Grady & Craik, 2000). The effect of aging on memory for emotional material has not been well studied. A recent study from our laboratory has sought to examine this issue (Denburg, Buchanan, Tranel, & Adolphs, in press). In the study, 80 healthy adults between the ages of 35 and 85 years were presented with a series of pictures that ranged from pleasant to neutral to unpleasant scenes (the same ones mentioned in connection with the studies by Buchanan et al. [2001] and Adolphs et al. [2001] on pages 51 and 52). Memory for these pictures was then tested 24 hours later. Results from this study showed that regardless of the age of the subject, memory for the emotionally arousing pictures was better than that for neutral pictures. Though there was an overall decline in memory function with age, there was only a small difference in the degree of memory modulation by emotion across the age range. These
64
findings indicate a sparing of the influence of emotion on memory with age, perhaps due to a more pronounced age-related degeneration of the hippocampus than for the amygdala. Studies designed to assess the relationship between age-related declines in brain volume and emotional memory function are necessary to test this hypothesis.
Summary and Integration The studies reviewed in this chapter illustrate the complex interrelationships among brain structures involved in the modulation of memory by emotion. Numerous brain structures—including the amygdala, hypothalamus, hippocampus, cingulate, insular and orbitofrontal cortices, among others—are involved in both the evaluation of and response to emotional stimuli. Findings from lesion studies and neuroimaging point to the amygdala as the most critical structure for the enhanced memory of stimuli encoded during emotional arousal. Our focus on the amygdala’s (and other structures’) role in modulation of declarative memory for stimuli does not exclude its role in other aspects of “emotional memory,” more broadly construed. In our view, the amygdala (together with structures like the ventral striatum and orbitofrontal cortex) helps to mount a coordinated, multisystem response to an emotional challenge. Such a response consists of concerted changes in many parameters of an organism’s body (autonomic, endocrine, and other changes) and brain. The changes in brain functioning are also multicomponent, and include, among others, changes in attention, alertness, and—the focus of this chapter—memory encoding and consolidation. The known neuroanatomical connectivity of the amygdala bears out this sketch; it receives rudimentary (LeDoux, 2000) as well as highly processed sensory information (Amaral, Price, Pitkanen, & Carmichael, 1992), which allows for quick and specific activation of response systems that influence behavior and subsequent memory formation. The specific role of the amygdala appears to be that of instigator, the catalyst that initially imbues incoming stimuli with emotional significance and triggers other neural regions to react accordingly. The detection and experience of emotion increases the processing of the emotional stimuli by focusing attention and enacting response systems (Frijda, 1987; Lang, Bradley, & Cuthbert, 1990). The amygdala plays an integral role in this enhanced processing through influence over numerous brain structures, including the hippocampus, hypothalamus, brain stem arousal centers, basal forebrain, and bidirectional connections with neocortex (McGaugh, 2000; Packard & Teather, 1998; Pitkänen, 2000). Along with the integrative, multisystem role played by structures like the amygdala in coordinating emotional responses, its role in memory is equally diverse. We can think of the ways in which emotion influences memory as changes in memory for (1) the emotion-inducing sensory properties of stimuli themselves (e.g., remembering that a picture showed mutilated bodies), (2) the
65
intrinsically neutral sensory properties of stimuli associated with a related emotional state (e.g., remembering that one saw a picture of landscape as part of story about a terrible disaster), (3) the emotional state itself in the perceiver (e.g., remembering that one felt upset), or (4) higher-order information about the emotion and the stimuli (e.g., remembering simply that there were some emotional stimuli or that one felt an emotion). None of these is mutually exclusive, and there is evidence to suggest that the amygdala may contribute to all of them. Interesting questions for the future remain about how these aspects of information are related and how they might depend differentially on certain brain structures. The findings we have reviewed place the amygdala in a modulatory position of the traditional medial temporal lobe memory system, its activity influencing memory formation only under conditions of emotional arousal. In response to emotion, the same brain structures are recruited in the service of both response output and memory formation. The finding that enhanced activity of response systems—autonomic, endocrine, and skeletal motor activity—accompanies increased mnemonic representation illustrates the coordinated activity of these brain structures in the service of survival. The lesion studies described suggest that the amygdala is necessary for the enhancement of memory for emotionally arousing stimuli. The dysfunction of one or both amygdalae (because of disease or surgery) results in reduced memory for emotionally arousing materials. The precise nature of such dysfunction—whether it is hyper- or hypoactivation— can lead either to compromised emotional memories or to the pathological exaggeration of such memories, as seen in traumatic and phobic memories that we have not reviewed here (Davis, Walker, & Lee, 1997; Öhman & Mineka, 2001). The nature of these studies has made it difficult for the assessment of consolidation or retrieval effects but clearly implicate the amygdala in the encoding of emotional material into memory. Similarly, functional neuroimaging studies have most consistently pointed to encoding as the memory stage at which the amygdala exerts its effects (but see Dolan, Lane, Chua, & Fletcher, 2000). The most consistent finding from the discussed studies is an association between amygdala activity during encoding and the enhancement of subsequent memory. The activation of the amygdala would seem to be a potent predictor of memory performance. Another pathway through which the amygdala may influence memory performance is via hormonal output (see chapter 3 for a more thorough discussion of these issues). The previously described pharmacological studies in animals have illustrated the effect of manipulations of both catecholamines and glucocorticoids on emotional memory performance. Direct connections between the amygdala and the hypothalamus mediate the release of both epinephrine and glucocorticoids during emotionally arousing situations (Davis, 1997). Following their release, these hormones exert actions throughout the central nervous system and specifically at the amygdala (Honkaniemi et al., 1992; Shepard,
66
Barron, & Myers, 2000). A great deal of animal research has focused on the bidirectional nature of the relationship between the amygdala and stress hormones (McGaugh & Izquierdo, 2000; Roozendaal, 2000). Enhancement of either adrenergic or glucocorticoid activity improves, but blockade of these hormones reduces, memory performance. These findings have been documented in both animals (Roozendaal, 2000) and humans (Buchanan & Lovallo, 2001; Cahill et al., 1994). The final common pathway of these hormones’ effects on memory performance appears to be at the amygdala—specifically through noradrenergic neurotransmission in the lateral/basolateral nuclei (Quirarte, Roozendaal, & McGaugh, 1997). As previously mentioned, manipulations of this activity within these nuclei influence hippocampal and cortical function (Escobar, Chao, & Bermudez-Rattoni, 1998; Ikegaya, Nakanishi, Saito, & Abe, 1997), providing a potential mechanism whereby the actions of these hormones at the amygdala could influence the formation of emotional memories. Even though human research has yet to show a relationship among stress hormones, amygdala activity, and emotional memory, animal research suggests such a relationship; future work will no doubt focus on elucidating this topic. Consistent with the finding that systemic hormones influence subsequent amygdala activity, one mechanism through which an emotional response could influence memory is via perception of the physiological response in the body. Vagal stimulation in both animals (Clark et al., 1995) and humans (Clark et al., 1999) results in increased memory performance. In fact, the sensory role of the vagus nerve is proposed as one mechanism whereby peripherally released epinephrine—which does not readily pass the blood-brain barrier—influences amygdala function in the formation of emotional memories (McGaugh, 2000). These results illustrate that the influence of the amygdala on memory could occur at multiple stages: (1) through rapid initial responses influencing neural information processing with a short latency after the occurrence of a stimulus, or (2) indirectly, through first triggering an emotional response in the body and subsequent central effects of the body’s physiological state (of which some of the latter may then also be mediated by the amygdala, albeit at a later time than mechanism 1). Most current studies have not permitted a separation of the emotion depicted in the stimulus, from the emotional reactions and feelings of the subject, and an important goal for the future will be to disentangle these issues. In all likelihood, a variety of structures participate in modulating the encoding, consolidation, and retrieval of our memories for emotional events, with respect to diverse components of information and at multiple temporal scales. Structures such as the amygdala and the ventral striatum perhaps provide an initial, rapid, automatic evaluation of the emotional significance of stimuli, or of the context within which they occur, and can then consequently modulate our attention to those stimuli (e.g., via basal forebrain), the strength of their memory consolidation (e.g., via hippocampus), or the nature of their contextual links to autobiographical information and to other items stored in memory (e.g.,
67
via prefrontal regions). Perhaps, at a later point, many of the same structures can modulate memory through less direct mechanisms. Given the enormous and varied amount of declarative knowledge that humans store over their lifetime, it seems prudent to acknowledge that multiple, complex neural mechanisms would exist to encode, order, and efficiently store such knowledge.
Note We thank Antonio and Hanna Damasio, Daniel Tranel, and Natalie Denburg for their participation in some of the studies reviewed. Supported in part by an NRSA grant from the National Institute on Aging to T.W.B. and from the National Institute of Mental Health to R.A.
References Abrisqueta-Gomez, J., Bueno, O. F., Oliveira, M. G., & Bertolucci, P. H. (2002). Recognition memory for emotional pictures in Alzheimer’s patients. Acta Neurologica Scandinavica, 105, 51–54. Adolphs, R., Cahill, L., Schul, R., & Babinsky, R. (1997). Impaired declarative memory for emotional material following bilateral amygdala damage in humans. Learning & Memory, 4, 291–300. Adolphs, R., Denburg, N. L., & Tranel, D. (2001). The amygdala’s role in longterm declarative memory for gist and detail. Behavioral Neuroscience, 115, 983– 992. Adolphs, R., Tranel, D., & Denburg, N. (2000). Impaired emotional declarative memory following unilateral amygdala damage. Learning & Memory, 7, 180–186. Alkire, M. T., Haier, R. J., Fallon, J. H., & Cahill, L. (1998). Hippocampal, but not amygdala, activity at encoding correlates with long-term, free recall of nonemotional information. Proceedings of the National Academy of Science, USA, 95, 14506–14510. Amaral, D. G., Price, J. L., Pitkanen, A., & Carmichael, S. T. (1992). Anatomical organization of the primate amygdaloid complex. In J. P. Aggleton (Ed.), The amygdala: Neurobiological aspects of emotion, memory, and mental dysfunction (pp. 1–66). New York: Wiley-Liss. Anderson, A. K., & Phelps, E. A. (2001). Lesions of the human amygdala impair enhanced perception of emotionally salient events. Nature, 411, 305–309. Babinsky, R., Calabrese, P., Durwen, H. F., Markowitsch, H. J., Brechtelsbauer, D., Heuser, L., & Gehlen, W. (1993). The possible contribution of the amygdala to memory. Behavioural Neurology, 6, 167–170. Baxter, M. G., & Murray, E. A. (2000). Reinterpreting the behavioral effects of amygdala lesions in non-human primates. In J. Aggleton (Ed.), The amygdala: A functional analysis (Vol. 2, pp. 545–568). Oxford: Oxford University Press. Bechara, A., Damasio, H., & Damasio, A. R. (2000). Emotion, decision making and the orbitofrontal cortex. Cerebral Cortex, 10, 295–307. Bechara, A., Damasio, H., Damasio, A. R., & Lee, G. P. (1999). Different contributions of the human amygdala and ventromedial prefrontal cortex to decisionmaking. Journal of Neuroscience, 19, 5473–5481.
68
Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. (1997). Deciding advantageously before knowing the advantageous strategy. Science, 275, 1293– 1295. Bechara, A., Tranel, D., Damasio, H., & Damasio, A. R. (1996). Failure to respond autonomically to anticipated future outcomes following damage to prefrontal cortex. Cerebral Cortex, 6, 215–225. Bianchin, M., Mello e Souza, T., Medina, J. H., & Izquierdo, I. (1999). The amygdala is involved in the modulation of long-term memory, but not in working or shortterm memory. Neurobiology of Learning and Memory, 71, 127–131. Bradley, M. M., Greenwald, M. K., Petry, M. C., & Lang, P. J. (1992). Remembering pictures: Pleasure and arousal in memory. Journal of Experimental Psychology: Learning Memory and Cognition, 18, 379–390. Brown, R., & Kulick, J. (1977). Flashbulb memories. Cognition, 5, 73–99. Brown, S., & Schafer, A. (1888). An investigation in the functions of the occipital and temporal lobes of the monkey’s brain. Philosophical Transactions of the Royal Society of London, 179, 303–327. Buchanan, T. W., Denburg, N. L., Tranel, D., & Adolphs, R. (2001). Verbal and nonverbal emotional memory following unilateral amygdala damage. Learning & Memory, 8, 326–335. Buchanan, T. W., & Lovallo, W. R. (2001). Enhanced memory for emotional material following stress-level cortisol treatment in humans. Psychoneuroendocrinology, 26, 307–317. Buckner, R. L., & Wheeler, M. E. (2001). The cognitive neuroscience of remembering. Nature Reviews Neuroscience, 2, 624–634. Burke, A., Heuer, F., & Reisberg, D. (1992). Remembering emotional events. Memory & Cognition, 20, 277–290. Cahill, L., Babinsky, R., Markowitsch, H. J., & McGaugh, J. L. (1995). The amygdala and emotional memory. Nature, 377, 295–296. Cahill, L., Haier, R. J., Fallon, J., Alkire, M. T., Tang, C., Keator, D., Wu, J., & McGaugh, J. L. (1996). Amygdala activity at encoding correlated with long-term, free recall of emotional information. Proceedings of the National Academy of Science, USA, 93, 8016–8021. Cahill, L., Haier, R. J., White, N. S., Fallon, J., Kilpatrick, L., Lawrence, C., et al. (2001). Sex-related difference in amygdala activity during emotionally influenced memory storage. Neurobiology of Learning and Memory, 75, 1–9. Cahill, L., & McGaugh, J. L. (1995). A novel demonstration of enhanced memory associated with emotional arousal. Consciousness and Cognition, 4, 410– 421. Cahill, L., Prins, B., Weber, M., & McGaugh, J. L. (1994). Beta-adrenergic activation and memory for emotional events. Nature, 371, 702–704. Cahill, L., & van Stegeren, A. (2003). Sex-related impairment of memory for emotional events with beta-adrenergic blockade. Neurobiology of Learning and Memory, 79, 81–88. Canli, T., Desmond, J. E., Zhao, Z., & Gabrieli, J. D. (2002). Sex differences in the neural basis of emotional memories. Proceedings of the National Academy of Science, USA, 99, 10789–10794.
69
Canli, T., Zhao, Z., Brewer, J., Gabrieli, J. D., & Cahill, L. (2000). Event-related activation in the human amygdala associates with later memory for individual emotional experience. Journal of Neuroscience, 20, RC99. Canli, T., Zhao, Z., Desmond, J., Glover, G., & Gabrieli, J. D. (1999). fMRI identifies a network of structures correlated with retention of positive and negative emotional memory. Psychobiology, 27, 441–452. Christianson, S.-Å., & Loftus, E. F. (1991). Remembering emotional events: The fate of detailed information. Cognition and Emotion, 5, 81–108. Clark, K. B., Krahl, S. E., Smith, D. C., & Jensen, R. A. (1995). Post-training unilateral vagal stimulation enhances retention performance in the rat. Neurobiology of Learning and Memory, 63, 213–216. Clark, K. B., Naritoku, D. K., Smith, D. C., Browning, R. A., & Jensen, R. A. (1999). Enhanced recognition memory following vagus nerve stimulation in human subjects. Nature Neuroscience, 2, 94–98. Cuenod, C. A., Denys, A., Michot, J. L., Jehenson, P., Forette, F., Kaplan, D., et al. (1993). Amygdala atrophy in Alzheimer’s disease. An in vivo magnetic resonance imaging study. Archives of Neurology, 50, 941–945. Damasio, A. R. (1994). Descartes’ error and the future of human life. Scientific American, 271, 144. Davis, M. (1997). Neurobiology of fear responses: The role of the amygdala. Journal of Neuropsychiatry and Clinical Neuroscience, 9, 382–402. Davis, M., Walker, D. L., & Lee, Y. (1997). Roles of the amygdala and bed nucleus of the stria terminalis in fear and anxiety measured with the acoustic startle reflex. Possible relevance to PTSD. Annals of the New York Academy of Science, 821, 305–331. de Kloet, E. R., Oitzl, M. S., & Joels, M. (1999). Stress and cognition: Are corticosteroids good or bad guys? Trends in Neuroscience, 22, 422–426. Denburg, N. L., Buchanan, T. W., Tranel, D., & Adolphs, R. (in press). Evidence for preserved emotional memory in normal elderly persons. Emotion. Dobbins, I. G., Kroll, N. E., Tulving, E., Knight, R. T., & Gazzaniga, M. S. (1998). Unilateral medial temporal lobe memory impairment, type deficit, function deficit, or both? Neuropsychologia, 36, 115–127. Dolan, R. J., Lane, R., Chua, P., & Fletcher, P. (2000). Dissociable temporal lobe activations during emotional episodic memory retrieval. Neuroimage, 11, 203–209. Eichenbaum, H., & Cohen, N. J. (2001). From conditioning to conscious recollection, memory systems of the brain. Oxford: Oxford University Press. Escobar, M. L., Chao, V., & Bermudez-Rattoni, F. (1998). In vivo long-term potentiation in the insular cortex, NMDA receptor dependence. Brain Research, 779, 314–319. Fink, G. R., Markowitsch, H. J., Reinkemeier, M., Bruckbauer, T., Kessler, J., & Heiss, W. D. (1996). Cerebral representation of one’s own past, neural networks involved in autobiographical memory. Journal of Neuroscience, 16, 4275– 4282. Frijda, N. H. (1987). Emotion, cognitive structure, and action tendency. Cognition and Emotion, 1, 115–143.
70
Gaffan, D., Murray, E. A., & Fabre-Thorpe, M. (1993). Interaction of the amygdala with the frontal lobe in reward memory. European Journal of Neuroscience, 5, 968–975. Goddard, G. V. (1964). Amygdaloid stimulation and learning in the rat. Journal of Comparative and Physiological Psychology, 58, 23–30. Gold, P. E., Hankins, L., Edwards, R. M., Chester, J., & McGaugh, J. L. (1975). Memory interference and facilitation with posttrial amygdala stimulation, effect on memory varies with footshock level. Brain Research, 86, 509–513. Gold, P. E., & van Buskirk, R. (1975). Facilitation of time-dependent memory processes with posttrial epinephrine injections. Behavioral Biology, 23, 509– 520. Golomb, J., de Leon, M. J., Kluger, A., George, A. E., Tarshish, C., & Ferris, S. H. (1993). Hippocampal atrophy in normal aging. An association with recent memory impairment. Archives of Neurology, 50, 967–973. Grady, C. L., & Craik, F. I. (2000). Changes in memory processing with age. Current Opinion in Neurobiology, 10, 224–231. Hamann, S., & Mao, H. (2002). Positive and negative emotional verbal stimuli elicit activity in the left amygdala. Neuroreport, 13, 15–19. Hamann, S. B., Cahill, L., McGaugh, J. L., & Squire, L. R. (1997). Intact enhancement of declarative memory for emotional material in amnesia. Learning & Memory, 4, 301–309. Hamann, S. B., Cahill, L., & Squire, L. R. (1997). Emotional perception and memory in amnesia. Neuropsychology, 11, 104–113. Hamann, S. B., Ely, T. D., Grafton, S. T., & Kilts, C. D. (1999). Amygdala activity related to enhanced memory for pleasant and aversive stimuli. Nature Neuroscience, 2, 289–293. Hamann, S. B., Ely, T. D., Hoffman, J. M., & Kilts, C. D. (2002). Ecstasy and agony, activation of the human amygdala in positive and negative emotion. Psychological Science, 13, 135–141. Hamann, S. B., Monarch, E. S., & Goldstein, F. C. (2000). Memory enhancement for emotional stimuli is impaired in early Alzheimer’s disease. Neuropsychology, 14, 82–92. Heuer, F., & Reisberg, D. (1990). Vivid memories of emotional events, the accuracy of remembered minutiae. Memory & Cognition, 18, 496–506. Honkaniemi, J., Pelto-Huikko, M., Rechardt, L., Isola, J., Lammi, A., Fuxe, K., et al. (1992). Colocalization of peptide and glucocorticoid receptor immunoreactivities in rat central amygdaloid nucleus. Neuroendocrinology, 55, 451–459. Hyman, B. T., Van Hoesen, G. W., & Damasio, A. R. (1990). Memory-related neural systems in Alzheimer’s disease, an anatomic study. Neurology, 40, 1721– 1730. Ikegaya, Y., Nakanishi, K., Saito, H., & Abe, K. (1997). Amygdala betanoradrenergic influence on hippocampal long-term potentiation in vivo. Neuroreport, 8, 3143–3146. Jack, C. R., Jr., Petersen, R. C., Xu, Y. C., Waring, S. C., O’Brien, P. C., Tangalos, E. G., et al. (1997). Medial temporal atrophy on MRI in normal aging and very mild Alzheimer’s disease. Neurology, 49, 786–794.
71
Jernigan, T. L., Archibald, S. L., Fennema-Notestine, C., Gamst, A. C., Stout, J. C., Bonner, J., et al. (2001). Effects of age on tissues and regions of the cerebrum and cerebellum. Neurobiology of Aging, 22, 581–594. Kazui, H., Mori, E., Hashimoto, M., Hirono, N., Imamura, T., Tanimukai, S., et al. (2000). Impact of emotion on memory, controlled study of the influence of emotionally charged material on declarative memory in Alzheimer’s disease. British Journal of Psychiatry, 177, 343–347. Kim, J. J., & Diamond, D. M. (2002). The stressed hippocampus, synaptic plasticity and lost memories. Nature Reviews Neuroscience, 3, 453–462. Klüver, H., & Bucy, P. C. (1938). ‘Psychic blindness’ and other symptoms following bilateral temporal lobectomy in rhesus monkeys. American Journal of Physiology, 5, 33–54. Kroll, N. E., Markowitsch, H. J., Knight, R. T., & von Cramon, D. Y. (1997). Retrieval of old memories, the temporofrontal hypothesis. Brain, 120, 1377–1399. LaBar, K. S., & Phelps, E. A. (1998). Arousal-mediated memory consolidation: Role of the medial temporal lobe in humans. Psychological Science, 9, 490–493. Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1990). Emotion, attention, and the startle reflex. Psychological Review, 97, 377–395. LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience, 23, 155–184. Levine, B., Black, S. E., Cabeza, R., Sinden, M., McIntosh, A. R., Toth, J. P., et al. (1998). Episodic memory and the self in a case of isolated retrograde amnesia. Brain, 121, 1951–1973. Liang, K. C., Juler, R. G., & McGaugh, J. L. (1986). Modulating effects of posttraining epinephrine on memory, involvement of the amygdala noradrenergic system. Brain Research, 368, 125–133. Lovallo, W. R. (1997). Stress and health: Biological and psychological interactions. Thousand Oaks, CA: Sage. Lupien, S. J., & Lepage, M. (2001). Stress, memory, and the hippocampus, can’t live with it, can’t live without it. Behavioral Brain Research, 127, 137–158. Maratos, E. J., Dolan, R. J., Morris, J. S., Henson, R. N. A., & Rugg, M. D. (2001). Neural activity associated with episodic memory for emotional context. Neuropsychologia, 39, 910–920. Maratos, E. J., & Rugg, M. D. (2001). Electrophysiological correlates of the retrieval of emotional and non-emotional context. Journal of Cognitive Neuroscience, 13, 877–891. Markowitsch, H. J., Calabrese, P., Wurker, M., Durwen, H. F., Kessler, J., Babinsky, R., Brechtelsbauer, D., Heuser, L., & Gehlen, W. (1994). The amygdala’s contribution to memory—A study on two patients with Urbach-Wiethe disease. Neuroreport, 5, 1349–1352. Markowitsch, H. J., Thiel, A., Reinkemeier, M., Kessler, J., Koyuncu, A., & Heiss, W. D. (2000). Right amygdalar and temporofrontal activation during autobiographic, but not during fictitious memory retrieval. Behavioural Neurology, 12, 181–190. McEwen, B. S. (2000). The neurobiology of stress, from serendipity to clinical relevance. Brain Research, 886, 172–189.
72
McEwen, B. S., & Sapolsky, R. M. (1995). Stress and cognitive function. Current Opinion in Neurobiology, 5, 205–216. McGaugh, J. L. (1983). Hormonal influences on memory. Annual Review of Psychology, 34, 297–323. McGaugh, J. L. (2000). Memory—A century of consolidation. Science, 287, 248–251. McGaugh, J. L., & Izquierdo, I. (2000). The contribution of pharmacology to research on the mechanisms of memory formation. Trends in Pharmacological Science, 21, 208–210. Moayeri, S. E., Cahill, L., Jin, Y., & Potkin, S. G. (2000). Relative sparing of emotionally influenced memory in Alzheimer’s disease. Neuroreport, 11, 653–655. Mori, E., Yoneda, Y., Yamashita, H., Hirono, N., Ikeda, M., & Yamadori, A. (1997). Medial temporal structures relate to memory impairment in Alzheimer’s disease, an MRI volumetric study. Journal of Neurology, Neurosurgery, and Psychiatry, 63, 214–221. Mu, Q., Xie, J., Wen, Z., Weng, Y., & Shuyan Z. (1999). A quantitative MR study of the hippocampal formation, the amygdala, and the temporal horn of the lateral ventricle in healthy subjects 40 to 90 years of age. American Journal of Neuroradiology, 20, 207–211. Murray, E. A. (1992). Medial temporal lobe structures contributing to recognition memory: The amygdaloid complex versus the rhinal cortex. In J. P. Aggleton (Ed.), The amygdala: Neurobiological aspects of emotion, memory, and mental dysfunction (pp. 453–470). New York: Wiley-Liss. Öhman, A., & Mineka, S. (2001). Fears, phobias, and preparedness, toward an evolved module of fear and fear learning. Psychological Review, 108, 483–522. Oitzl, M. S., & de Kloet, E. R. (1992). Selective corticosteroid antagonists modulate specific aspects of spatial orientation learning. Behavioral Neuroscience, 106, 62–71. Packard, M. G., Cahill, L., & McGaugh, J. L. (1994). Amygdala modulation of hippocampal-dependent and caudate nucleus-dependent memory processes. Proceedings of the National Academy of Science, USA, 91, 8477–8481. Packard, M. G., & Teather, L. A. (1998). Amygdala modulation of multiple memory systems, hippocampus and caudate-putamen. Neurobiology of Learning and Memory, 69, 163–203. Parker, A., Wilding, E., & Akerman, C. (1998). The Von Restorff effect in visual object recognition memory in humans and monkeys. The role of frontal/perirhinal interaction. Journal of Cognitive Neuroscience, 10, 691–703. Phelps, E. A., LaBar, K. S., & Spencer, D. D. (1997). Memory for emotional words following unilateral temporal lobectomy. Brain and Cognition, 35, 85–109. Pitkänen, A. (2000). Connectivity of the rat amygdaloid complex. In J. P. Aggleton (Ed.), The amygdala: A functional analysis. Oxford: Oxford University Press. Quirarte, G. L., Roozendaal, B., & McGaugh, J. L. (1997). Glucocorticoid enhancement of memory storage involves noradrenergic activation in the basolateral amygdala. Proceedings of the National Academy of Science, USA, 94, 14048–14053. Reisberg, D., & Heuer, F. (1992). Remembering the details of emotional events. In E. Winograd & U. Neisser (Eds.), Affect and accuracy in recall: Studies of “flashbulb” memories (pp. 163–190). Cambridge, MA: Harvard University Press.
73
Rolls, E. T. (2000a). Memory systems in the brain. Annual Review of Psychology, 51, 599–630. Rolls, E. T. (2000b). The orbitofrontal cortex and reward. Cerebral Cortex, 10, 284– 294. Roozendaal, B. (2000). Glucocorticoids and the regulation of memory consolidation. Psychoneuroendocrinology, 25, 213–238. Roozendaal, B., Williams, C. L., & McGaugh, J. L. (1999). Glucocorticoid receptor activation in the rat nucleus of the solitary tract facilitates memory consolidation, involvement of the basolateral amygdala. European Journal of Neuroscience, 11, 1317–1323. Schmolck, H., Buffalo, E. A., & Squire, L. R. (2000). Memory distortions develop over time: Recollections of the O. J. Simpson trial verdict after 15 and 32 months. Psychological Science, 11, 39–45. Schooler, J. W., & Eich, E. (2000). Memory for emotional events. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 379–392). Oxford: Oxford University Press. Scott, S. A., DeKosky, S. T., & Scheff, S. W. (1991). Volumetric atrophy of the amygdala in Alzheimer’s disease, quantitative serial reconstruction. Neurology, 41, 351–356. Scoville, W. B., & Milner, B. (1957). Loss of recent memory after bilateral hippocampal lesions. Journal of Neurology, Neurosurgery, and Psychiatry, 20, 11–21. Seidlitz, L., & Diener, E. (1998). Sex differences in the recall of affective experiences. Journal of Personality and Social Psychology, 74, 262–271. Shepard, J. D., Barron, K. W., & Myers, D. A. (2000). Corticosterone delivery to the amygdala increases corticotropin-releasing factor mRNA in the central amygdaloid nucleus and anxiety-like behavior. Brain Research, 861, 288–295. Squire, L. (1987). Memory and brain. New York: Oxford University Press. Squire, L. R., & Kandel, E. R. (2000). Memory: From mind to molecules. New York: Scientific American Library. Tomaz, C., Dickinson-Anson, H., McGaugh, J. L., Souza-Silva, M. A., Viana, M. B., & Graeff, F. G. (1993). Localization in the amygdala of the amnestic action of diazepam on emotional memory. Behavioral Brain Research, 58, 99–105. Tranel, D. & Hyman, B. (1990). Neuropsychological correlates of bilateral amygdala damage. Archives of Neurology, 47, 349–355 Tulving, E., Kapur, S., Craik, F. I., Moscovitch, M., & Houle, S. (1994). Hemispheric encoding/retrieval asymmetry in episodic memory, positron emission tomography findings. Proceedings of the National Academy of Science, USA, 91, 2016–2020. Tulving, E., Kapur, S., Markowitsch, H. J., Craik, F. I., Habib, R., & Houle, S. (1994). Neuroanatomical correlates of retrieval in episodic memory, auditory sentence recognition. Proceedings of the National Academy of Science, USA, 91, 2012–2015. Van Hoesen, G. W., Augustinack, J. C., & Redman, S. J. (1999). Ventromedial temporal lobe pathology in dementia, brain trauma, and schizophrenia. Annals of the New York Academy of Science, 877, 575–594. Weiskrantz, L. (1956). Behavioral changes associated with ablation of the amygdaloid complex in monkeys. Journal of Comparative and Physiological Psychology, 49, 381–391.
74
Yerkes, R. M., & Dodson, J. D. (1908). The relation of strength of stimulus to rapidity of habit formation. Journal of Comparative Neurology and Psychology, 18, 31– 39. Zola-Morgan, S., Squire, L. R., Alvarez-Royo, P., & Clower, R. P. (1991). Independence of memory functions and emotional behavior, separate contributions of the hippocampal formation and the amygdala. Hippocampus, 1, 207–220. Zola-Morgan, S., Squire, L. R., & Amaral, D. G. (1986). Human amnesia and the medial temporal region, enduring memory impairment following a bilateral lesion limited to field CA1 of the hippocampus. Journal of Neuroscience, 6, 2950– 2967. Zola-Morgan, S., Squire, L. R., & Amaral, D. G. (1989). Lesions of the amygdala that spare adjacent cortical regions do not impair memory or exacerbate the impairment following lesions of the hippocampal formation. Journal of Neuroscience, 9, 1922–1936.
75
3
. , , . , .
T
rauma affects memory. We have known that for millennia, but only recently have we been able to ask and answer pointed questions about why and how. Only by elucidating the neurological consequences of trauma, which until recently had been deemed a purely “psychological” event, can we unravel its impact on memory. In this chapter we (1) briefly discuss the nature and organization of memory; (2) attempt to define the notion of trauma and the related phenomenon of stress; (3) consider the impact of trauma and stress on the brain and, by extension, on the memory functions subserved by the brain; (4) offer an integrated account connecting these data with several clinical manifestations of trauma, stress, and memory; and (5) use this account to resolve an apparent tension between two parallel literatures: one describing a trauma- or stress-based impairment of memory and another describing an emotion-based enhancement of memory.
The Nature of Memory:
A Taxonomy
One of the key insights of recent cognitive- and neuroscience-based analyses of memory is that multiple systems underlie this function. Unlike earlier perspectives, which viewed memory as a single system, perhaps subserved by a restricted part of the brain, the modern consensus view is that several types of memory, each obeying rather different rules of operation, and each subserved by discrete neural systems, interact to produce our subjective sense of remembering. This insight is critical if we are to understand the impact of any experience, normal or traumatic, on memory, because it raises the possibility that different kinds of memory will face different fates in the presence of trauma.
76
76
We needn’t go into much detail about the nature of these multiple memory systems or the types of information they process. Various views suggest how to classify and arrange these systems (cf. Schacter & Tulving, 1994); most agree on a distinction between two broad classes of memory. First, there are memories of the events in our lives and the knowledge of the world that we obtain from those events. Typically, this class of memory, referred to as explicit memory, can be explicitly retrieved on demand. Second, there are memories for the skills, procedures, and habits we acquire through experience. These implicit memories are not so readily made explicit; indeed, in most instances these memories cannot be expressed explicitly and are evident only in behavior. A distinction between explicit and implicit memory is important but not sufficient for understanding the impact of trauma on remembering. Within explicit memory, there is a further crucial subdivision between episodic and semantic memories. Episodic memory incorporates the specific context of an experienced event, including the time and place of its occurrence. Semantic memory, on the other hand, is concerned with the knowledge one acquires during events but is itself separated from the specific event in question. Thus, our knowledge about the meaning of words, and facts about the world, though acquired in the context of some specific experience, appears to be stored in a form that is not bound to the originating context. The sensory features of experiences, their positive and negative attributes, other objects with which they might have been associated— knowledge of this kind constitutes a part of semantic memory. The exact relation between episodic and semantic memory remains to be unraveled; however, there is considerable agreement that they are separable in the brain. Evidence suggests a third, emotional memory, system that mediates the encoding and storage of emotionally charged events. This system pertains to learning about fearful and unpleasant stimuli, although some evidence suggests it plays a role in memory for pleasant information as well (see chapter 2, this volume; Hamann, Ely, Grafton, & Kilts, 1999). Apparently, each of these memory systems is subserved by a different brain area, and, consequently, each is open to differential impact by exposure to stressors and to trauma. Episodic memory seems to be the most influenced by trauma and stress and is neuroanatomically linked to the hippocampal formation in the medial temporal lobe. But before embarking on a description of the hippocampal system and detailing its response to stress, we first define stress and trauma.
What Are Stress and Trauma? Despite the widespread use of “stress” and “trauma,” these terms lack precise definitions. Although “stressor” is unambiguous according to the Oxford English Dictionary (OED), which gives it a single definition, “a single condition or agent that constitutes a stress for an organism,” the term “stress” is utterly ambiguous.
77
The OED provides at least nine distinct senses of the word when it is used as a noun. These range from stress defined as “emphasis,” to stress defined as “physical hardship, strain or pressure.” Furthermore, the OED provides at least five distinct meanings for “stress” used as a verb. “Trauma” is similarly ambiguous. Although the OED considers it only as a noun, it gives five distinct senses for its use, including two that are germane here: 1. Pathology. A wound, or external bodily injury in general; also the condition caused by this. 2. Psychoanalysis and Psychiatry. A psychic injury, especially one caused by emotional shock the memory of which is repressed and remains unhealed; an internal injury, esp. to the brain, which may result in a behavioral disorder of organic origin. Also, the state or condition so caused.
In the scientific literature, both trauma and stress are often treated as static things, as if they were well-defined entities characterized by psychophysiological measures that identify and quantify them (Kim & Diamond, 2002). Such is clearly not the case, at least not at this stage in our understanding. Both stress and trauma are events, that is, dynamic processes associated with various physical and psychological responses that can differ in magnitude and expression. A common working definition of stress is any event that seriously disturbs the physiological or psychological homeostasis of an organism. Cannon (1929) was the first to use ‘homeostasis’ to characterize processes that are essential to the maintenance of life and to highlight the body’s ability to control its internal environment within narrow limits. Selye (1956) initiated the first major study of the effects of stress on systemic regulation, defining stress as any severe threat to homeostasis that could result in death. More recently, however, McEwen (2000, 2001) has suggested that because homeostasis applies to a limited number of systems (such as pH, body temperature) that are essential for survival, another term, “allostasis,” may better describe the processes disturbed during stress. Allostasis refers to the systems that actively maintain homeostasis even when life and death are not at stake. Kim and Diamond (2002) argue that a good definition of stress must address the following two facts. First, stress is determined not by a given environmental situation but by how an organism perceives and reacts to a situation. This definition permits individual differences in response to stressors, as well as variations in response to a fixed stressor over time and experience within an individual. The need to take experience, personal biases, and temperament into account should be obvious to anyone who has learned to engage in public speaking; at first a nerve-racking experience, for most, public speaking eventually becomes easier and less anxiety-provoking. Second, there is no physiological state that always defines stress. Although most investigators rely on elevations of stress hormones
78
(e.g., glucocorticoids) as an indicator of stress, such hormones can also increase in response to physical exertion and daily changes in the sleep/wake cycle (e.g., Kanaley, Weltman, Pieper, Weltman, & Hartman, 2001; Plihal & Born, 1999; Van Cauter & Turek, 2001). With these limitations in mind, Kim and Diamond (2002) presented a threepart definition of stress, which they claim is broadly applicable across species and paradigms. They argue that, first, stress is always accompanied by high levels of physiological arousal, as measured by behavior (motor behaviors) or neurochemistry (increased stress hormones). Second, stress must be perceived as threatening and something that would be avoided if possible. For example, giving a speech is considered threatening, and many people would avoid it if possible. When the speech cannot be avoided, however, the task is accompanied by increased physiological excitability (heart rate, sweating, glucocorticoid increases, etc.). Finally, stress depends on whether an organism perceives that it has control over the stressful experience. This component reflects evidence from animal and human studies demonstrating that perceived control can profoundly mitigate the experience of stress (e.g., Fox & Dwyer, 2000; Maier & Watkins, 1998; Prince & Anisman, 1990). We agree with the first component of the definition but take issue with the second and third components. We suggest that “stressor” should not always mean something negative. Instead, it should simply designate a dynamic “strain” or “pressure” exerted on an individual. This strain or pressure could be pleasant (e.g., skydiving, sexual activity) or unpleasant (e.g., a loss of social status), but in both cases it disturbs the individual’s allostatic balance. Further, the data suggest that controllability can sometimes mitigate the action of a stressor but at other times has no impact at all. People often have control over events they experience as stressful; giving speeches and willingly jumping out of airplanes are two good examples. Thus, although controllability is certainly important and may, in some cases, both mitigate the perceived negativity of stress and determine its magnitude, we assert that stress reactions, even intense stress reactions, occur in the presence of perceived control. For our purposes, then, we use “stress” to designate activation of a dynamic physiological state—a system that, according to current knowledge, consists of the hypothalamic-pituitary-adrenal (HPA) axis. We use the activation of this system as a proxy for stress, bearing in mind the noted qualification that the mere presence of this physiological state does not indicate inevitable stress. Further details of the physiology of stress will be provided in the section that follows. Moreover, we use the term “stressor” to designate the environmental conditions that elicit or influence this physiological system, recognizing the need to specify the formal relations between given environmental stressors and the stress response. We do not consider psychological trauma the equivalent of intense stress. Although stress coupled with environmental variables may produce a traumatic
79
psychological state, and a traumatic psychological state may induce stress (i.e., serve as a stressor), a traumatic state is, by our reckoning, a quasi-independent mental state that systematically interacts with stress and stressors.1
Effects of Stress: Brain Physiology
and Anatomy
As mentioned, stressors activate the HPA axis. Although the stress response system is complex, to date much of the research focus has been on the HPA axis and the hippocampus. Stress leads to the release of corticotropin-releasing factor (CRF) from the hypothalamus and subsequently triggers adrenocorticotropic hormone (ACTH) release from the pituitary gland. In response to ACTH, stress hormones called glucocorticoids are secreted and released from the adrenal glands. Circulating glucocorticoids then feed back onto the pituitary and hypothalamus to inhibit the secretion of CRF and ACTH. The hippocampus (and certain cortical areas, particularly the frontal cortex) plays an important role in this negative feedback system as well, helping the pituitary and hypothalamus inhibit HPA activity (Feldman & Conforti, 1980, 1985; Jacobson & Sapolsky, 1991; Sapolsky & Meaney, 1986). Hippocampal involvement in the regulation of the stress response turns out to be critically important to any discussion of stress and memory, because, as we noted, this structure has long been implicated in explicit memory. In particular, the hippocampal formation has been linked to episodic memory, to context, and to spatial maps (e.g., O’Keefe & Nadel, 1978). Debate rages over aspects of this story, but the link between the hippocampal system and episodic memory seems firm. The seminal studies of Scoville and Milner (1957), for example, made it clear that bilateral ablation of the hippocampal complex resulted in profound loss of memory. We now know that damage to the hippocampus results in a specific form of memory dysfunction, in particular that involving spatial/contextual and episodic memories (Hirsch, 1974; Nadel, 1991, 1994; Nadel & Willner, 1980; Rosenbaum et al., 2000), and, further, that stress appears to impair precisely these types of memory (see Lupien & McEwen, 1997). Together with the discovery of dense concentrations of stress hormone receptors in the hippocampus (e.g., McEwen, Weiss, & Schwartz, 1968), these findings provide a basis for the claim that stress hormone modulation of hippocampal activity underlies the damaging effects of stress in animals and humans (see Bremner, 1999; McEwen, 2000, for reviews). Thus, increases in stress hormones not only trigger inhibition of HPA activity via hippocampal mediation but at high enough levels also impair the neuronal structure and function of the hippocampus. Glucocorticoids at abnormally high or abnormally low concentrations can, by changing the physiological characteristics of large populations of neurons, disrupt hippocampal function. Thus,
80
a moderate level of circulating glucocorticoids appears to enhance memory, whereas maximal or minimal levels appear to disrupt it. The stress response typically involves a dramatic increase in the level of circulating glucocorticoids, which can produce changes in hippocampal neuronal physiology and anatomy. In short, stress appears to alter hippocampal morphology, disrupt neurogenesis, and block synaptic plasticity thought to underlie memory formation (see McEwen, 2000, 2001, for reviews). For example, both glucocorticoid exposure and restraint stress over 21 days have been shown to result in reduced dendritic branching in pyramidal neurons of the CA3 region of the rat hippocampus (Magarinos & McEwen, 1995; Watanabe, Gould, & McEwen, 1992; Woolley, Gould, & McEwen, 1990). Psychosocial stress leads to atrophy of dendrites as well, in the tree shrew (Magarinos, McEwen, Flugge, & Fuchs, 1996) and in rats (McKittrick et al., 2000). In addition, stress and glucocorticoid elevation can produce alterations in synaptic terminal structure in CA3 (Magarinos, Verdugo, & McEwen, 1997). With even longer periods of exposure to stress and glucocorticoids, initial evidence suggested that overt loss of hippocampal neurons occurred (e.g., Sapolsky, Krey, & McEwen, 1985; Uno, Ross, Else, Suleman, & Sapolsky, 1989). Such findings led to assumptions that neuronal death associated with glucocorticoid release might underlie the severe memory impairments seen in posttraumatic stress disorder (see later discussion). More recent work, however, raises questions about whether glucocorticoids are themselves toxic or simply create conditions for other factors to exert neurotoxic effects (Sapolsky, 1996). Nonetheless, chronic exposure to stressors is undoubtedly associated with hippocampal impairment. For example, post-mortem analysis of chronically stressed monkeys showed extensive damage in the CA3 region of the hippocampus; subsequent research suggested that high levels of the glucocorticoid cortisol were responsible for the hippocampal damage, as cortisol-secreting pellets implanted in the monkey hippocampus resulted in the same pattern of damage. In addition to having neurotoxic effects on hippocampal neurons, chronic exposure to stressors may also inhibit the growth of new neurons (granule cells) in the dentate gyrus, thereby changing its structure. A stress-induced decrease in neurogenesis has been demonstrated in the dentate gyrus of rats, tree shrews, and primates (Gould, Tanapat, McEwen, Flugge, & Fuchs, 1998; Gould & Tanapat, 1999; Tanapat, Hastings, Rydel, Galea, & Gould, 2001) and has been hypothesized as the primary mechanism of stress-related hippocampal impairment (Gould & Gross, 2002; Gould & Tanapat, 1999). Glucocorticoids appear to mediate these stress-induced changes in neurogenesis. For example, normalizing glucocorticoid levels was shown to prevent predator odor stress from inhibiting granule cell production (Tanapat et al., 2001). In addition, corticosterone treatment has been shown to decrease dentate cell proliferation, whereas adrenalectomy increases the production of new granule cells (Cameron & Gould, 1994; Cameron & McKay, 1999).
81
In addition to these structural effects, stress and stress hormones also modulate the excitability of hippocampal neurons. Acute stress and glucocorticoid release can impair hippocampal plasticity (Diamond, Fleshner, & Rose, 1994; Foy, Stanton, Levine, & Thompson, 1987). Long-term potentiation (LTP) and the related phenomenon of primed burst potentiation (PBP) are two processes by which hippocampal neuronal responsiveness changes in response to experience (McEwen, 2001). Thus, these forms of potentiation are often conceptualized as physiological models of memory. Studies show that hippocampal LTP can be impaired by the administration of glucocorticoids (e.g., Diamond & Rose, 1994; Filipini, Gijsbers, Birmingham, & Dubrovsky, 1991). Diamond, Bennet, Fleshner, and Rose (1992) reported an inverted U-shaped relationship between the level of circulating glucocorticoids and the extent of LTP, demonstrating that glucocorticoids facilitated LTP at low levels but disrupted it at high levels. Stressinduced disruption of LTP targets CA1 and the dentate gyrus and has been shown to last up to 48 hours in rats and 24 hours in mice (e.g., Garcia, Musleh, Tocco, Thompson, & Baudry, 1997; Shors, Gallegos, & Breindl, 1997). These studies show that LTP can be disrupted by stress and that changes can last for days after exposure to a single stressor; however, these changes are typically not permanent. More recent studies suggest that PBP may be even more sensitive to psychological stress (e.g., predator exposure) than is LTP (Mesches, Fleshner, Heman, Rose, & Diamond, 1999).
Stress and Brain Anatomy:
Evidence for Hippocampal
Volume Reduction
Given the evidence of hippocampal impairment in response to stress, one might expect to see gross structural changes in the hippocampi of certain patient populations. Indeed, the human hippocampus does appear to show signs of atrophy as a result of severe stress or persistently elevated glucocorticoids. Reductions have been noted in disorders associated with prolonged glucocorticoid elevations, such as Cushing’s syndrome, recurrent major depression, schizophrenia, posttraumatic stress disorder, and in some cases of normal aging (Starkman et al., 1999; Sheline et al., 1996; Bremner et al., 1993; Bremner, Randall, Capelli, et al., 1995; Gurvits et al., 1996; Fukuzako et al., 1996; Convit et al., 1995). For example, Cushing’s syndrome arises from adrenocorticotropic hormone or corticotropin-releasing hormone secreting tumors (Sapolsky, 2000). Those with the disorder thus have hypercortisolemia and often exhibit cognitive dysfunction (Sapolsky, 2000; Starkman et al., 1999). Using magnetic resonance imaging (MRI), researchers have reported selective hippocampal atrophy in individuals suffering from Cushing’s syndrome for approximately 1 to 4 years (Starkman, Gebarski, Berent, & Schteingart, 1992). Although a trauma control group was not tested in this study, hippocampal reduction was strongly correlated with cortisol levels.
82
MRI studies of PTSD have revealed a pattern of evidence many consider indicative of hippocampal atrophy. When Vietnam veterans with combat-related PTSD were compared to healthy control subjects, the PTSD patients were found to have an 8% reduction in right hippocampal volume (Bremner, Randall, Scott, et al., 1995). This difference in volume was associated with short-term memory impairments as assessed by the WMS-Logical memory test. Similar results were obtained in childhood abuse survivors (physical and sexual) with PTSD (Bremner et al., 1997). In this sample, however, left hippocampal volume was 12% smaller than that of controls. Although right hippocampi were 3.8% smaller than those of controls, this difference did not reach statistical significance. In both studies, control subjects were matched for sex, age, race, height, weight, socioeconomic status, education, and alcohol abuse. Other studies finding similar differences in hippocampal volume include Gurvits et al. (1996) (26% smaller bilateral volume in Vietnam combat veterans with PTSD compared to non-PTSD combat veterans) and Stein, Koverola, Hanna, Torchia, & McClarty (1997) (5% smaller left hippocampal volume in women who reported sexual abuse than in controls). Concerns about the reliability and interpretability of these studies, which center on the co-occurrence of hippocampal atrophy with other disorders (e.g., depression, substance abuse), led Bremner and other researchers to revisit their dataset and statistically exclude those subjects who had abused alcohol or other substances during the 6 months prior to their MRIs. Despite these changes, results remained suggestive of a relationship between PTSD and reduced hippocampal volume not explainable by alcohol or drug abuse (see Bremner, 2001b). Though in general MRI studies are suggestive of stress-related hippocampal atrophy, interpretation of the results is compromised by cross-sectional designs and the use only of subjects with chronic PTSD. Consequently, one cannot use these results to infer causality. Reduced hippocampal volume could either precede the traumatic event, making individuals vulnerable to developing PTSD, or could develop in the aftermath of the trauma as a consequence of prolonged exposure to stress hormones or other uncontrolled factors. To correct for these confounds, Bonne et al. (2001) conducted a prospective MRI study examining the hippocampal volume of individuals exposed to acute traumas (traumas experienced by subjects admitted to a hospital emergency room). The hippocampal volume of those who developed PTSD 6 months later did not differ from those who had not developed PTSD at the 6-month followup. Although these results are inconsistent with earlier studies of hippocampal volume reductions and PTSD, they may be explained by differences in the nature or duration of trauma. The earlier studies tested individuals exposed to chronic traumatization, such as child abuse or war experience, whereas Bonne et al. tested individuals exposed to acute stressors, perhaps of lesser severity.2 Another possibility is that the earlier studies failed to screen tightly for confounding factors such as alcoholism and depression, although Bremner (2001b) attempted to do so. In any case, the conservative conclusion is that tentative evidence
83
exists for reduced hippocampal volume in PTSD. Further studies that better control for comorbidity, type and duration of trauma, and duration of PTSD are needed. Nonetheless, stress-induced deficits in memory may exist in the absence of gross neuroanatomical changes, reflecting either functional or subtle structural impairments.
Effects of Stress on Memory: Data We have shown that exposure to trauma may result in damage to brain regions important for memory function. Now we turn to evidence supporting a relationship between stressors and memory function itself, a relationship with a long history. Stress-related memory impairments have been reported in a variety of clinical settings. As we mentioned, patient populations with chronically elevated levels of stress hormones, such as Cushing’s syndrome, schizophrenia, some forms of recurrent depression, PTSD,3 dementia of the Alzheimer’s type, and asthmatic patients treated with the glucocorticoid prednisone (Keenan, Jacobson, Soleymani, & Newcomer, 1995; Mauri et al., 1993; Rasmusson et al., 2001; Sapolsky, 2000; Sheline, Sanghavi, Mintun, & Gado, 1999; Starkman et al., 1999), are characterized by impaired memory function. However, despite consistent correlations between chronic elevations in stress hormones and memory impairment, these data are limited by the problems endemic to all naturalistic studies. Because nonrepresentative sampling, possible noncausal associations, and confounding variables such as extraneous disease factors may be responsible for memory impairments (e.g., Sapolsky, 2000), these observations must be treated with caution. The correlations do, however, serve as a starting point and encourage us to take a closer look at the nature of memory and its interactions with stress and trauma. A growing body of randomized, placebo-controlled studies, using reasonably precise definitions of stress, has demonstrated that exposure to uncontrollable stressors can impair memory. This research has proceeded along two separate but related paths, one dedicated to the study of relations between episodic memory and long-term/chronic stress and the other to relations between episodic memory and short-term/acute stress. The ethics inherent in human research make experimental studies of relations between memory and chronic stressors difficult if not impossible, leaving the field to naturalistic research designs. Experimental studies of relations between memory and acute stressors, such as stressful speeches or participation in highrisk tasks (e.g., skydiving), though difficult ethically, are at least possible. Finally, the ethics governing animal research permit the induction of short-term (acute) and long-term or repeated (chronic) stress using stressors such as tail-shock, restraint, swimming in cold water, and more naturalistic stressors such as
84
predator exposure. We consider each of these literatures briefly in the following sections. The administration of stress-related hormones permits well-controlled studies of relations between memory and acute stress in humans. Acute doses of stress hormones such as glucocorticoids (e.g., cortisol) can be safely administered to humans over relatively brief periods. In some studies, agonists (corticocortisone) or antagonists (e.g., dexamethasone) of naturally occurring glucocorticoids are used instead of cortisol. Glucocorticoid antagonists reduce the level of circulating glucocorticoids, and both abnormally high and abnormally low levels of glucocorticoids have been shown to produce memory deficits (Conrad & Roy, 1996; Luine, Villegas, Martinez, & McEwen, 1994; Vaher, Luine, Gould, & McEwen, 1994).
Acute Stressors: Memory in Humans Inducing acute stress in the laboratory typically involves exposing the subject, whether animal or human, to a brief, one-time stressor (e.g., a speech task, a predator’s odor) or administering a single dose of glucocorticoids. In both cases, glucocorticoid levels increase, usually for no longer than several hours in humans. However, the cognitive effects of acute stress or glucocorticoid administration are often apparent for several hours to as long as a day; moreover, they are usually reversible and are often specific to the task or experimental situation (McEwen, 2001). These effects are typically specific to memory function, with spatial and verbal-episodic memory often showing the greatest impairment. Wolkowitz, Reus, and Weingartner (1990) and Wolkowitz et al. (1993), for example, examined recall of previously learned verbal-episodic material after the administration of dexamethasone or prednisone and found that these glucocorticoids interfered with word recall. In the dexamethasone study, participants learned a list of 12 semantically associated words, which they were first asked to recall following a 90-second distractor task and then asked to recognize in a larger list of related distractor words. Participants were tested 1 week before and 1 day after administration of 1 mg of dexamethasone. Although this single dose of dexamethasone did not change recall, it did increase errors of commission, or false recognition of related words not presented on the study list. The prednisone study used a similar test of memory, but here both recall and recognition of previously learned words were required 24 hours after presentation of the list, a manipulation that created a test of longer-term episodic memory. Participants were given 80 mg of prednisone daily for 5 days. Word recognition and recall were tested once during an initial 5-day placebo period, again after 4 days of prednisone administration, and once again 7 days after discontinuation of the treatment. Again, participants receiving prednisone showed no differences in free recall but falsely recognized more semantically related words than
85
did controls. Thus, administration of dexamethasone and prednisone resulted in difficulties recognizing words presented in standard word lists (see also Bender, Lerner, & Poland, 1991), a difficulty characterized by an impaired ability to discriminate previously studied information from new, but similar, information. Newcomer, Craft, Hershey, Askins, and Bardgett (1994) also investigated the effects of dexamethasone on measures of memory, as well as attention and perceptual function. Dexamethasone treatment administered over 4 days (in successive doses of 0.5, 1, 1, and 1 mg) was associated with impaired recall of verbal material originally presented in a paragraph (Wechsler, 1945). This impairment was strongest on the 4th day of treatment and was specific to memory for verbal material. The presence of dexamethasone did not affect measures of attention or perception. Unlike Wolkowitz et al. (1990, 1993), Newcomer et al. (1994) found that dexamethasone treatment resulted in errors of omission (information in the paragraph was forgotten) but not errors of commission (related but nonpresented information was added to the material presented in the paragraph). The differences between the Wolkowitz et al. (1990) and Newcomer et al. studies are likely due to differences in the strategies required by the two tasks. The recall task Newcomer et al. used required the verbatim recall of story-like information presented in paragraph format. Given these instructions, normal subjects usually make more omission than commission errors, as thematically related intrusions errors rarely occur in this type of task. By comparison, semantically associated word lists, like the ones used in Wolkowitz et al. (1990), are known to produce commission errors at high rates (see Roediger & McDermott, 1995). We have recently shown that psychologically induced stress increases commission errors on semantically associated word lists (Payne, Nadel, Allen, Thomas, & Jacobs, 2002), a result we will discuss in more detail. Results of the first few experimental studies indicate that pharmacological doses of the glucocorticoids dexamethasone and prednisone impair verbal episodic memory. However, caution should be used when thinking of this memory impairment as causally related to stress-induced increases in cortisol. Although changes in recall and recognition could result from increased glucocorticoid receptor binding in a manner similar to the direct administration of cortisol (the traditional explanation), an alternative explanation is possible. Dexamethasone suppresses cortisol, and because very high or very low levels of cortisol can affect the biological structures on which memory depends, the observed changes could result from dexamethasone-induced decreases in cortisol secretion—and, hence, a failure to bind to glucocorticoid receptors. This uncertainty, along with differences in the binding of dexamethasone and cortisol to type I and type II glucocorticoid receptors in the hippocampus, limits our confidence in the generality of conclusions reached when dexamethasone and similar ligands are used to examine the effects of cortisol or stress in humans (Newcomer et al., 1999).
86
Similar concerns motivated a second study by Newcomer et al. (1999), which examined interactions between hydrocortisone (cortisol) and explicit memory for words, again using a paragraph recall task. Two fixed oral doses of hydrocortisone (160 mg/d or 40 mg/d) were administered over 4 days. The lower-dose treatment was selected to approximate physiological levels of cortisol associated with mild to moderate stress, and the higher-dose treatment was selected to approximate physiological levels of cortisol associated with severe stress. Sampling of plasma cortisol and memory testing were conducted at baseline (day 0), on days 1 and 4, and after a 6-day washout period (day 10). Impaired paragraph recall was reported in the presence of the higher, but not the lower, dose of hydrocortisone. As in their previous study (Newcomer et al., 1994), this effect was maximal on the 4th day of treatment. Newcomer et al. (1999) did not observe the immediate effects of glucocorticoids seen in some studies (e.g., Wolkowitz et al., 1990) but instead showed that several days of cortisol exposure at doses and plasma concentrations associated with high stress could impair the recall of verbal material presented in a paragraph format. An earlier study by Kirschbaum, Wolf, Wippich, and Hellhammer (1996), however, reported that a single, low dose of hydrocortisone (10 mg) led to a deficit in verbal episodic memory. In this study, subjects who received hydrocortisone recalled fewer words (via a cued recall test) from a previously learned word list than did control subjects when recall occurred 60 minutes after receiving the drug. Thus, hydrocortisone apparently can produce a deficit in at least one type of verbal memory (cued word recall) at smaller doses and with shorter delays between administration and test than suggested by the results of the Newcomer et al. (1999) experiment. Again, the differences in these two experiments might depend on the nature of the material recalled, the delay between learning and test, or different task demands. Although the details vary from experiment to experiment, the administration of drugs simulating acute stress clearly changes the recall pattern of words presented in word lists or paragraphs, both of which measure aspects of explicit verbal memory. Many of the glucocorticoid-administration studies are limited in that the physiological effects of cortisol are longer-lived than those of the standard memory experiment (which typically lasts no longer than an hour or two). Consider the designs used by Kirschbaum et al. (1996) and Newcomer et al. (1999) already described. In each case cortisol levels were probably elevated continuously throughout the study, making it difficult to determine what aspect of memory was impaired. In an attempt to sort this out, de Quervain, Roozendaal, Nitsch, McGaugh, and Hock (2000) administered 25 mg doses of cortisone (which is quickly absorbed and transformed into cortisol) at different stages of a verbal delayed free-recall experiment. In this way they could evaluate the impact of elevated cortisol on each of three distinct memory processes: acquisition, storage, and retrieval. Participants learned word lists composed of unrelated words and received cortisone either (1) before list learning, (2) immediately after
87
list learning, or (3) 24 hours later, but immediately before recall. Impairment in recall was found only when cortisone was given immediately before recall, thus implicating retrieval as the memory stage most affected by a low dose of cortisol. Although a 25 mg dose of hydrocortisone administered at encoding did not affect recall of verbal material presented in paragraphs (Newcomer et al., 1999), the same low dose of cortisone did affect the retrieval of words presented in lists (de Quervain et al., 2000). This discrepancy could have arisen from differences in the kind of information recalled (stories vs. word lists), or it might reflect a greater impact of stress on retrieval than on encoding or consolidation. In addition to explicit memory for words, spatial memory also appears to suffer in the presence of stress. To our knowledge, spatial memory in humans has not yet been assessed with the direct administration of glucocorticoids. However, Laurance, Hardt, Nadel, and Jacobs (2001) found that exposure to a social stressor (public speaking) slows the rate at which spatial information is acquired in a virtual environment. The impact of stress in this study was specific to spatial map-based performance, whereas performance based on individual cues, and therefore dependent on different neural circuits, remained intact. Public speaking tasks have also resulted in verbal episodic deficits, as measured by listlearning tasks, in healthy elderly subjects (Lupien et al., 1997) and in younger adults (Kirschbaum et al., 1996).
Acute and Chronic Stressors:
Memory in Animals
Animal studies of stress and memory yield results that are generally consistent with the human studies; exposure to glucocorticoids, acute or long-term, is often associated with deleterious effects on memory tasks dependent on hippocampal integrity (e.g., Bodnoff et al., 1995; Luine et al., 1994; Ohl & Fuchs, 1999; see Kim & Diamond, 2002; Lupien & Lepage, 2001, for reviews). As in humans, however, acute stress effects are often short-term and reversible, whereas repeated or long-term stress effects can lead to frank cognitive dysfunction (McEwen & Sapolsky, 1995) that may result from the changes in hippocampal morphology already described. Administering corticosterone to rats facilitates the extinction of forms of conditioned shock-avoidance that depend on the hippocampal formation (e.g., Bohus & Lissak, 1968; Greidanus, 1970; Roozendaal & McGaugh, 1996). Spatial memory, also known to depend on the hippocampus, is impaired in rats exposed to stressors or given corticosterone and in transgenic mice with elevated corticosterone levels. For example, 21 days of restraint was correlated with impairments in spatial memory as measured by performance in the radial arm maze task (Luine et al., 1994). Exposure to stressors has also been shown to disrupt hippocampal-dependent object recognition (e.g., Clark, Zola, & Squire, 2000) and
88
working memory (Diamond, Fleshner, Ingersoll, & Rose, 1996; Diamond, Park, Heman, & Rose, 1999). Likewise, long-term subcutaneous implants of glucocorticoids, which mimic the effects of chronic stress, impair new learning and memory for maze escape behaviors in rats. In one study, the severity of such deficits was correlated with the number of damaged cells in the CA3 region of the hippocampus (Arbel, Kadar, Silberman, & Levy, 1994). A thorough review of the effects of stress in animals is beyond the scope of this chapter, but the reader is directed to excellent reviews (de Kloet, Oitzl, & Joels, 1999; Kim & Diamond, 2002; Lupien & Lepage, 2001; Lupien & McEwen, 1997; McEwen, 2000). Collectively, findings from these studies converge on the idea that elevated levels of glucocorticoids released during stress disrupt normal hippocampal physiology, which, in turn, leads to changes in specific forms of memory. These effects do not, however, occur alone. Instead, they occur in the context of a larger stress response. Other neurochemical systems interact with glucocorticoids to mediate the effects of stress on memory and hippocampal function, including norepinephrine, serotonin, dopamine, amino acids (e.g., GABA, glutamate), and immune compounds. Because many reviews focus exclusively on the glucocorticoids, a brief mention of the other systems will be appropriate here. Exposure to stress causes the release of norepinephrine from the locus coeruleus (a nucleus in the pontine region of the brainstem), and this is often thought of as the “first wave” response to stress. The norepinephrine system responds almost immediately to a stressor, whereas the HPA axis, or “second wave” response, reacts more slowly. Norepinephrine binds to receptors throughout the brain and facilitates motor responsiveness, thus governing the “flight or fight” response (e.g., Stoddard, Bergdall, Townsend, & Levin, 1986). Moreover, norepinephrine has potentiating effects on memory for emotional experiences, a finding that may be mediated by the numerous norepinephrine receptors located in the amygdala (e.g., McGaugh, 2000). Norepinephrine can reach very high levels and still facilitate memory for emotion, even while similarly high levels of glucocorticoids impair contextual and verbal episodic memory. Exposure to stress also produces serotonin release in the hippocampus, medial prefrontal cortex (PFC), and amygdala, among other brain regions (e.g., Inoue, Tsuchiya, & Koyama, 1994; Kaehler, Singewald, Sinner, Thurnher, & Philippu, 2000), and stress-induced alterations in serotonin impact memory function. For example, substances that reduce extracellular serotonin levels appear to block the effects of stress on memory and hippocampal function, suggesting that serotonin release during stress might contribute to hippocampal damage (McEwen et al., 1997; McEwen, 2000). Tianeptine, an antidepressant that works by enhancing serotonin reuptake (and therefore reducing extracellular serotonin levels), blocks both stress- and corticosterone-induced dendritic atrophy of neurons in the CA3 region of the hippocampus (Watanabe, Gould, Daniels, Cameron, & McEwen, 1992).
89
Inhibition of hippocampal neurons and associated memory deficits may also be mediated by a stress-induced increase in proinflammatory cytokines. Specifically, interleukin-1 (IL-1), which is released by macrophages in response to stress, may mediate stress-induced memory impairments by stimulating serotonin release in the hippocampus (Linthorst, Flachskamm, Holsboer, & Reul, 1994) or directly interfering with LTP (Bellinger, Madamba, & Siggins, 1993) by disrupting calcium influx mechanisms (Cunningham, Murray, O’Neill, & O’Connor, 1996). In any case, IL-1 may play a key role in mediating hippocampusspecific memory impairments. For example, IL-1 induction has been found to impair performance on spatial (i.e., hippocampal-dependent) versions of the Morris water maze task but to leave performance on the nonspatial versions unaffected (Gibertini, Newton, Friedman, & Klein, 1995). IL-1 has also been shown to interfere with fear conditioning to contextual but not discrete cues (Aubert, Vega, Dantzer, & Goodall, 1995). The stress response also involves the dopaminergic system, primarily the system that innervates the medial PFC. This system appears to be vulnerable to mild and brief stress, and some studies suggest that stress has a negative effect on dopaminergic function. This is important because this region, which appears essential for the normal functions associated with working memory, shares responsibility with the hippocampus for processing the source of information (source is a type of contextual information specifying where, or from whom, something was learned) and plays a role in episodic binding. Dopamine turnover is necessary to ensure proper PFC function; in line with this observation, recent evidence suggests that working memory is impaired by exposure to extreme or chronic stressors (Arnsten, 2000). Dopamine may play a role in emotional memory as well, perhaps mediating the cognitive modulation of fear responses (see Pani, Porcella, & Gessa, 2000). Benzodiazepines play a role in stress and memory by potentiating the effects of the inhibitory neurotransmitter GABA. Blocking benzodiazepine receptors increases the experience of anxiety, while Valium, a benzodiazepine receptor agonist, decreases it. Results from several studies suggest that chronic stress reduces benzodiazepine receptor binding in the frontal cortex, not only increasing the subjective experience of stress and anxiety but also perhaps impairing hippocampal function (McEwen, 2000). Moreover, recent experiments in young animals show that inducing stress through maternal deprivation also reduces benzodiazepine receptor binding in the frontal lobes, as well as the amygdala and locus coeruleus (Caldji, Francis, Sharma, Plotsky, & Meaney, 2000; cf. Bremner & Vermetten, 2001). Finally, excitatory amino acids and endogenous opiate peptides may be involved in the stress-induced impairment of hippocampal function (Moghaddam, Bolinao, Stein-Behrens, & Sapolsky, 1994). In brief, excitatory amino acids (e.g., glutamate) may abet glucocorticoids in causing pyramidal cell loss in the hippocampus (e.g., Virgin et al., 1991). Stress- and glucocorticoid-induced atrophy
90
of dendrites can be blocked by the anti-epileptic drug phenytoin (Dilantin), which works primarily by blocking the release of glutamate. Some evidence suggests that glucocorticoids and stress elevate levels of glutamate in the hippocampus and other brain areas (Stein-Behrens, Lin, & Sapolksy, 1994) and that a high concentration of glutamate is toxic to neurons. In addition, excitatory amino acids aid in the damage of hippocampal neurons in culture and appear to exacerbate kainic acid-induced damage, as well as ischemic damage, to the hippocampus (Stein-Behrens et al., 1992). NMDA receptors are thought to be involved, because blockade of such receptors is another effective means of preventing stress-induced atrophy of dendrites (McEwen et al., 1995). Clearly, then, glucocorticoids are accompanied by, and in some cases modulated or mediated by, other neurochemicals that respond to stressors. This point is important because it suggests that any effects of stress on memory may reflect complex cascades; thus, understanding stress- or trauma-induced memory impairments in humans, to which we now turn our attention, will not be a simple matter.
Chronic Stress: Memory in Humans As in animals, the memory-impairing effects of acute stress or glucocorticoid administration are generally reversible in humans (Lupien & McEwen, 1997; McEwen, 2000).4 Whether the effects of chronic stress are reversible, however, remains an open question. Research generally suggests that chronic glucocorticoid elevation is associated with memory dysfunction in humans (e.g., Mauri et al., 1993). Although some studies examining the long-lasting effects of Cushing’s disease suggest that moderate recovery from cognitive deficit after cortisol reduction may be possible (Starkman et al., 1999), other studies suggest more permanent damage (Sheline et al., 1996; see also Bremner, 2001b). Memory deficits associated with aging in humans appear related to progressive dysregulation of HPA activity with increasing age. For instance, Lupien et al. (1994) reported that elderly individuals whose cortisol levels increased over a 4-year period performed significantly poorer on tests of explicit memory than individuals whose cortisol levels declined over the same period (see also Lupien et al., 1998). These observations led to the clinically relevant idea that some individuals who endure life-threatening experiences with associated increases in stressrelated hormones might experience a dysregulation of their memory systems, perhaps for specific traumatic events and for autobiographical events in general. Traumatic Stress and Memory in Humans: The Special Case of Posttraumatic Stress Disorder For every 100 people who endure horrifying experiences (e.g., Holocaust survivors, victims of physical or sexual abuse or rape, war veterans, etc.), about 15 will develop posttraumatic stress disorder (American Psychiatric Association, 1994 [DSM-IV]), which characteristically involves a collection of memory
91
problems (see Bremner, 1999; Lemieux & Coe, 1995; Yehuda et al., 1995). Diagnostic criteria for PTSD include previous exposure to “an event or events that involved actual or threatened death or serious injury, or a threat to the physical integrity of self or others.” In addition, the person’s response must have involved intense fear, helplessness, or horror (DSM-IV). PTSD is characterized by intrusive recollections of the traumatic experience, avoidance of the event and reminders of the event, hyperarousal, dissociation, sleep disturbance, emotional numbing, and, paradoxically, increased emotional responsivity and anxiety (Emilien et al., 2000; Golier, Yehuda, & Southwick, 1997; Saigh & Bremner, 1999). Epidemiological studies suggest that the lifetime prevalence of PTSD in the general population ranges from 8 to 9 per 100 and that roughly 60% of these cases become chronic (e.g., Breslau, Davis, Andreski, & Peterson, 1991; Davidson, Roth, & Newman, 1991; Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). Prevalence rates may be even higher in children and adolescents and are twice as frequent in women as in men (e.g., Costello, Erkanli, Fairbank, & Angold, 2002; Kessler et al., 1995; see Fairbank, Ebert, & Caddell, 2001, for a more comprehensive review). Clinicians have long reported that memory difficulties are also associated with PTSD, noting that war veterans often forget appointments and what they had for breakfast yet describe the central aspects of their combat experiences in excruciating detail, as if those experiences happened yesterday (Bremner, 2001b). Although forgetting appointments was once understood as a psychoanalytical defense mechanism and treated as “resistance” on the part of the patient, a growing body of empirical research supports a more biological explanation for the memory problems seen in PTSD (Bremner, 2001b). Both short-term memory and episodic memory appear to suffer in those diagnosed with PTSD (Bremner et al., 1993; Bremner, 2001a, b; Torrie, 1944). Severe memory deficits have been measured in PTSD using portions of the Wechsler Adult Intelligence Scale (WAIS), the Wechsler Memory Scale (WMS), and the Verbal Selective Reminding Test (vSRT). For example, veterans with PTSD performed more poorly than healthy controls on the logical memory component of the WMS (about 2 of 5 show impaired immediate recall and 1 of 2 show impaired delayed recall; Bremner et al., 1993). More recently, Vasterling et al. (2002), using the Rey Auditory Verbal Learning Test (AVLT), showed that Vietnam veterans with PTSD manifested greater deficits on this working memory task than controls (Vietnam Veterans without diagnosed psychopathology). These authors demonstrated that the cognitive impairments associated with PTSD are independent of estimated verbal general intelligence (see, however, McNally & Shin, 1995). Similar results were obtained from a sample of adults reporting childhood sexual abuse (Bremner, Randall, Capelli, et al., 1995). These individuals performed more poorly than controls on the WMS Logical component for immediate and delayed recall, as well as percent retention. They also had difficulty with
92
the immediate recall of words as measured with the vSRT. These impairments cannot be explained by differences in general intelligence as reflected by IQ, which did not differ in either study. Vietnam veterans, Desert Storm veterans, and victims of rape diagnosed with PTSD all have impaired verbal episodic memory in comparison to control subjects (Jenkins, Langlais, Delis, & Cohen, 1998; Uddo, Vasterling, Brailey, & Sutker, 1993; Vasterling, Brailey, Constans, Borges, & Sutker, 1997; Vasterling, Brailey, Constans, & Sutker, 1998; Yehuda et al., 1995, but see Stein et al., 1997). Although these studies show that PTSD and memory deficits are related, design limitations prevent unequivocal causal conclusions. The first limitation is associated with the well-understood problems of quasi-experimental designs. The second concerns a failure to examine memory performance in the context of other neuropsychological functions, which leaves open the question of whether PTSD patients exhibit selective memory impairments or more general cognitive impairments. The third concerns lack of control over the potential relations among different cognitive functions, leaving us asking, for example, if memory changes in PTSD can be explained more generally by changes in attention, concentration, visuospatial abilities, or other cognitive functions related to memory. The fourth concerns what accounts for changes in memory observed under these conditions; memory deficits may be due to development of PTSD symptoms per se, to the result of trauma exposure generally, or to some unknown set of factors. A fifth limitation concerns comorbidity, that is, whether other disorders associated with changes in memory (e.g., depression, alcoholism) might account for many of the memory effects observed in PTSD. Horner and Hamner (2002) reviewed 19 studies examining cognitive function in PTSD and concluded that converging evidence for memory deficits in PTSD does exist. They caution, however, that methodological problems, particularly comorbidity (e.g., substance abuse, mood disorders, and other anxiety disorders), were a problem in most studies. Sixteen of the 19 studies reported evidence of attention or memory difficulties, or both, in PTSD. However, 15 of these studies included patients with other comorbid disorders, reflecting the difficulty recruiting cases of “pure” PTSD—and raising the question of whether such a pure sample exists. An important study by Gilbertson, Gurvits, Lasko, Orr, and Pitman (2001), which was not discussed in the Horner and Hamner (2002) review, examined performance on a range of neuropsychological tests including memory, attention, visuo-spatial skills, and executive functioning. These authors reported that, although those meeting criteria for PTSD were impaired on all these tasks compared to controls, performance on the attention and memory tasks uniquely predicted a PTSD diagnosis. In addition, verbal memory emerged as distinguishable from generalized attentional differences, as well as from the severity of the triggering trauma, general intelligence, depression, alcohol use, or a history of
93
developmental learning problems (Gilbertson et al.). This study provides some of the first evidence for a specific association between changes in episodic memory and the presence of PTSD. Autobiographical memory also suffers in PTSD. Individuals diagnosed with PTSD often have overly general (i.e., nonspecific, lacking context and detail) memories of the past (McNally, Lasko, Macklin, & Pitman, 1995) and show selective deficits when retrieving specific personal information in response to cue words having positive (e.g., “kindness”), neutral (e.g., “appearance”), or negative (e.g., “panic”) emotional valence (Kaspi, McNally, & Amir, 1995; McNally, Litz, Prassas, Shin, & West, 1994; McNally, 1998). When individuals with PTSD were asked to retrieve a personal event memory triggered by a cue, they tended to retrieve generic memories without reference to a specific event. That is, rather than reporting contextually specific memories with specific source (time, place) information, they appear to report schema- or knowledge-based narratives. Moreover, they report difficulty retrieving specific positively valenced personal memories, suggesting that trauma-related thoughts hamper access to positive or neutral memories. As these results clearly demonstrate, memory problems are a hallmark of PTSD. In addition to general memory effects, those with the disorder tend to either vividly re-experience (i.e., “flashbacks”), or fail to remember, information related to their traumas. PTSD can thus be characterized as a disorder that paradoxically includes excessive and uncontrolled memory retrieval (e.g., intrusive memories, nightmares, flashbacks) and memory disruption (e.g., traumatic amnesia, fragmented memory).
Stress and Memory: What Is
Forgotten and Why
Although the foregoing discussion makes it clear that stress can impair memory, the data leave one wondering precisely what aspect(s) of memory are impacted by stress. All aspects, and all systems of memory, obviously are not entirely disrupted; if they were, researchers would be left with nothing to measure. So what, exactly, does stress cause us to fail to learn, fail to store, or fail to retrieve? In this section, we suggest that (1) contextual memories (i.e., memories of the spatial and environmental aspects of experience), (2) detailed memories (i.e., those details that allow episodic memories to be distinguished from one another), and (3) coherent, explicit autobiographical memories (i.e., the conscious, contextually, or episodically grounded recall of material with details that fit together in sensible, meaningful ways) suffer in the presence of high levels of stress.5 What evidence is there that memory for stressful experiences lacks coherence, context, and episodic detail? Initial support came primarily from anecdotal clinical reports, which suggested that experiences encoded during high levels of stress
94
are often fragmented (Bremner, 1999; Golier et al., 1997; Gray & Lombardo, 2001; van der Kolk, Hopper, & Osterman, 2001; van der Kolk, 1994). The bits and pieces of experience are not related to one another as a whole. That is, these pieces and patches are not bound together in a way that produces a “good gestalt,” or a coherent episode. Fragmentation is an important feature of PTSD. Patients sometimes describe “gaps” in recalled experiences, not only of trauma but also of other personal experiences (Bremner, 1999). Some laboratory evidence suggests that stress-induced fragmentation can emerge in experiments assessing memory for verbal material (e.g., Payne et al., 2002, described later); it is, however, most commonly observed in response to trauma. Rather than simply retrieving fragments and reporting them as such, some individuals who have experienced trauma appear to make educated guesses about memory in a process called “narrative smoothing.” Jacobs and Nadel (1998) offered a neurobiological theory of reconstructed memory taking this phenomenon into account. Starting with the observation that multiple memory systems underlie the storage of episodic information, they suggested that retrieving an autobiographical memory entails accessing and integrating fragments of information stored in these different systems. Many different types of information go into the retrieval of a specific experience. There are the elements composing the episode: what happened; who engaged in which actions; what it all looked like, sounded like, smelled like; etc. In addition to this episodic “content,” there is also spatial or contextual information. This is a particularly important type of information because all episodes occur someplace and thus are defined partly by the space or context in which they occur. It can be difficult if not impossible to divorce memories of episodes from memories of context, a notion that led us to suppose that context serves as an organizing frame to which the elements of an episodic memory trace are attached (Jacobs & Nadel, 1998; Nadel & Payne, 2002; Nadel, Payne, & Jacobs, 2002). Consider a hypothetical traumatic war experience; a memory of this experience might include smells (the jungle, unwashed bodies), sounds (gunfire and cursing), sights (the flash of a sniper’s weapon, the sight of a wounded combatant), tactile feelings (the humidity, thorns slashing skin), actions (diving for cover, returning fire), and emotions (anger and fear). Each of these independent features is stored in the relevant part of the brain, typically, but not exclusively, in the neocortex. The hippocampus and adjacent medial temporal regions appear to be critical to a process by which these disparate fragments of information from multiple brain regions are bound into a unified memory trace at the time of retrieval (Mitchell, Johnson, Raye, Mather, & D’Esposito, 2000; Mitchell, Johnson, Raye, & D’Esposito, 2000). At the same time, the hippocampus appears to be critical for placing an autobiographical memory in time, place, and context. Hence, the disruption of hippocampal function by high levels of stress and glucocorticoids could disrupt the storage of information about place, time, and context during a
95
traumatic experience, as well as the binding process that connects features of the experience to the context and to one another (Bremner, Krystal, Charney, & Southwick, 1996; Jacobs & Nadel, 1998). The idea that memories are disaggregated during storage and then reaggregated during retrieval has several implications in explaining interactions between extreme stress and memory (Jacobs & Nadel, 1998). By disrupting the hippocampally based binding function, stress may leave the pieces, patches, and context of a memory trace disconnected. So memories formed in the presence of stressors, and hence under the influence of high levels of glucocorticoids, will be disjointed and fragmented, lacking spatial or temporal context, and thus creating deficits in episodic memory. Although traumatic stress appears to disrupt hippocampally based features of memory (e.g., context, episodic detail), quite the contrary appears to be the case for emotional features of memory based in other brain regions.6 The amygdala, a brain region adjacent to the hippocampus, is known to process, and perhaps store, emotional memories (Fanselow & LeDoux, 1999; LeDoux, 2000). The amygdala is active during emotionally charged conditions, and the emotional memories dependent on this region are highly resistant to forgetting (LeDoux, 1992). This “stamping in” of emotional memories is at least partly mediated by the stress hormone norepinephrine, which, as noted earlier, also reaches high levels during stressful experiences (see McGaugh, 2000, for a review). Hence, two prominent stress hormones, both released during intense stress, have opposite effects on different structures in the brain. High levels of cortisol disrupt hippocampal function, impairing episodic (e.g., de Quervain et al., 2000; Payne et al., 2002) and spatial memory (Laurance et al., 2001). In contrast, high levels of both norepinephrine and cortisol facilitate amygdalar functions and the emotional memories dependent on them (see Buchanan & Lovallo, 2001; Metcalfe & Jacobs, 1996, 1998, 2000). Jacobs and Nadel (1998) suggested that in the absence of an intact hippocampus-based memory system, the amygdala-based system stores emotional information unbound to the spatio-temporal context of the relevant events. This process results in a pool of emotional memories encoded without a coherent spatio-temporal frame to organize them. This pool is, essentially, a population of sensory and perceptual fragments acquired during the traumatic event (Jacobs, Laurance, Thomas, Luzcak, & Nadel, 1996). By this model, traumatic events do not typically lead to a complete eradication of memory but rather to the storage of fragments lacking an organizing framework. Devoid of this framework, traumatized individuals resort to “narrative smoothing”; retrieved memories are composed of re-aggregated representations cobbled together by a narrative based on gist, inference, and educated guesswork (Jacobs & Nadel, 1998; Nadel & Jacobs, 1998; see also Shimamura, 1997). By this model, a fragmentation of normally coherent elements of an experience resulting from disruption of hippocampally based contextual encoding,
96
coupled with the sparing or even enhancement of amygdala-based emotional encoding, leads to a theoretically interesting clinical picture of trauma and memory. Van der Kolk and Fisler (1995) presented evidence that traumatic events are remembered initially as disconnected images and waves of disjointed sensations and emotions. The emotions can emerge as punctate waves (van der Kolk & Fisler, 1995) or as persistent anxiety without an apparent source (e.g., Freud, 1926). Clinicians sometimes suspect that clients presenting with such generalized (or “freefloating”) anxiety—an emotional disturbance without obvious anchors in autobiographical memory—have suffered past traumas. Claims about generalized anxiety are supported by some empirical evidence, which suggests that memory disturbance in response to trauma may be associated with dysregulation of the autonomic nervous system, heightened arousal or vigilance, and increased startle reflex (Joseph, 1996). Indeed, emotional dysregulation is also characteristic of PTSD, and while patients often suffer from extreme anxiety, they may also exhibit isolation of affect (described in the following) and describe emotional blunting and feelings of numbness. Cortisone therapy (a common treatment for rheumatoid arthritis) has been associated with both anxiety and panic, on the one hand, and with affective dissociation, depersonalization, and emotional flatness, on the other (Clark, Bauer, & Cobb, 1952). These observations are reminiscent of a defense mechanism proposed late in Freud’s career: defensive “undoing” (Freud, 1926). Freud thought of anxiety as a marker that emerges when the mind cannot cope with the threatening information assailing it. Thus, it pushes autobiographical information associated with the trauma out of awareness, and anxiety emerges instead. Like much of Freud’s work, defensive undoing is an intuitively appealing idea but difficult to test. In contrast, the biological response to stress offers a testable and more rigorous explanation of this phenomenon. If high levels of cortisol disrupt normal neuronal function of the structure responsible for the storage of critical contextual information (i.e., the hippocampus), then coherent memories of trauma will be rendered inaccessible.7 Moreover, if physiological processes impair cognitive but not emotional aspects of memory, as suggested by the opposite actions of catecholamines on amygdala function and cortisol on hippocampal function, then feelings of anxiety disconnected from the corresponding “factual” or autobiographical memory of a traumatic event might be expected. In other words, emotional memory for the traumatic event might continuously enter awareness in the form of generalized feelings of anger, fear, or uneasiness (referred to as “body memories” by van der Kolk, 1994) unbound to the originating episodic context. Without a contextual framework to relate emotional memories to associated traumatic events, these emotional memories surface as vague disconnected feelings of unease, anger, or fear that are troublesome. Anxiety is thus experienced as “generalized” and “freefloating.”
97
Isolation of Affect in PTSD:
A Paradoxical Memory Deficit
Perhaps the most common pattern of deficit seen in PTSD involves what we described: a disruption of episodic memory, along with the experience of extreme anxiety apparently devoid of a source. Clinicians also describe cases exhibiting the opposite pattern of forgetting, however, when a person retains episodic information concerning a traumatic event but loses the corresponding emotional response (see chapter 11). Modern clinicians use the terms “emotional blunting” or “emotional numbing” to describe this pattern (e.g., Golier et al,, 1997; chapter 11). However, this observation is reminiscent of Freud’s concept of “isolation of affect,” when people may recount traumatic experiences in the absence of the emotion that must have initially accompanied them; these individuals are, in fact, often overtly blasé about the upsetting information they recount (also called “splitting off of affect”; Freud, 1926). Thus, isolation of affect involves a lack of emotion and may contribute to experiential aspects of PTSD (Harvey & Bryant, 1998; van der Kolk, 1991). Freud speculated that individuals experiencing emotional numbness do not repress traumatic memories, thereby holding disturbing information out of awareness. Rather, unlike the failure to remember in “undoing,” individuals experiencing emotional numbness have no trouble recollecting the episodic aspects of the trauma. They do, however, have difficulty consciously accessing the emotion that originally belonged to it. By this conjecture, these individuals remember only the cognitive aspects of trauma (Mitchell & Black, 1995). To bring memory of the emotion into awareness would be too threatening, unduely straining the ego. Although there is little or no epidemiological data on which to rest the case, isolation of affect (emotional blunting) lends itself to a biological explanation. One possibility (we speculate on it in the next section) is that stress disconnects normal chains of communication among normally interacting hippocampal, amygdala, and cortical networks. A Speculative Neurobiological Account of Isolation of Affect Metcalfe and Jacobs (1996, 1998, 2000), working within the theoretical framework just outlined, explored interactions between two learning/memory systems, (1) a cool system, based in interactions between hippocampal and frontal regions, that is slow, emotionally neutral, flexible, integrative, episodic, strategic, and highly contextualized and (2) a hot system, based in the amygdala, that is the basis of emotionality, fears, and passions, that is fast and impulsive and controlled by innate and acquired releasing stimuli. Stress has a different impact on the cool hippocampal-based system and the hot amygdala-based system. Exposure to intense stressors or trauma disrupts the hippocampus, leaving disaggregated memories ruled by the cool system. Exposure to these same stressors, however,
98
enhances the hot system, encoding context-free emotional memories rapidly and efficiently. This approach seems to account for many features of recovered memory, specific phobia, panic attacks, and PTSD (Jacobs & Nadel, 1985, 1998; Nadel & Jacobs, 1996; Metcalfe & Jacobs, 1996), with one glaring exception. As we have described, modern clinicians often describe an emotional blunting associated with the presence of PTSD; some even describe this as a “core defining feature” of PTSD (e.g., Litz & Gray, 2002). To offer an account of emotional blunting (i.e., isolation of affect), we must consider two possibilities, neither well supported empirically. The first possibility is that chronic exposure to glucocorticoids damages neurons in the amygdala, disrupting its processing of emotional information. The most serious weakness of this proposal is that chronic stress does not appear to damage the amygdala. Indeed, there is some evidence that, in the rat, predictable exposure to a stressor simultaneously disrupts dendritic arborization in the hippocampus but enhances this same process in the amygdala (Vyas, Mitra, Rao, & Chattarji, 2002). Under these conditions, one expects to observe disruption of hippocampal-mediated memory (e.g., contextualization) and enhancement of amygdala-mediated memory (e.g., emotional memory). In contrast, chronic exposure to unpredictable stressors leaves the dendritic arborization process relatively intact in the hippocampus but induces dendritic atrophy in the basolateral nucleus of the amygdala (Vyas et al., 2002). Under these conditions, one expects to observe highly contextualized episodic and emotional memories but greatly disrupted processing of emotional memories. That is, one expects broad emotional flattening. In the presence of a damaged amygdala, however, one would expect a global disruption of emotional processing, yet this apparently does not happen to individuals who experience isolation of affect. Instead, these individuals appear to be unaware of their emotional reactions while simultaneously showing enhanced behavioral and physiological reactivity (see, for example, Jacobs, Nadel, & Hayden, 1992). This finding recalls work reported by Peter Lang and his colleagues and introduces a second possible explanation of isolation of affect. Lang and Cuthbert (1984) conceptualized anxiety as including three loosely coupled response systems (constellations) of overt behavior, verbal report (cognition), and physiological activation. Within this framework, those who experience isolation of affect behave as if they do not experience subjective emotion (i.e., “feelings”) but do experience heightened arousal at the level of both behavioral and physiological constellations (Jacobs et al., 1992). It is as if the physiological concomitants of emotion do not enter awareness. The physiological changes occur but without corresponding emotional valence. More formally, affect is not expressed in the cognitive constellation, while heightened emotion (e.g., hypervigilance, in-
99
creased startle, etc.) remains expressed in both behavioral and physiological constellations. We account for this pattern of responding in isolation of affect by making several potentially defensible assumptions about interactions among stress, stressors, and the functions of hippocampus, amygdala, and PFC.
Stress and the Frontal Cortex:
Stress-Induced Forgetting Beyond
the Hippocampus
We have focused almost exclusively on the hippocampus and amygdala. The frontal lobes, however, have received increasing attention in the stress literature. Recent experiments using functional magnetic resonance imaging (fMRI) (Henson, Shallice, & Dolan, 1999) and electrophysiological techniques (Otten & Rugg, 2002) have demonstrated an important role for the PFC in memory (Raye, Johnson, Mitchell, Nolde, & D’Esposito, 2000; Buckner, Logan, Donaldson, & Wheeler, 2000). Traditionally, the PFC was thought to contribute mainly to encoding processes through its presumed role in working memory, which allows information to be maintained temporarily in an active state (Baddeley, 1986). New evidence from PET and fMRI studies, however, shows that the PFC is also involved in memory retrieval (see Henson et al., 1999; Nyberg & Cabeza, 2000; Rugg, Flethcer, Chua, & Dolan, 1999) and that the PFC may share, with the hippocampus, responsibilities for binding episodic material (Mitchell, Johnson, Raye, & D’Esposito, 2000). The frontal cortex is also involved in the modulation of emotion and inhibitory responses (e.g., Arnsten, 2001; Rosenkranz & Grace, 2002). Moreover, damage to this area, particularly if it is confined to the prefrontal region, leads to perseveration, difficulty choosing correct responses, and distortions and confabulations of memory. For example, patients with frontal lobe damage tend to make an abnormally large number of confident false recognition errors, in spite of good recall and recognition generally (Curran, Schacter, Norman, & Galluccio, 1997; Delbecq-Derouesne, Beauvois, & Shallice, 1990; Melo, Winocur, & Moscovitch, 1999; Parkin, Bindschaedler, Harsent, & Metzler, 1996). With the existence of glucocorticoid receptors in the frontal lobes (reviewed later), stressrelated increases in commission errors likely are not caused solely by a disrupted hippocampus. An important review by Lupien and Lepage (2001) underscores the contribution of frontal regions to stress-related effects on episodic and emotional memory. These authors argue that the hippocampus has been studied at the expense of other brain regions likely involved in glucocorticoid action and that attempts to explain stress impairments on memory solely on the basis of the impact of stress on the hippocampus may be incomplete.
100
Lupien and Lepage (2001) note that both Type I (mineralocorticoid, MR) and Type II (glucocorticoid, GR) stress receptors are present in cortical regions, with a preferential and dense distribution in the PFC. Hence, stress may exert effects not only on hippocampal neurons but also on neurons critical for the normal functioning of the PFC. If one takes the influence of stress and corticosteroids on human memory to reflect the action of GRs, then one must acknowledge that both hippocampal and PFC function are likely involved. Differential MR and GR action in the hippocampus and PFC may underlie some of the differences in the magnitude and nature of memory impairments reviewed in this chapter. The hippocampus, mainly through MR, but also moderate GR, receptor binding, may be a primary site of short-term, reversible effects that occur in response to stress (e.g., de Quervain et al., 2000; Newcomer et al., 1994, 1999; Payne et al., 2002; Wolkowitz et al., 1990, 1993). MRs must be totally saturated before GRs become activated, and significant reductions in LTP do not take place until both MRs and GR receptors are occupied. Thus, experiments using humans (for obvious ethical reasons) likely rarely attain the levels of stress needed to produce lasting changes in memory. On the other hand, exposure to trauma may lead to the widespread and sustained saturation of both MR and GR receptors in the PFC and hippocampus. These conditions cannot be ethically reproduced in the laboratory and thus may emerge only in the context of naturally occurring chronic stress in the patient populations described earlier (e.g., Cushing’s syndrome, PTSD, major recurrent depression, etc.). In such cases, extensive binding of cortisol to both receptor types in the hippocampus and PFC may lead to long-lasting and harmful changes in neuronal functioning in the entire memory circuit, resulting in dramatic changes in how environmental information is encoded, stored, and retrieved. Evidence that excessively high levels of norepinephrine (e.g., during exposure to uncontrollable stress) impair cognitive functions of the PFC (Arnsten, 1998) makes this concept all the more feasible. Consequently, exposure to intense stress appears to affect the PFC, as well as the hippocampus and amygdala. Models of stress and memory that focus solely on the hippocampus and amygdala cannot adequately account for isolation of affect, but perhaps the addition of the PFC can explain this phenomenon. Does the PFC play a role in the modulation of emotions? What happens to that role under conditions of extreme stress? And, if we can specify that role adequately, how does this help us understand isolation of affect? We start by assuming that the PFC plays two roles in emotional processing and regulation: (1) inhibitory control of emotional responses generated by an amygdala-centered system and (2) subjective awareness of that emotion (e.g., Amaral, Price, Pitkanen, & Carmichael, 1992; Reiman, Lane, Ahern, Schwartz, & Daivdson, 2000; Rosenkranz & Grace, 2002; Anderson, Bechara, Damasio, Tranel, & Damasio, 1999). Under the model outlined earlier, these assumptions have two important implications. First, disruption of the PFC should disinhibit
101
the amygdala and thereby amplify the behavioral and physiological constellations of emotions controlled by it. Second, disruption of the PFC should disrupt the subjective experience of emotionality. Under these conditions, we expect individuals to experience high levels of emotional arousal in the behavioral and physiological constellations but not in awareness. This process would then account for narrow (or cognitive) emotional flattening with enhanced reactivity in the behavioral and physiological constellations. Although the model remains to be rounded out and tested, the welldocumented processes that simultaneously impair PFC- and hippocampalbased memories and facilitate amygdala-based memories provide empirical physiological support for defensive undoing and emotional blunting, or the isolation of affect.
Trauma and Memory:
What Is Remembered?
Intense stress is associated with memory disruption, either partial or complete, relating to the trauma itself or to episodic memory in general. Yet, paradoxically, other cases of trauma provoke an inability to forget (hypermnesia) and a desire to talk about the experiences constantly (e.g., Terr, 1991, 1993). Along similar lines, some individuals report spontaneous flashbacks and hallucinations, as well as severe emotional upset in relation to the trauma. Sometimes these flashbacks appear to be triggered by environmental stimuli (e.g., the sound of a firecracker or an engine backfiring) and are reportedly so real that the person feels he or she is re-experiencing, rather than remembering, the event. The fact that people report flashbacks suggests that some representation of traumatic events has been stored but is not always accessible. Interestingly, flashbacks reportedly plague trauma survivors regardless of whether they report amnesia for traumatic events or experience hypermnesia and talk about them obsessively.8 Further, whether they experience flashbacks may be unrelated to general level of memory function (e.g., Joseph, 1996). A thorough investigation of flashbacks is beyond the scope of this chapter. Nonetheless, we briefly note that an understanding of the neurobiological processes involved may help elucidate flashback phenomena. Although fear is tied to specific stimuli in a person’s immediate environment, anxiety appears to be a secondary response (LeDoux, 1992). Pathways of fear involve interactions between cortical sensory processing areas and the amygdala, where emotionally neutral sensory signals are transmitted to receive their affective connotation. Depending on whether the emotional processing mechanisms of the amygdala are activated directly by incoming sensory information or indirectly by cognitive processes organized elsewhere, a person will either experience fear, a primary response, or anxiety, a secondary response (Davis, 1992; Davis & Shui, 1999; Gray, 2000). The former pathway may initiate flashbacks, which elicit fear di-
102
rectly and are experienced in the here and now. The secondary pathway may characterize “normal” reactivation of trauma memories, which should elicit negative affect and anxiety that are correctly referenced to the past. This latter pathway appears to be disrupted during the trauma, leading to decontextualized and fragmented memory for the experience. Acknowledging the differing contributions of these two pathways, we might expect the content of flashbacks to be highly emotional, as fear circuits become activated. The fact that amygdala function is preserved during intense stress fits this suggestion. Moreover, due to stress-induced disruption of the structures underlying context and detail memory (i.e., hippocampus and PFC) during the trauma, we might expect flashbacks to lack both contextual information and details outside the emotional center of the traumatic event. In the absence of contextual detail, flashbacks should consist mainly of the central content or theme of the traumatic experience.9 In other words, we expect the content of flashbacks to be rarely, if ever, about the peripheral details of a traumatic experience. Rather, we expect the images that emerge during a flashback to be the most horrible, emotionally salient, and central features of the traumatic experience. These considerations force us to examine another literature, one that seemingly unfolded parallel to the trauma and memory literature.
The Emotion and Memory Literature We have focused almost exclusively on memory impairment that results as a consequence of stress and trauma. Many readers, however, particularly those with a background in emotion, may be wondering how these data fit with evidence seemingly pointing in the opposite direction—namely, that memory often improves in response to emotion, even intense emotion (see also chapter 2 here). A key finding in the emotion and memory literature is that memory for emotional events is superior to memory for similar, neutral events (e.g., Heuer & Reisberg, 1990; chapter 1 here). Most people have ready examples of emotional events that are remembered vividly “as if they happened yesterday,” and a substantial literature confirms that some emotional experiences are unforgettable (see chapter 1). People also report vivid recollections for intensely negative experiences accompanied by strong negative emotion (e.g., Porter & Birt, 2001). So we now have one literature demonstrating that high levels of stress can disrupt memory for stressful experiences and another literature demonstrating that intense negative emotion can improve memory for negative experiences. The key to this apparent conflict, we think, is embedded in an as yet unacknowledged difference between strong negative emotion and stress. Although stressful experiences are usually negative, and although substantial overlap may connect negative emotion and stress, we claim that they are not the same. In the following pages, we try to address the tension between these two literatures directly. Although at first blush data from the two research traditions seem
103
contradictory, we suggest that a solution resides in the differential contributions of hippocampus-based circuits and amygdala-based circuits. Can intensely negative memories be vivid and intact in some cases and disrupted and fragmented in others? Can traumatic memories be both vivid and fragmented? The short answer to both questions, we think, is yes. The key to resolving this apparent paradox likely resides in (1) the magnitude of the stress, (2) the corresponding amount of stress hormones released during emotional experiences of different intensities (i.e., the degree of stress inherent in a given emotional experience), and (3) the degree to which changes in these stress hormones affect different memory structures. The thrust of our argument resides in the (possibly nonlinear) relations between experienced emotion and the magnitude or intensity of the stress response. However, we briefly point out two additional difficulties that may help account for the contradictory literatures. The first difficulty is definitional. It concerns the often-overlooked fact that there are multiple memory systems. Memory is not a single entity that goes up or down in response to emotional arousal; rather, memory is an aggregate of many systems of which emotional memory is just one. If this is true, emotional arousal (even in the absence of stress) may facilitate some attributes of “memory” yet disrupt others. The second concerns how “emotion” is conceptualized and where and how it comes into play in memory experiments. In the stress/trauma literature, emotion (emotional memory) is a dependent variable, one of several types of memory differentially affected by stress, the independent variable. Conversely, in the emotion literature, emotion (or emotional arousal) is often used as an independent variable, which influences different types of memory (usually enhancing memory for emotional content: words, pictures, videos, experiences, etc.). Subtle differences between emotion and stress need untangling, but distinguishing between emotional content and emotional arousal that may be associated with that content is critical. We consider each of these problems. Something about emotionality leads to particularly vivid and long-lasting memories. Attempts to isolate mechanisms underlying this effect have led to discoveries anchored in research techniques derived from cognitive theory (e.g., chapter 1) and in research techniques derived from learning/neuroscientific theory (e.g., McGaugh, 2000, 2002). The Learning/Neuroscientific Approach to Emotion and Memory Research A crucial contribution of the neuroscience approach is the unshakable notion that specific neural and hormonal mechanisms underlie the enhanced consolidation of memory for emotional materials. One group of researchers (Cahill, Babinsky, Markowitsch, & McGaugh, 1995; Cahill & McGaugh, 1998), for example, has demonstrated that stress hormones interact with the amygdala and, through that interaction, enhance the storage of explicit memories. This group has shown that adrenergic activity enhances (e.g., O’Carroll, Drysdale, Cahill, Shajahan,
104
& Ebmeier, 1999; Southwick et al., 2002) and blockade of adrenergic activity reduces (e.g, Cahill, Prins, Weber, & McGaugh, 1994) the consolidation of explicit memory for emotional materials in humans. The activation of beta-adrenergic receptors, located on neurons in the basolateral nucleus of the amygdala, appears critical for this noradrenergic mediation of emotional memory consolidation (McGaugh, 2002; McGaugh & Roozendaal, 2002). In addition to adrenergic influences on memory, McGaugh, Cahill and their associates point out that low doses of glucocorticoids can enhance memory as well, suggesting that both epinephrine (and norepinephrine) and cortisol enhance memory consolidation (Cahill & McGaugh, 1998). Indeed, the enhancing effects of cortisol on memory for emotional (i.e., nonhippocampal-dependent) information have been demonstrated in animal studies of aversive conditioning and in human studies that use emotional memory tasks (e.g., Buchanan & Lovallo, 2001; Roozendaal, 2000). For example, Buchanan and Lovallo showed that 20 mg of hydrocortisone facilitated memory for emotional pictures, but not neutral pictures, relative to a placebo control. Again, the basolateral amygdala (BLA) is the specific nucleus thought to underlie these effects. The BLA is not only a locus of interaction between norepinephrine and cortisol (McEwen, 2000) but also a major hub of communication between the amygdala and hippocampus. As such, the BLA is thought to be a critical locus regulating norepinephrine and glucocorticoid influences on memory consolidation enhancement effects (Roozendaal & McGaugh, 1997). These findings appear to contradict the stress data we reviewed. Yet a close examination of the designs, the data, and the measures to obtain them leads to a proposal that could resolve this apparent contradiction, given a clear understanding of features composing emotional memory. The design of a typical study examining “emotion and memory” goes something like this. Human subjects watch a series of slides depicting successive moments in a story that is either emotionally arousing or emotionally neutral. A tape-recorded narrative accompanies the slides, helping subjects interpret the scenes and synthesize them into a story. Many studies have used variations on the “doctor/mechanic” slide sequences originally developed by Heuer and Reisberg (1990). In both stories, a mother and young boy visit the father at work. In the emotional series, the father is a surgeon performing explicit surgeries (where viscera or severed legs are in plain view); in the neutral series, the father is a mechanic working on a broken car, and disconnected car parts are in plain view. Studies in this tradition focus mainly on consolidation processes, so after waiting for a time (usually minutes after viewing the slides; although see Cahill et al., 1994), participants receive a substance that either enhances or blocks adrenergic activity. The participants receive a single test of memory, usually after a long delay. The typical result shows enhanced memory for the emotional slide sequence with adrenergic facilitation and decreased memory for the emotional slide sequence under adrenergic blockade. Neutral slides,
105
on the other hand, remain unaffected by adrenergic alterations (e.g., Cahill et al., 1994). These studies have made critical contributions to our understanding of the neural and hormonal mechanisms underlying emotional memory enhancement. Nevertheless, they tempt us to believe, perhaps prematurely, that emotional arousal improves emotional memory overall. Despite having borrowed emotional memory materials and designs from cognitive researchers, these researchers measure memory differently. Those from the learning-based tradition often obtain high-level aggregate measures of memory, measures that may not be sensitive to subtle nuances obtained when measuring the components of memory directly (although see Adolphs, Denburg, & Tranel, 2001). For example, Cahill and colleagues assess memory using traditional recall and recognition measures, focusing on the overall “amount” of memory preserved posttreatment. In contrast, cognitive researchers often examine qualitative differences within emotional memory—for example, differentiating memory for central or gist-like information from memory for peripheral detail (see chapter 1), often showing that only the former is enhanced by emotion. Hence, although arousal may lead to broad enhancement of emotional memory, we do not yet have the necessary data to draw such a conclusion. Until the effects of emotion on gist-like information and peripheral detail are directly compared, it may be safest to assume that memory enhancement maps onto a fairly robust pattern of findings in the cognitive literature—that emotional arousal promotes memory for an event’s gist or center, often at the expense of other (contextual) details (Adolphs et al., 2001; Metcalfe & Jacobs, 1998, 2000; chapter 2 here). The Cognitive Approach to Emotion and Memory Research: Central Versus Peripheral Information Results anchored in the cognitive tradition are generally consistent with the notion that increased levels of norepineprhine (and arousal) enhance recall of memory for emotional events. An early finding in the emotion and memory literature was that individuals initially remember highly unpleasant events more poorly than neutral events but that, over time, the pattern reverses, with enhanced recall for arousing events after a delay (e.g., Goodman, Hirschman, Hepps, & Rudy, 1991; Osborne, 1972). Clark, Milberg, and Ross (1983) and Kleinsmith and Kaplan (1963, 1964) demonstrated that highly arousing events are initially less well remembered than low-arousal or neutral events, but a recovery effect for the high-arousal events occurs after a delay. The eventual enhancement of memory by arousal is not surprising because such memories may be essential for survival. For instance, remembering to avoid a certain poisonous food could save one’s life (Bremner, Southwick, & Charney, 1997, 1999; McEwen, 2000). It seems safe to assume that negative emotional events lead to overall enhanced memory for these events, particularly after a delay. Recent research, however, suggests that the story isn’t quite so straightforward. First, newer studies show that emotional arousal can promote imme-
106
diate memory enhancement, implicating an effect of arousal on encoding, in addition to consolidation, processes (see Hamann, 2001, for a review). Second, and more important, many studies suggest that emotion enhances episodic memory but in a way often limited to the central and thematic features of such experiences (Burke, Heuer, & Reisberg, 1992; Christianson & Loftus, 1987, 1991; Loftus, Loftus, & Messo, 1987). People tend to remember the gist or affective significance of a stimulus or event at the expense of peripheral details (Christianson, 1992, Goodman et al., 1991; Ochsner & Schacter, 2000). This tendency may appear in the “weapon focus” phenomenon, when witnesses to crimes recall information about the weapon used during an assault at the expense of other important details (Stanny & Johnson, 2000). Loftus and Burns (1982), for example, reported that subjects who watched an emotionally arousing videotape of a violent bank robbery (in which a little boy is shot in the face) did not recall the events immediately preceding the attack as well as control subjects who watched a less emotional videotape (in which bank robbers simply ran past the little boy, leaving him unharmed). Similarly, Goodman et al. (1991) showed that receiving an unpleasant inoculation leads to enhanced memory for central information related to the procedure; likewise, Christianson and Loftus (1991) found enhanced memory for central information but impaired memory for peripheral information within an emotional event, as compared to a similar, neutral event. Burke, Heuer, and Reisberg (1992) demonstrated the same emotion-driven enhancement of central or gist-like information but added to our understanding of “centrality” by showing that memory improved both for visually central information and information associated with an event’s central theme or plot. As in other studies, however, memory suffered for details that were not associated with central events (see also Christianson & Loftus, 1991). Christianson and Loftus (1990) found that people describing their most traumatic memories reported more central than peripheral detail information. Study after study has shown that memory for peripheral detail suffers while memory for gist or thematic content improves with heightened emotionality (see chapter 1 for a comprehensive review and more detailed scrutiny of these data). There are several possible explanations for this phenomenon (see Heuer & Reisberg, 1992, also chapter 1 here). For example, central information may be better remembered simply because people attend to it longer. Increased memory for emotional information, however, can occur even when viewing time for emotional and nonemotional information is held constant (Christianson et al., 1991; Ochsner & Schacter, 2000). Another possibility is that the nature of attention may differ for emotional and nonemotional events, leading to better memory. Emotional events could also be associated with more elaborate processing and more rehearsal of the information to which people attend (Ochsner & Schacter, 2000). For example, witnesses to emotional attention-grabbing episodes (e.g., accident, crime, assault), compared with a neutral attention-
107
grabbing episode, may be more concerned with what they have just seen and think about (rehearse) it more often. According to Christianson and Safer (1995), this process of elaboration implies a psychological focusing on salient information that is the source of emotional stress, while at the same time limiting access to the information in the mental periphery. Although most discussion of these findings focuses on memory differences for central detail versus peripheral detail, there is another way to conceptualize the findings. As Christianson (1992) points out, central detail represents the source of the emotional arousal; it includes the most relevant information for extracting the emotional significance of a stimulus or event. So there may be an alternative definition of what centrality is and what it means for traumatic memory. First, central information may represent a concentration of experience, where disproportionate emphasis is placed on emotion. Whatever is emotional about an experience will therefore be well remembered, whereas more neutral information will not fare as well. Thus, emotion, likely through the activation of the adrenergic system, may be responsible for the memory enhancement of central experiences, because centrality and the source of emotionality often overlap. Second, central information may also represent a schematization of experience, where peripheral details are not well remembered but the gist or essence of an event is particularly well preserved. The process of schematization occurs when preserved emotional details are woven into a meaningful narrative memory on the basis of schematic inference. The result is a gist-based memory that may be biased toward central and emotional features of experience. By this view, schematized memories are likely thematically consistent, but they may not be accurate. Our notion of schematized memories dovetails with evidence suggesting that emotional memories are prone to errors and can even be entirely false (see chapter 1). Due to the way they are formed, schematically consistent memories are not always accurate, especially if one believes that “the truth is in the details.” Centrality and gist are quite compatible within this interpretation and likely represent the same information. Gist memories, we say, are composed of preserved bits of central information, linked together by schemabased inferences. Glucocorticoid binding at hippocampal receptor sites may shed light on this process, particularly if one thinks of peripheral details as contextually important features (i.e., specifying source information, spatial and temporal relationships) that rely on a functional hippocampus and central information as gist-like and relatively noncontextual, capturing the essence of experience but missing many of the episodic details that would produce complete and veridical memories. By this view, then, central information represents not only the emotional tone but also the overall gist or theme of a stressful experience. Central or gist-like information may survive, whereas detail memory may suffer, in response to emotional stress.
108
This claim brings us to the inevitable place where the words “emotion” and “stress” must be pulled apart. Despite obvious overlap between the two, emotional arousal is not always sufficiently intense to elicit a stress response, and occasionally a stress response is elicited but is relatively mild (and thus may enhance memory; see Cahill & McGaugh, 1998). Thus, while the pattern of enhanced gist and impaired detail memory may be influenced by definitional problems with these terms and limits in memory measurement (i.e., in the neuroscientific experiments previously described), the most compelling reason for such a pattern concerns the magnitude of negative emotion and, critically, the degree of HPA-axis activity associated with negative emotion. Stress is distinct from ordinary emotion insofar as HPA-axis activity and cortisol reach high enough levels for memory impairment. Whether emotion promotes memory for gist and centrality while disrupting memory for noncentral details may depend on the level of stress elicited by an emotional experience.10 Along these lines, we suspect that HPA-axis activity may be associated with the viewing of emotional slides and videotapes used in most emotional memory experiments. This hypothesis is easy to test if cortisol measures are taken during these procedures. A study of this nature is currently under way in our laboratory, with the goal of testing our predictions about how gist and detail memory fare in response to stress. Our suggestion finds support in other studies as well. There is good evidence in the memory literature that gist and detail memory are different (Neisser, 1981; Reyna & Kiernan, 1994; see also chapter 11 here, which distinguishes between “core memory” and “narrative memory”). The fuzzy trace theory of Brainerd and Reyna (1998), for example, posits two types of processing that eventually lead to two different representations of experience. During encoding, verbatim and gist traces are formed in parallel, creating a hierarchy of independent representations at varying levels of precision. In many cases, gist representations are more easily retrieved than detail representations, and we suggest that stress is one such case. Interestingly, fuzzy trace theory has gained attention as a theoretical framework that could explain a variety of false memory errors, and we note here that preservation of gist at the expense of detail may inspire “filling in,” or narrative smoothing, setting the stage for emotional memories that are true to gist yet inaccurate (see Golier et al., 1997; Heuer & Reisberg, 1992). Studies demonstrating increased memory for central, thematic, and appraisalrelevant information (e.g., Burke et al., 1992) are consistent with the results of stress studies by Payne et al. (2002) and Lindsay and Jennings (2000). Both studies sought to determine the effect of stress on false memory as assessed by the Deese-Roediger-McDermott (DRM) paradigm (Deese, 1959; Roediger & McDermott, 1995). In this work, participants studied a number of lists of semantically related nouns (e.g., candy, sour, sugar, bitter, chocolate, cake, etc.), followed by a four-item recognition task consisting of three types of words: words
109
presented in the original list (e.g., candy), words not presented in the original list and not related to the theme of the list (e.g., hat), and “critical lure” words not presented in the original list but highly related to the them of the list (e.g., sweet). Payne et al. showed that experiencing an acute stressor followed by a brief waiting period dramatically increased false recognition of the critical lures. In particular, thematically related or gist-based false alarms (incorrect identification of critical lures as present in the original list, or “commission errors”) were amplified by the psychologically induced stressor. This increase in false alarms using the DRM paradigm has also been observed in chronically anxious participants (Lindsay & Jennings, 2000), in traumatized patients with and without the diagnosis of PTSD (Zoellner, Foa, Brigidi, & Przeworski, 2000), and in sexually abused women with PTSD (Bremner, Shobe, & Kihlstrom, 2000). Both Payne et al. (2002) and Lindsay and Jennings (2000) concluded that elevated stress hormones increase the production of false memories in the DRM paradigm. One must, however, ask “false memory for what”? Only on very rare occasions did stressed subjects falsely recognize unrelated distractor words, and their rate of false recognition of these distractors did not differ from that of nonstressed control subjects. Stressed subjects exhibited false memory of a particular type; they falsely remembered more words related to the thematic structure of the word list and therefore made more gist-based errors than their nonstressed counterparts. We suggest that this may be due to impaired contextual and detail memory and, as a result, an overreliance on gist memory (see Payne et al., 2002). These findings converge on the idea that stress disrupts memory for spatial and contextual details of experience, leading to an overreliance on thematic information.11 A closer look at the studies described earlier supports this idea. Recall that in Wolkowitz et al. (1990, 1993; see also Deptula, 1983), dexamethasone treatment was associated with significantly greater rates of commission errors, or intrusions (self-generated words) into free recall, than control (placebo) treatment. It was also associated with greater rates of false recognition of semantically related distractor words. There was, however, no difference between the treatment and control groups on correct free recall or total hits on the recognition task. The same pattern emerged for people given prednisone, though the tendency to false alarm to related distractors disappeared 7 days after discontinuation of treatment. Although not conceptualized as false memory experiments, the work of Deptula (1983) and Wolkowitz et al. (1990, 1993) found the same false remembering of related but nonpresented information as did Lindsay and Jennings (2000) and Payne et al. (2002). Excessive reliance on gist information implies some “filling in” of missing context and details and leads to errors logically consistent with the theme of an experience but nonetheless incorrect. Indeed, “innocent confabulation” and “narrative smoothing” during traumatic memory retrieval may stem from an overreliance on gist information. When people face decontextualized fragments, they may infer thematic material based
110
on gist to make sense of their experience. Thus, many memories formed under stress may in some sense be false, and many false memories may reflect gist-based narrative smoothing. Although the context and details of experience may be incorrect, the gist may well be accurate, an important consideration for the validity of memories for traumatic experience. The relation between emotional stress and memory for central or gist-like information clearly warrants further exploration. Nonetheless, the research reviewed here tempts one to conclude that focusing on the central or thematic features of an emotionally arousing event, perhaps at the expense of memory for context and detail, may be a characteristic reaction to stress.
Putting It All Together To summarize, stress, on the whole, appears to disrupt types of memory that depend on a functional hippocampus and PFC. Disruption of these structures is associated with impairment in episodic memory. Episodic memory, however, appears to be composed of different parts: memory for context (spatial memory), memory for events that occur within the specific context, and thematic (gist) information, derived by inference. Trauma appears to disrupt memory for the context (e.g., Jacobs & Nadel, 1985, 1998) and the details of experienced events (e.g., Bartlett, 1932/1995). The thematic content of the memory, on the other hand, appears to survive intact, perhaps because it is facilitated by an amygdalabased modulation of emotional aspects of experience. We suggest, speculatively, that this same pattern of impairment and sparing might characterize how memory is disrupted in PTSD. We have also pointed out that traumatic memories can apparently suffer two rather different fates: In one case, episodic information is disrupted while emotion is preserved; for example, emotion might be remembered in the form of intrusive fear and recurrent images of the trauma (flashbacks) or as persistent anxiety (free-floating anxiety). All episodic information associated with the trauma is not necessarily forgotten. Rather, the memory survives in a fragmented form that renders those fragments susceptible to narrative smoothing. Narrative smoothing does not reflect intentional fabrication but an automatic and inferential use of preserved emotional and gist information that fills in gaps and smooths over inconsistencies. In the second case, described in detail in the previous section, memory for traumatic events is preserved, but the associated emotion is cut off (isolation of affect). Based on the central idea that trauma results in abnormal functioning in various brain regions, with consequences for the memory roles these structures play, we have described patterns of disruption that account for the abnormal intrusion of traumatic memories into consciousness (i.e., flashbacks), the lack of recall for traumatic memory (amnesia), and the fragmentation and subsequent filling in of traumatic memories, depending on the specific circumstances. We
111
hope these ideas stimulate debate and experimentation and, with proper development, effective intervention.
Notes The writing of this chapter was supported by grants to L. N. from the Flinn Foundation and the McDonnell-Pew Cognitive Neuroscience Program. Address correspondence to any of the authors at: Psychology Department, University of Arizona, Tucson, AZ 89719 (
[email protected],
[email protected], or wjj@u. arizona.edu). 1. But what is trauma? Trauma is an event that often has a specified role in the etiology of a mental disorder. But what is a mental disorder? Building on the work of Klein (1978) and Wakefield (1992), we use the term “mental disorder” to designate an authentic dysregulation of adaptive emotional, cognitive, or behavioral systems designed through natural or sexual selection to solve specific adaptive problems. A “disorder” becomes noticeable when a dysregulated system becomes maladaptive (dysfunctional), that is, consistently fails to solve the adaptive problem for which it was designed. Accordingly, the presence of a dysregulated system (a mental disorder) may be inferred from measurable maladaptive (dysfunctional) characteristics at an emotional, cognitive, behavioral, physiological, or anatomical level. Note that this definition of mental disorder differs from that offered by Klein (1978) and Wakefield (1992) in two ways. First, the term “harmful” has been removed from the phrase “harmful dysfunction.” The term has been a source of both controversy and confusion (see McNally [2001] for a brief summary). Second, the notion of sexual selection has been added to natural selection as a “gold standard” against which one might determine ‘normality’ (see Miller [2000] for a delightful introduction to the notion of sexual selection and the scope of its influence). It is far beyond the reach of this chapter to discuss the implications of these moves. Suffice it to say that these distinctions provide theoretically useful predictions about distinctions between extant judgments of disorder and nondisorder. 2. The nature of traumatic events leading to the emergency room visits was not described in the Bonne et al. (2001) article. However, subjects were not included in the study if they suffered head trauma or physical injury that required hospitalization. 3. Although PTSD has been associated with persistently low levels of cortisol (hypocortisolemia) (Yehuda, 1997; Yehuda, Southwick, Nussbaum, Giller, & Mason, 1991), some cases, particularly in the early stages of the disorder, show elevated levels of stress hormones (e.g., Pitman & Orr, 1990; see Bremner, 2001b). 4. “Reversible” here implies that memory function eventually returns to baseline in acute stress studies. 5. These three types of memory are not mutually exclusive. For example, “context” can be a part of detail memory, helping to distinguish one episode from another, and “coherence” is a direct outcome of contextual information and details being combined properly to yield a veridical representation of experience. 6. It behooves the reader to distinguish between the effects of emotion on memories of various forms and the effects of traumatic stress on emotional memories. The
112
data we discuss here apply to the latter. In the section entitled “The Emotion and Memory Literature,” we discuss effects of emotion on memory (traditionally conceptualized) and provide a brief critique of the methods and interpretations offered by that literature. 7. This idea finds some support in the controversial evidence pointing to the existence of psychogenic amnesia, where trauma memories, or even memory for one’s entire identity, can be temporarily lost (Emilien et al., 2000; Markowitsch, 1999; but see Kihlstrom, 2001). 8. We do not know if these two reactions to trauma reflect different underlying brain functions, individual differences wherein some talk about traumatic events and others attempt to forget them, the level of stress experienced, the nature of the stressor (events that inspire guilt, shame, or severe physical threat may be less likely to be discussed), or some combination of these factors. 9. See Nadel et al. (2002) and Nadel and Payne (2002) for definitions of context and detail and an explanation of how gist-based memories result from a lack of both. 10. An important distinction in the emotional memory literature is that some of the material we remember is itself emotional, while other material is emotionally neutral but encountered while one is emotionally aroused by some other experience. In our view, the source of emotional arousal is not critical for the pattern of enhanced gist/impaired detail memory to emerge because elevated stress and HPA activity determine the pattern. Given sufficient elevation, regardless of its source, this HPA activity will differentially influence hippocampus-based and amygdala-based memory function, provided that the stress response occurs within a specific timeline. 11. This information on “gist” and “thematic information” brings to mind work on schemas and scripts—representations that describe knowledge people can abstract from common, frequently occurring events (e.g., Abelson, 1981; Mandler, 1984)—and recalls the classic studies of Bartlett (1932/1995), who demonstrated that schemas can influence retrieval. His classic “War of the Ghosts” study showed that people reconstruct their memories based on schemas as they try to fill in missing details and make sense out of incomprehensible information.
References Abelson, R. P. (1981). Psychological status of the script concept. American Psychologist, 36, 715–729. Adolphs, R., Denburg, N. L., & Tranel, D. (2001). The amygdala’s role in long-term declarative memory for gist and detail. Behavioral Neuroscience, 115, 983–992. Amaral, D. G., Price, J. L., Pitkanen, A., & Carmichael, S. T. (1992). Anatomical organization of the primate amygdaloid complex. In J. P. Aggleton (Ed.), The amygdala: Neurobiological aspects of emotion, memory and mental dysfunction (pp. 1–66). New York: Wiley-Liss. Anderson, S. W., Bechara, A., Damasio, H., Tranel, D., & Damasio, A. R. (1999). Impairment of social and moral behavior related to early damage in human prefrontal cortex. Nature Neuroscience, 2, 1032–1037. Arbel, I., Kadar, T., Silberman, M., & Levy, A. (1994). The effects of long-term corticosterone administration on hippocampal morphology and cognitive performance of middle-aged rats. Brain Research, 657, 227–235.
113
Arnsten, A. (1998). Catecholamine modulation of prefrontal cortical cognitive function. Trends in Cognitive Sciences, 2, 436–447. Arnsten, A. (2000). Stress impairs prefrontal cortical function in rats and monkeys: Role of dopamine D1 and norepinephrine alpha-1 receptor mechanisms. Progress in Brain Research, 126, 183–192. Arnsten, A. F. T. (2001). Dopaminergic and noradrenergic influences on cognitive functions mediated by prefrontal cortex. In M. V. Solanto, A. F. T. Arnsten, & F. X. Castellanos (Eds.), Stimulant drugs and ADHD: Basic and clinical neuroscience (pp. 185–208). London: Oxford University Press. Aubert, A., Vega, C., Dantzer, R., & Goodall, G. (1995). Pyrogens specifically disrupt the acquisition of a task involving cognitive processing in the rat. Brain, Behavior and Immunity, 9, 129–148. Baddeley, A. (1986). Working memory. Oxford: Clarendon. Bartlett, F. C. (1932/1995). Remembering: A study in experimental and social psychology. New York: Cambridge University Press. Bellinger, F. P., Madamba, S., & Siggins, G. R. (1993). Interleukin 1 b inhibits synaptic strength and long-term potentiation in the rat CA1 hippocampus. Brain Research, 628, 227–234. Bender, B. G., Lerner, J. A., & Poland, J. E. (1991). Association between corticosteroids and psychologic change in hospitalized asthmatic children. Annals of Allergy, 66, 414–419. Bodnoff, S. R., Humphreys, A. G., Lehman, J. C., Diamond, D. M., Rose, G. M., & Meaney, M. J. (1995). Enduring effects of chronic corticosterone treatment on spatial learning, synaptic plasticity, and hippocampal neuropathology in young and mid-aged rats. Journal of Neuroscience, 15, 61–69. Bohus, B., & Lissak, K. (1968). Adrenocortical hormones and avoidance behaviour of rats. International Journal of Neuropharmacology, 7, 301–306. Bonne, O., Brandes, D., Gilboa, A., Gomori, J. M., Shenton, M. E., Pitman, R. K., et al. (2001). Longitudinal MRI study of hippocampal volume in trauma survivors with PTSD. American Journal of Psychiatry, 158, 1248–1251. Brainerd, C. J., & Reyna, V. F. (1998). Fuzzy-trace theory and children’s false memories. Journal of Experimental Child Psychology, 71, 81–129. Bremner, J. D. (1999). Does stress damage the brain? Biological Psychiatry, 45, 797– 805. Bremner, J. D.. (2001a). A biological model for delayed recall of childhood abuse. In J. Freyd & A. Deprince (Eds.), Trauma and cognitive science: A meeting of minds, science, and human experience (pp. 165–183). The Haworth Maltreatment & Trauma Press. Bremner, J. D. (2001b). Hypotheses and controversies related to effects of stress on the hippocampus: An argument for stress-induced damage to the hippocampus in patients with posttraumatic stress disorder. Hippocampus, 11, 75–81. Bremner, J. D., Krystal, J. H., Charney, D. S., & Southwich, S. M. (1996). Neural mechanisms in dissociative amnesia for childhood abuse: Relevance to the current controversy surrounding the “False Memory Syndrome.” America Journal of Psychiatry, 153, FS71–82.
114
Bremner, J. D., Licinio, J., Darnell, A., Krystal, J. H., Owens, M., Southwick, S. M., et al. (1997). Elevated CSF corticotropin-releasing factor concentrations in posttraumatic stress disorder. American Journal of Psychiatry, 154, 624–629. Bremner, J. D., Randall, P. R., Capelli, S., Scott, T., McCarthy, G., & Charney, D. S. (1995). Deficits in short-term memory in adult survivors of childhood abuse. Psychiatry Research, 59, 97–107. Bremner, J. D., Randall, P. R., Scott, T. M., Bronen, R. A., Delaney, R. C., Seibyl, J. P., et al. (1995). MRI-based measurement of hippocampal volume in posttraumatic stress disorder. American Journal of Psychiatry, 152, 973–981. Bremner, J. D., Randall, P., Vermetten, E., Staib, L., Bronen, R. A., Mazure, C. M., et al. (1997). MRI-based measurement of hippocampal volume in posttraumatic stress disorder related to childhood physical and sexual abuse: A preliminary report. Biological Psychiatry, 41, 23–32. Bremner, J. D., Scott, T. M., Delaney, R. C., Southwick, S. M., Mason, J. W., Johnson, D. R., et al. (1993). Deficits in short-term memory in post-traumatic stress disorder. American Journal of Psychiatry, 150, 1015–1019. Bremner, J. D., Shobe, K. K., & Kihlstrom, J. F. (2000). False memories in women with self-reported childhood sexual abuse: An empirical study. Psychological Science, 11, 333–337. Bremner, J. D., Southwick, S. M., & Charney, D. S. (1997). Neuroanatomical correlates of the effects of stress on memory: Relevance to the validity of memories of childhood abuse. In P. S. Appelbaum, L. A. Uyehara, & M. R. Elin (Eds.), Trauma and memory (pp. 61–92). London: Oxford. Bremner, J. D., Southwick, S. M., & Charney, D. S. (1999). The neurobiology of PTSD: An integration of animal and human research. In J. D. Bremner & P. Saigh (Eds.), Posttraumatic stress disorder: A comprehensive text (pp 103–143). New York: Allyn and Bacon. Bremner, J. D. Southwick, S. M., Johnson, D. R., Yehuda, R., & Charney, D. S. (1993). Childhood physical abuse in combat-related posttraumatic stress disorder. American Journal of Psychiatry, 150, 235–239. Bremner, J. D., & Vermetten, E. (2001). Stress and development: Behavioral and biological consequences. Development and Psychopathology, 13, 473–489. Breslau, N., Davis, G. C., Andreski, P., & Peterson, E. (1991). Traumatic events and post-traumatic stress disorder in an urban population of young adults. Archives in General Psychiatry, 48, 216–222. Buchanan, T. W., & Lovallo, W. R. (2001). Enhanced memory for emotional material following stress-level cortisol treatment in humans. Psychoneuroendocrinology, 26, 307–317. Buckner, R. L., Logan, J., Donaldson, D. I., & Wheeler, M. E. (2000). Cognitive neuroscience of episodic memory encoding. Acta Psychologica, 105, 127–139. Burke, A., Heuer, F., & Reisberg, D. (1992). Remembering emotional events. Memory & Cognition, 20, 277–290. Cahill, L., Babinsky, R., Markowitsch, H. J., & McGaugh, J. L. (1995). The amygdala and emotional memory. Nature, 377, 295–296. Cahill, L., & McGaugh, J. L. (1998). Mechanisms of emotional arousal and lasting declarative memory. Trends in Neuroscience, 21, 294–299.
115
Cahill, L., Prins, B., Weber, M., & McGaugh, J. L. (1994). Beta adrenergic activation and memory for emotional events. Nature, 371, 702–704. Caldji, C., Francis, D., Sharma, S., Plotsky, P. M., & Meaney, M. J. (2000). The effects of early rearing environment on the development of GABA-A and central benzodiazepine receptor levels and novelty-induced fearfulness in the rat. Neuropsychopharmacology, 22, 219–229. Cameron, H. A., & Gould, E. (1994). Adult neurogenesis is regulated by adrenal steroids in the dentate gyrus. Neuroscience, 61, 203–209. Cameron, H. A., & McKay, R. (1999). Restoring production of hippocampal neurons in old age. Nature Neuroscience, 2, 894–897. Cannon, W. B. (1929). Bodily changes in pain, hunger, fear, and rage. 2nd ed. New York: Appleton. Christianson, S.-Å. (1992). Emotional stress and eyewitness memory: A critical review. Psychological Bulletin, 112, 284–309. Christianson, S.-Å., & Loftus, E. F. (1987). Memory for traumatic events. Applied Cognitive Psychology, 1, 225–239. Christianson, S.-ÅA, & Loftus, E. F. (1990). Some characteristics of people’s traumatic memories. Bulletin of the Psychonomic Society, 28, 195–198. Christianson, S.-Å., & Loftus, E. F. (1991). Remembering emotional events: The fate of detailed information. Cognition & Emotion, 5, 81–108. Christianson, S.-Å., Loftus, E. F., Hoffman, H., & Loftus, G. R. (1991). Eye fixations and memory for emotional events. Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 693–701. Christianson, S.-Å., & Safer, M. A. (1995). Emotional events and emotions in autobiographical memories. In D. C. Rubin (Ed.), Remembering our past: Studies in autobiographical memory (pp. 218–243). Cambridge: Cambridge University Press. Clark, L. D., Bauer, W., & Cobb, S. (1952). Preliminary observations on mental disturbances occurring in patients under therapy with cortisone and ACTH. New England Journal of Medicine, 246, 205–216. Clark, M. S., Milberg, S., & Ross, J. (1983). Arousal cues arousal-related materials in memory: Implications for understanding effects of mood on memory. Journal of Verbal Learning and Verbal Behavior, 22, 633–649. Clark, R. E., Zola, S. M., & Squire, L. R. (2000). Impaired recognition memory in rats after damage to the hippocampus. Journal of Neuroscience, 20, 8853–8860. Conrad, C. D., & Roy, E. J. (1996). Dentate gyrus destruction and spatial learning impairment after corticosteroid removal in young and middle-aged rats. Hippocampus, 5, 1–15. Convit, A., DeLeon, M. J., Tarshish, C., De Santi, S., Kluger, A., Rusinek, H., et al. (1995). Hippocampal volume losses in minimally impaired elderly. Lancet, 345, 266–266. Costello, E. J., Erkanli, A., Fairbank, J. A., & Angold, A. (2002). The prevalence of potentially traumatic events in childhood and adolescence. Journal of Traumatic Stress, 15, 99–112. Cunningham, A. J., Murray, C. A., O’Neill, L. A., & O’Connor, J .J. (1996). Interleukin 1-b (IL-1b) and tumor necrosis factor (TNF) inhibit long-term potentiation in the rat dentate gyrus in vitro. Neuroscience Letters, 203, 17–20.
116
Curran, T., Schacter, D. L., Norman, K. A., & Galluccio, L. (1997). False recognition after a right frontal lobe infarction: Memory for general and specific information. Neuropsychologia, 35, 1035–1049. Davidson, J. R. T., Roth, S., & Newman, E. (1991). Treatment of posttraumatic stress disorder with fluxetine. Journal of Traumatic Stress, 4, 419–423. Davis, M. (1992). The role of the amygdala in fear and anxiety. Annual Review of Neuroscience, 15, 353–375. Davis, M., & Shui, C. (1999). The extended amygdala: Are the central nucleus of the amygdala and the bed nucleus of the stria terminalis differentially involved in fear versus anxiety? In: J. McGinty (Ed.), Advancing from the ventral striatum to the extended amygdala: Implications for neuropsychiatry and drug use: In honor of Lennart Heimer (pp. 281–291). New York: New York Academy of Sciences. Delbecq-Derouesne, J., Beauvois, M. F., & Shallice, T. (1990). Preserved recall versus impaired recognition. Brain, 113, 1045–1074. de Quervain, D. J.-F., Roozendaal, B., Nitsch, R. M., McGaugh, J. L., & Hock, C. (2000). Acute cortisone administration impairs retrieval of long-term declarative memory in humans. Nature Neuroscience, 3, 313–314. Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal of Experimental Psychology, 58, 17–22. De Kloet, E. R., Oitzl, M. S., & Joels, M. (1999). Stress and cognition: Are corticosteroids good or bad guys? Trends in Neuroscience, 22, 422–426. Deptula, D. J. (1983). Lateralized brain dysfunction in depression: An analysis of memory. Dissertation Abstracts International. Diamond, D. M., Bennet, M. C., Fleshner, M., & Rose, G. M. (1992). Inverted U relationship between the level of peripheral corticosterone and the magnitude of hippocampal primed burst potentiation. Hippocampus, 2, 421–430. Diamond, D. M., Fleshner, M., Ingersoll, N. & Rose, G. M. (1996). Psychological stress impaires spatial working memory: Relevance to electrophysiological studies of hippocampal function. Behavioral Neuroscience, 110, 661–672. Diamond, D. M., Fleshner, M., & Rose, G. M. (1994). Psychological stress repeatedly blocks hippocampal primed burst potentiation in behaving rats. Behavioural Brain Research, 62, 1–9. Diamond, D. M., Park, C. R., Heman, K. L., & Rose, G. M. (1999). Exposing rats to a predator impairs spatial working memory in the radial arm water maze. Hippocampus, 9, 542–552. Diamond, D. M., & Rose, G. M. (1994). Stress impairs LTP and hippocampaldependent memory. Annals of New York Academy of Sciences, 746, 411–414. Emilien, G., Penasse, C., Charles, G., Martin, D., Lasseaux, L., & Waltregny, A. (2000). Post-traumatic stress disorder: Hypotheses from clinical neuropsychology and psychopharmacology research. International Journal of Psychiatry in Clinical Practice, 4, 3–18. Fairbank, J. A., Ebert, L., & Caddell, J. M. (2001). Posttraumatic stress disorder. In H. E. Adans & P. B. Sutker, (Eds.), (pp. 183–209). Comprehensive handbook of psychopathology (3rd ed., pp. 183–209). New York: Plenum. Fanselow, M. S., & LeDoux, J. E. (1999). Why we think plasticity underlying Pavlovian fear conditioning occurs in the basolateral amygdala. Neuron, 23, 229–232.
117
Feldman, S., & Conforti, N. (1980). Participation of the dorsal hippocampus in the glucocorticoid negative-feedback effect on adrenocortical activity. Neuroendocrinology, 30, 52–55. Feldman, S., & Conforti, N. (1985). Modifications of adrenocortical responses following frontal cortex stimulation in rats with hypothalamic differentiations and medial forebrain bundle lesions. Neuroscience, 15, 1045–1047. Filipini, D., Gijsbers, K., Birminghan, M. K., & Dubrovsky, B. (1991). Effects of adrenal steroids and their reduced metabolites on hippocampal long-term potentiation. Journal of Steroid Biochemistry and Molecular Biology, 40, 87–92. Fox, M. L., & Dwyer, D. J. (2000). Reducing stress by increasing control. Clinical Leadership and Management Reviews, 14, 114–117. Foy, M. R., Stanton, M. E., Levine, S., & Thompson, R. F. (1987). Behavioral stress impairs long-term potentiation in rodent hippocampus. Behavioral Neural Biology, 48, 138–149. Freud, S. (1926). Inhibitions, symptoms, and anxiety. London: Hogarth Press and the Institute of Psycho-analysis. Fukuzako, H., Fukuzako, T., Hashiguchi, T., Hokazono, Y., Takeuchi, K., Hirakawa, K., et al. (1996). Reduction in hippocampal formation volume is caused mainly by its shortening in chronic schizophrenia: Assessment by MRI. Biological Psychiatry, 39, 938–945. Garcia, R., Musleh, W., Tocco, G., Thompson, R. F., & Baudry, M. (1997). Timedependent blockade of STP and LTP in hippocampal slices following acute stress in mice. Neuroscience Letters, 233, 41–44. Gibertini, M., Newton, C., Friedman, H., & Klein, T. W. (1995). Spatial learning impairment in mice infected with Legionella pneumophila or administered exogenous interleukin-1–b. Brain, Behavior and Immunity, 9, 113–128. Gilbertson, M. W., Gurvits. T. V., Lasko, N. B., Orr, S. P., & Pitman, R. K. (2001). Multivariate assessment of explicit memory function in combat veterans with posttraumatic stress disorder. Journal of Traumatic Stress, 14, 413–432. Golier, J. A., Yehuda, R., & Southwick, S. M. (1997). Memory and posttraumatic stress disorder. In P. S. Appelbaum, L. A. Uyehara, & M. R. Elin (Eds.), Trauma and memory (pp. 225–242). London: Oxford. Goodman, G. G., Hirschman, J. E., Hepps, D. H., & Rudy, L. (1991). Children’s memory for stressful events. Merrill-Palmer Quarterly, 37, 109–158. Gould, E., & Gross, C. G. (2002). Neurogenesis in adult mammals: Some progress and problems. Journal of Neuroscience, 22, 619–623. Gould, E., & Tanapat, P. (1999). Stress and hippocampal neurogenesis. Biological Psychiatry, 46, 1472–1479. Gould, E., Tanapat, P., McEwen, B. S., Flugge, G., & Fuchs, E. (1998). Proliferation of granule cell precursors in the dentate gyrus of adult monkeys is diminished by stress. Proceedings of the National Academy of Science USA, 95, 3168–3171. Gray, J. A. (2000). The neuropsychology of anxiety: An enquiry into the function of the septo-hippocampal system. New York: Oxford. Gray, M. J., & Lombardo, T. W. (2001). Complexity of trauma narratives as an index of fragmented memory in PTSD: A critical analysis. [Special issue]. Applied Cognitive Psychology, 15, S171–S186.
118
Greidanus, T. J. B. (1970). Effects of steroids on extinction of an avoidance response in rats. A structure-activity relationship study. Progress in Brain Research, 32, 185–191. Gurvits, T. G., Shenton, M. E., Hokama, H., Ohta, H., Lasko, N.B., Gilbertson, M.W., et al. (1996). Magnetic resonance imaging study of hippocampal volume in chronic, combat-related posttraumatic stress disorder. Biological Psychiatry, 40, 1091–1099. Hamann, S. (2001). Cognitive and neural mechanisms of emotional memory. Trends in Cognitive Sciences, 5, 394–400. Hamann, S. B., Ely, T. D., Grafton, S. T., & Kilts, C. D. (1999). Amygdala activity related to enhanced memory for pleasant and aversive stimuli. Nature Neuroscience, 2, 289–293. Harvey, A. G., & Bryant, R. A. (1998). Relationship of acute stress disorder and posttraumatic stress disorder following motor vehicle accidents. Journal of Consulting and Clinical Psychology, 66, 507–512. Henson, R. N. A., Shallice, T., & Dolan, R. J. (1999). Right prefrontal cortex and episodic memory retrieval: A functional MRI test of the monitoring hypothesis. Brain, 122, 1367–1381. Heuer, F., & Reisberg, D. (1990). Vivid memories of emotional events: The accuracy of remembered minutiae. Memory & Cognition, 18, 496–506. Heuer, F., & Reisberg, D. (1992). Emotion, arousal, and memory for detail. In S. Christianson (Ed.), The handbook of emotion and memory: Research and theory (pp. 151–180). Hillsdale, NJ: Erlbaum. Hirsh, R. (1974). The hippocampus and contextual retrieval of information from memory: A theory. Behavioral Biology, 12, 421–444. Horner, M. D., & Hamner, M. B. (2002). Neurocognitive functioning in posttraumatic stress disorder. Neuropsychology Review, 12, 15–30. Inoue, T., Tsuchiya, K., & Koyama, T. (1994). Regional changes in dopamine and serotonin activation with various intensity of physical and psychological stress in the rat brain. Pharmacology, Biochemistry, & Behavior, 49, 911–920. Jacobs, W. J., & Nadel, L. (1985). Stress-induced recovery of fears and phobias. Psychological Review, 92, 512–531. Jacobs, W. J., & Nadel, L. (1998). Neurobiology of reconstructed memory. Psychology, Public Policy and Law, 4, 1110. Jacobs, W. J., Nadel, L., & Hayden, V. (1992). Anxiety disorders. In D. J. Stein & J. Young (Eds.), Cognitive science and clinical disorders. (pp. 211–233) New York: Academic Press. Jacobs, W. J., Laurance, H. E., Thomas, K. G. F., Luzcak, S. E., & Nadel, L. (1996). On the veracity and variability of recovered traumatic memory. Traumatology, 2. Available http://www.rdz.stjohns.edu: 80/trauma/traumaj.html Jacobson, L., & Sapolsky, R. M. (1991). The role of the hippocampus in feedback regulation of the hypothalamic-pituitary-adrenal axis. Endocrinology Reviews, 12, 118–134. Jenkins, M. A., Langlais, P. J., Delis, D., & Cohen, R. (1998). Learning and memory in rape victims with posttraumatic stress disorder. American Journal of Psychiatry, 155, 278.
119
Joseph, R. (1996). Neuropsychiatry, neuropsychology, and clinical neuroscience. Baltimore: Williams & Wilkins. Kaehler, S. T., Singewald, N., Sinner, C., Thurnher, C., & Philippu, A. (2000). Conditioned fear and inescapable shock modify the release of serotonin in the locus coeruleus. Brain Research, 859, 249–254. Kanaley, J. A., Weltman, J. Y., Pieper, K. S., Weltman, A., & Hartman, M. L. (2001). Cortisol and growth hormone responses to exercise at different times of day. Journal of Clinical Endocrinology and Metabolism, 86, 2881–2889. Kaspi, S. P., McNally, R. J., & Amir, N. (1995). Cognitive processing of emotional information in posttraumatic stress disorder. Cognitive Therapy and Research, 19, 319–330. Keenan, P. A., Jacobson, M. W., Soleymani, R. M., & Newcomer, J. W. (1995). Commonly used therapeutic doses of glucocorticoids impair explicit memory. Annals of the New York Academy of Sciences, 761, 400–402. Kessler, R. C., Sonnega, A., Bromet, E., Hughes, M., & Nelson, C. B. (1995). Posttraumatic stress disorder in the national comorbidity survey. Archives of General Psychiatry, 52, 1048–1060. Kihlstrom, J. F. (2001). Dissociative disorders. In P. B. Sutker & H. E. Adams (Eds.), Comprehensive handbook of psychopathology (3rd ed.). New York: Kluwer Academic/Plenum Publishers. Kim, J. J., & Diamond, D. M. (2002). The stressed hippocampus, synaptic plasticity and lost memories. Nature Reviews Neuroscience, 3, 453–462. Kirschbaum, C., Wolf, O. T., May, M., Wippich, W., & Hellhammer, D. H. (1996). Stress- and treatment-induced elevations of cortisol levels associated with impaired declarative memory in healthy adults. Life Sciences, 58, 1475–1483. Klein, D. F. (1978). A proposed definition of mental illness. In R. L. Spitzer & D. F. Klein(Eds.), Critical issues in psychiatric diagnosis (pp. 41–71). New York: Raven Press. Kleinsmith, L. J., & Kaplan, S. (1963). Paired-associate learning as a function of arousal and interpolated interval. Journal of Experimental Psychology, 65, 190–193. Kleinsmith, L. J., & Kaplan, S. (1964). The interaction of arousal and recall interval in nonsense syllable paired-associate learning. Journal of Experimental Psychology, 67, 124–126. Lang, P. J., & Cuthbert, B. N. (1984). Affective information processing and the assessment of anxiety. Journal of Behavioral Assessment, 6, 369–395. Laurance, H. E., Hardt, O. M., Nadel, L., & Jacobs, W. J. (2001). Stress affects spatiotemporal processing: Support for a model of traumatic memory. Poster session presented at the 109th meeting of the American Psychological Association. LeDoux, J. E. (1992). Emotion and the amygdala. In J. P. Aggleton (Ed.), The amygdala: Neurobiological aspects of emotion, memory, and mental dysfunction (pp. 339–351). New York: Wiley-Liss. LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience, 23, 155–184. Lemieux, A. M., & Coe, C. L. (1995). Abuse-related posttraumatic stress disorder: Evidence for chronic neuroendocrine activation in women. Psychosomatic Medicine, 57, 105–115.
120
Lindsay, A. L., & Jennings, J. M. (2000). Chronic anxiety and false memory: Potential implications for medial temporal lobe functioning. Poster session presented at the 2000 meeting of the Society for Neuroscience. Linthorst, A. C., Flachskamm, C., Holsboer, F., & Reul, J. M. (1994). Local administration of recombinant human interleukin-1b in the rat hippocampus increases serotonergic neurotransmission, hypothalamic-pituitary-adrenocortical axis activity and body temperature. Endocrinology, 135, 520–532. Litz, B. T., & Gray, M. J. (2002). Emotional numbing in posttraumatic stress disorder: Current and future research directions. Australian and New Zealand Journal of Psychiatry, 36, 198–204. Loftus, E. F., & Burns, T. (1982). Mental shock can produce retrograde amnesia. Memory & Cognition, 10, 318–323. Loftus, E. F., Loftus, G., & Messo, J. (1987). Some facts about “weapon focus.” Law and Human Behavior, 11, 55–62. Luine, V., Villegas, M., Martinez, C., & McEwen, B. S. (1994). Repeated stress causes reversible impairments of spatial memory performance. Brain Research, 639, 167–170. Lupien, S. J., de Leon, M., de Santi, S., Convit, A., Tarshish, C., Nair, N.P.V., et al. (1998). Cortisol levels during human aging predict hippocampal atrophy and memory deficits. Nature Neuroscience, 1, 69–73. Lupien, S. J., Gaudreau, S., Tchiteya, B. M., Maheu, F., Sharma, S., Nair, N. P. V., et al. (1997). Stress-induced declarative memory impairments in healthy elderly subjects: Relationship to cortisol reactivity. Journal of Clinical Endocrinology and Metabolism, 82, 2070–2075. Lupien, S. J., Lecours, A. R., Lussier, I., Schwartz, G., Nair, N. P. V., & Meaney, M. J. (1994). Basal cortisol levels and cognitive deficits in human aging. Journal of Neuroscience, 14, 2893–2903. Lupien, S. J., & Lepage, M. (2001). Stress, memory, and the hippocampus: Can’t live with it, can’t live without it. Behavioural Brain Research, 127, 137–158. Lupien, S. J., & McEwen, B. S. (1997). The acute effects of corticosteroids on cognition: Integration of animal and human model studies. Brain Research Reviews, 24, 1–27. Magarinos, A. M., & McEwen, B. S. (1995). Stress-induced atrophy of apical dendrites of hippocampal CA3c neurons: Involvement of glucocorticoid secretion and excitatory amino acid receptors. Neuroscience, 69, 89–98. Magarinos, A. M., McEwen, B. S., Flugge, G., & Fuchs, E. (1996). Chronic psychosocial stress causes apical dendritic atrophy of hippocampal CA3 pyramidal neurons in subordinate tree shrews. Journal of Neuroscience, 16, 3534– 3540. Magarinos, A. M., Verdugo, J. M., & McEwen, B. S. (1997). Chronic restraint stress alters synaptic terminal structure in hippocampus. Proceedings of the National Academy of Sciences, 94, 14002–14008. Maier, S. F., & Watkins, L. R. (1998). Stressor controllability, anxiety, and serotonin. Cognitive Therapy Research, 22, 595–613. Mandler, J. M. (1984). Stories, scripts, and scenes: Aspects of schema theory. Hillsdale, NJ: Erlbaum.
121
Markowitsch, H. J. (1999). Functional neuroimaging correlates of functional amnesia. Memory, 5/6, 561–583. Mauri, M., Sinforiani, E., Bono, G., Vignati, F., Berselli, M. E., Attanasio, R., et al. (1993). Memory impairment in Cushing’s disease. Acta Neurologica Scandinavica, 87, 52–55. McEwen, B. S. (2000). The neurobiology of stress: From serendipity to clinical relevance. Brain Research, 886, 172–189. McEwen, B. S. (2001). Neurobiology of interpreting and responding to stressful events: Paradigmatic role of the hippocampus. In B. S. McEwen (Ed.). Handbook of physiology. Section 7: The endocrine system. Volume 4: Coping with the environment: Neural and endocrine mechanisms. New York: Oxford University Press. McEwen, B. S., Albeck, D., Cameron, H., Chao, H. M., Gould, E., Hastings, N., et al. (1995b). Stress and the brain: A paradoxical role for adrenal steroids. In G. D. Litwack (Ed.), Vitamins and hormones (pp. 371–402). New York: Academic Press. McEwen, B. S., Conrad, C. D., Kuroda, Y., Frankfurt, M., Magarinos, A. M., & McKittrick, C. (1997). Prevention of stress-induced morphological and cognitive consequences. European Neuropsychopharmacology, 7(suppl.), 322–328. McEwen, B. S., & Sapolsky, R. M. (1995). Stress and cognitive function. Current Opinion in Neurobiology, 5, 205–216. McEwen, B. S., Weiss, J. M., & Schwartz, L. S. (1968). Selective retention of corticosterone by limbic structure in rat brain. Nature, 220, 911–912. McGaugh, J. L. (2000). Memory—A century of consolidation. Science, 287, 248–251. McGaugh, J. L. (2002). The amygdala regulates memory consolidation. In L. R. Squire & D. L. Schacter (Eds.), Neuropsychology of memory. 3rd ed. New York: Guilford Press. McGaugh, J. L., & Roozendaal, B. (2002). Role of adrenal stress hormones in forming lasting memories in the brain. Current Opinion in Neurobiology, 12, 205–210. McKittrick, C.nR., Magarinos, A. M., Blanchard, D. C., Blanchard, R. J., McEwen, B. S., & Sakai, R. R. (2000). Chronic social stress reduces dendritic arbors in CA3 of hippocampus and decreases binding to serotonin transporter sites. Synapse, 36, 85–94. McNally, R. J. (1998). Experimental approaches to cognitive abnormality in posttraumatic stress disorder. Clinical Psychology Review, 18, 971–982. McNally, R. J. (2001). On Wakefield’s harmful dysfunction analysis of mental disorder. Behaviour Research & Therapy 39, 309–314. McNally, R. J., Lasko, N. B., Macklin, M. L., & Pitman, R. K. (1995). Autobiographical memory disturbance in combat-related posttraumatic stress disorder. Behaviour Research and Therapy, 33, 619–630. McNally, R .J., Litz, B. T., Prassas, A., Shin, L. M., & West, F. W. (1994). Emotional priming of autobiographical memory in post-traumatic stress disorder. Cognition & Emotion, 8, 351–367. McNally, R. J., & Shin, L. M. (1995). Association of intelligence with severity of posttraumatic stress disorder symptoms in Vietnam combat veterans. American Journal of Psychiatry, 152, 936–938. Melo, B., Winocur, G., & Moscovitch, M. (1999). False recall and false recognition: An examination of the effects of selective and combined lesions to the medial
122
temporal lobe/diencephalon and frontal lobe structures. Cognitive Neuropsychology, 16, 343–359. Mesches, M. H., Fleshner, M., Heman, K. L., Rose, G. M., & Diamond, D. M. (1999). Exposing rats to a predator blocks primed burst potentiation in the hippocampus in vitro. Journal of Neuroscience, 19, RC18(1–5). Metcalfe, J., & Jacobs, W. J. (1996). An interactive hot system/cool system view of memory under stress. PTSD Research Quarterly, 7, 1–6. Metcalfe, J., & Jacobs W. J. (1998). Emotional memory: The effects of stress on ‘cool’ and ‘hot’ memory systems. In D. L. Medin (Ed.), The psychology of learning and motivation: Vol. 38. Advances in research and theory (pp. 187–222). San Diego, CA: Academic Press. Metcalfe, J., & Jacobs, W. J. (2000). Can ‘hot’ human emotions be captured by ‘cool’ mathematical models? In E. Tulving (Ed.), Memory, consciousness, and the brain: The Tallinn Conference (pp. 228–242). Philadelphia: Psychology Press. Miller, G. F. (2000). The mating mind: How sexual choice shaped the evolution of human nature. New York: Doubleday. Mitchell, K. J., Johnson, M. K., Raye, C. L., & D’Esposito, M. (2000). fMRI evidence of age-related hippocampal dysfunction in feature binding in working memory. Cognitive Brain Research, 10, 197–206. Mitchell, K. J., Johnson, M. K., Raye, C. L., Mather, M., & D’Esposito, M. (2000). Aging and reflective processes of working memory: Binding and test load deficits. Psychology and Aging, 15, 527–541. Mitchell, S. A., & Black, M. J. (1995). Freud and beyond: A history of modern psychoanalytic thought. New York: BasicBooks. Moghaddam, B., Bolinao, M. L., Stein-Behrens, B., & Sapolsky, R. (1994). Glucocorticoids mediate the stress-induced extracellular accumulation of glutamate. Brain Research, 655, 251–254. Nadel, L. (1991). The hippocampus and space revisited. Hippocampus, 1, 221– 229. Nadel, L. (1994). Multiple memory systems: What and why. An update. In D. L. Schacter & E. Tulving (Eds.), Memory systems 1994 (pp. 39–63). Cambridge, MA: MIT Press. Nadel, L., & Jacobs, W. J. (1996). The role of the hippocampus in PTSD, panic, and phobia. In N. Kato (Ed.), The hippocampus: Functions and clinical relevance (pp. 455– 463). Amsterdam: Elsevier Science. Nadel, L., & Jacobs, W. J. (1998). Traumatic memory is special. Current Directions in Psychological Science, 7, 154–157. Nadel, L., & Payne, J. D. (2002). The hippocampus, wayfinding, and episodic memory. In P. Sharp (Ed.), The neural basis of navigation: Evidence from single cell recording (pp. 235–247). Boston: Kluwer Academic Publishers. Nadel, L., Payne, J. D., & Jacobs, W. J. (2002). The relationship between episodic memory and context: Clues from memory errors made while under stress. Physiological Research, 51, S3–S11. Nadel, L., & Willner, J. (1980). Context and conditioning: A place for space. Physiological Psychology, 8, 218–228. Neisser, U. (1981). John Dean’s memory: A case study. Cognition, 9, 1–22.
123
Newcomer, J. W., Craft, S., Hershey, T., Askins, K., & Bardgett, M. E. (1994). Glucocorticoid-induced impairment in declarative memory performance in adult humans. Journal of Neuroscience, 14, 2047–2053. Newcomer, J. W., Selke, G., Melson, A. K., Hershey, T., Craft, S., Richards, K., et al. (1999). Decreased memory performance in healthy humans induced by stresslevel cortisol treatment. Archives of General Psychiatry, 56, 527–533. Nyberg, L., & Cabeza, R. (2000). Brain imaging of memory. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 501–519). Oxford: Oxford University Press. O’Carroll, R. E., Drysdale, E., Cahill, L., Shajahan, P., & Ebmeier, K. P. (1999). Stimulation of the noradrenergic system enhances and blockage reduces memory for emotional material in man. Psychological Medicine, 29, 1083–1088. Ochsner, K. N., & Schacter, D. L. (2000). A social cognitive neuroscience approach to emotion and memory. In J. C. Borod (Ed.), The neuropsychology of emotion (pp. 163–193). New York: Oxford University Press. Ohl, F., & Fuchs, E. (1999). Differential effects of chronic stress on memory processes in the tree shrew. Cognitive Brain Research, 7, 379–387. O’Keefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map. Oxford: Clarendon Press. Osborne, J. (1972). Short- and long-term memory as a function of individual differences in arousal. Perceptual and Motor Skills, 34, 587–593. Otten, L. J., & Rugg, M. D. (2002). The birth of memory. Trends in Neurosciences, 25, 279–281. Pani, L., Porcella, A., & Gessa, G. L. (2000). The role of stress in the pathophysiology of the dopaminergic system. Molecular Psychiatry, 5, 14–21. Parkin, A. J., Bindschaedler, C., Harsent, L., & Metzler, C. (1996). Pathological false alarm rates following damage to the left frontal cortex. Brain and Cognition, 32, 14–27. Payne, J. D., Nadel, L., Allen, J. J. B., Thomas, K. G. F., & Jacobs, W. J. (2002). The effects of experimentally-induced stress on false recognition. Memory, 10, 1–6. Pitman, R., & Orr, S. (1990). Twenty-four hour urinary cortisol and catecholamine excretion in combat-related posttraumatic stress disorder. Biological Psychiatry, 27, 245–247. Plihal, W., & Born, J. (1999). Memory consolidation in human sleep depends on inhibition of glucocorticoid release. Neuroreport, 10, 2741–2747. Porter, S., & Birt, A. (2001). Is traumatic memory special? A comparison of traumatic memory characteristics with memory for other emotional life experiences. Applied Cognitive psychology, 15, S101–S117. Prince, C. R., & Anisman, H. (1990). Situation specific effects of stressor controllability on plasma corticosterone changes in mice. Pharmacology, Biochemistry, and Behavior, 37, 613–621. Rasmusson, A. M., Lipschitz, D. S., Wang, S., Hu, S., Vojvoda, D., Bremner, J. D., et al. (2001). Increased pituitary and adrenal reactivity in premenopausal women with posttraumatic stress disorder. Biological Psychiatry, 50, 965–977.
124
Raye, C. L., Johnson, M. K., Mitchell, K. J., Nolde, S. F., & D’Esposito, M. (2000). fMRI investigations of left and right PFC contributions to episodic remembering. [Special issue]. Psychobiology, 28, 197–206. Reiman, E. M., Lane, R. D., Ahern, G. L., Schwartz, G. E., & Davidson, R. J. (2000). Positron emission tomography in the study of emotion, anxiety, and anxiety disorders. In R. D. Lane & L. Nadel (Eds.), Cognitive neuroscience of emotion (pp. 389–406). Oxford: Oxford University Press. Reyna, V. F., & Kiernan, B. (1994). Development of gist versus verbatim memory in sentence recognition: Effects of lexical familiarity, semantic content, encoding instructions, and retention interval. Developmental Psychology, 30, 178–191. Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory & Cognition, 21, 803–814. Roozendaal, B. (2000). Glucocorticoids and the regulation of memory consolidation. Psychoneuroendocrinology, 25, 213–238. Roozendaal, B., & McGaugh, J. L. (1996). Amygdaloid nuclei lesions differentially affect glucocorticoid-induced memory enhancement in an inhibitory avoidance task. Neurobiology of Learning and Memory, 65, 1–8. Roozendaal, B., & McGaugh, J. L. (1997). Glucocorticoid receptor agonist and antagonist administration into the basolateral but not central amygdala modulates memory storage. Neurobiology of Learning and Memory, 67, 176–179. Rosenbaum, R. S., Priselac, S., Kohler, S., Black, S. E., Gao, F, Nadel, L., et al. (2000). Remote spatial memory in an amnesic person with extensive bilateral hippocampal lesions. Nature Neuroscience, 3, 1044–1048. Rosenkranz, J. A., & Grace, A. A. (2002). Cellular mechanisms of infralimbic and prelimbic prefrontal cortical inhibition and dopaminergic modulation of basolateral amygdala neurons in vivo. Journal of Neuroscience, 22, 324– 337. Rugg, M. D., Fletcher, P. C., Chua, P. M.-L., & Dolan, R. J. (1999). The role of the prefrontal cortex in recognition memory and memory for source: An fMRI study. NeuroImage, 10, 520–529. Saigh, P. A., & Bremner, J. D. (1999). The history of posttraumatic stress disorder. In P. A. Saigh & J. D. Bremner (Eds.), Posttraumatic stress disorder: A comprehensive text. Needham Heights, MA: Allen & Bacon. Sapolsky, R. M. (1996). Why stress is bad for your brain. Science, 273, 749–750. Sapolsky, R. M. (2000). Glucocorticoids and hippocampal atrophy in neuropsychiatric disorders. Archives of General Psychiatry, 57, 925–935. Sapolsky, R. M., Krey, L. C., & McEwen, B. S. (1985). Prolonged glucocorticoid exposure reduces hippocampal neuron number: implications for aging. Journal of Neuroscience, 5, 1222–1227. Sapolsky, R. M., & Meaney, M. J. (1986). Maturation of the adrenocortical stress response: Neuroendocrine control mechanisms and the stress hyporesponsive period. Brain Research, 396, 64–76. Schacter, D. L., & Tulving, E. (Eds.), (1994). Memory systems 1994. Cambridge, MA: MIT Press.
125
Scoville, W. B., & Milner, B. (1957). Loss of recent memory after bilateral hippocampal lesions. Journal of Neurology, Neurosurgery, and Psychiatry, 20, 11–21. Selye, H. (1956). The stress of life. New York: McGraw-Hill. Sheline, Y. I., Sanghavi, M., Mintun, M. A., & Gado, M. H. (1999). Depression duration but not age predicts hippocampal volume loss in medically healthy women with recurrent major depression. Journal of Neuroscience, 19, 5034–5043. Sheline, Y. I., Wang, P. W., Gado, M. H., Csernansky, J. C., & Vannier, M. W (1996). Hippocampal atrophy in recurrent major depression. Proceedings from the National Academy of Sciences USA, 93, 3908–3913. Shimamura, A. P. (1997). Recollection: Perspectives on reinstated memory and child trauma. In J. D. Read & D. S. Lindsay (Eds.), Recollections of trauma (pp. 253–266). New York: Plenum Press. Shors, T. J., Gallegos, R. A., & Breindl, A. (1997). Transient and persistent consequences of acute stress on long-term potentiation (LTP), synaptic efficacy, theta rhythms and bursts in area CA1. Synapse, 26, 209–217. Southwick, S. M., Davis, M., Horner, B., Cahill, L., Morgan, C. A., Golc, P. E., et al. (2002). Relationship of enhanced norepinephrine activity during memory consolidation to enhanced long-term memory in humans. American Journal of Psychiatry, 159, 1420–1422. Stanny, C. J., & Johnson, T. C. (2000). Effects of stress induced by a simulated shooting on recall by police and citizen witnesses. American Journal of Psychology, 113, 359–386. Starkman, M. N., Gebarski, S. S., Berent, S., & Schteingart, D. E. (1992). Hippocampal formation volume, memory dysfunction, and cortisol levels in patients with Cushing’s syndrome. Biological Psychiatry, 32, 756–765. Starkman, M. N., Giordani, B, Gebrski, S. S., Berent, S., Schork, M. A., & Schteingart, D. E. (1999). Decrease in cortisol reverses human hippocampal atrophy following treatment of Cushing’s disease. Biological Psychiatry, 46, 1595–1602. Stein-Behrens, B. A., Elliott, E. M., Miller, C. A., Schilling, J. W., Newcombe, R., & Sapolsky, R.M. (1992). Glucocorticoids exacerbate kainic acid-induced extracellular accumulation of excitatory amino acids in the rat hippocampus. Journal of Neurochemistry, 58, 1730–1735. Stein-Behrens, B. A., Lin, W. J., & Sapolsky, R. M. (1994). Physiological elevations of glucocorticoids potentiate glutamate accumulation in the hippocampus. Journal of Neurochemistry, 63, 596–602. Stein, M. B., Koverola, C., Hanna, C., Torchia, M. G., & McClarty, B. (1997). Hippocampal volume in women victimized by childhood sexual abuse. Psychological Medicine, 27, 951–959. Stoddard, S. L., Bergdall, V. K., Townsend, D. W., & Levin, B. E. (1986). Plasma catecholamines associated with hypothalamically-elicited flight behavior. Physiology & Behavior, 37, 709–715. Tanapat, P., Hastings, N. B., Rydel, T. A., Galea, L. A., & Gould, E. (2001). Exposure to fox odor inhibits cell proliferation in the hippocampus of adult rats via an adrenal hormone-dependent mechanism. Journal of Comparative Neurology, 437, 496–504.
126
Terr, L. (1991). Childhood traumas: An outline and overview. American Journal of Psychiatry, 148, 10–20. Terr, L. (1993). Unchained memories. New York: Basic Books. Torrie, A. (1944). Psychosomatic casualties in the Middle East. Lancet, 29, 139– 143. Uddo, M., Vasterling, J. T., Brailey, K., & Sutker, P. B. (1993). Memory and attention in posttraumatic stress disorder. Journal of Psychopathology and Behavioral Assessment, 15, 43–52. Uno, H., Ross, T., Else, J., Suleman, M., & Sapolsky, R. (1989). Hippocampal damage associated with prolonged and fatal stress in primates. Journal of Neuroscience, 9, 1705–1711. Vaher, P., Luine, V., Gould, E., & McEwen, B. (1994). Effects of adrenalectomy on spatial memory performance and dentate gyrus morphology. Brain Research, 656, 71–76. van Cauter, E., & Turek, F. W. (2001). Roles of sleep-wake and dark-light cycles in the control of endocrine, metabolic, cardiovascular, and cognitive function. In B. S. McEwen & H. M. Goodman (Eds.), Handbook of physiology: Section 7, The endocrine system. New York: Oxford University Press. van der Kolk, B. A. (1991). The biological response to psychic trauma: Mechanisms and treatment of intrusion and numbing. Anxiety Research, 4, 199–212. van der Kolk, B. A. (1994). The body keeps the score: Memory and the evolving psychobiology of posttraumatic stress. Harvard Reviews Psychiatry, 1, 253–265. van der Kolk, B. A., & Fisler, R. (1995). Dissociation and the fragmentary nature of traumatic memories: Review and experimental confirmation. Journal of Traumatic Stress, 8, 505–525. van der Kolk, B. A., Hopper, J. W., & Osterman, J. E. (2001). Exploring the nature of traumatic memory: Combining clinical knowledge with laboratory methods. Journal of Aggression, Maltreatment & Trauma, 4, 9–31. Vasterling, J. J., Brailey, K., Allain, A. N., Duke, L. M., Constans, J. I., & Sutker, P. B. (2002). Attention, learning, and memory performances and intellectual resources in Vietnam veterans: PTSD and no disorder comparisons. Neuropsychology, 16, 5–14. Vasterling, J .J., Brailey, K., Constans, J. I., Borges, A., & Sutker, P. B. (1997). Assessment of intellectual resources in Gulf War veterans: Relationship to PTSD. Assessment, 4, 51–59. Vasterling, J. J., Brailey, K., Constans, J. I., & Sutker, P. B. (1998). Attention and memory dysfunction in posttraumatic stress disorder. Neuropsychology, 12, 125–133. Virgin, C. E., Taryn, P. T. H., Packan, D. R., Tombaugh, G. C., Yang, S. H., Horner, H. C., et al. (1991). Glucocorticoids inhibit glucose transport and glutamate uptake in hippocampal astrocytes: Implications for glucocorticoid neurotoxicity. Journal of Neurochemistry, 57, 1422–1428. Vyas, A., Mitra, R., Rao, S. B. S., & Chatterji, S. (2002). Chronic stress induces contrasting patterns of dendritic remodeling in hippocampal and amygdaloid neurons. Journal of Neuroscience, 22, 6810–6818.
127
Wakefield, J. C. (1992). Disorder as harmful dysfunction: A conceptual critique of DSM-III-R’s definition of mental disorder. Psychological Review, 99, 232–247. Watanabe, Y., Gould, E., Daniels, D. C., Cameron, H., & McEwen, B. S. (1992). Tianetine attenuates stress-induced morphological changes in the hippocampus. European Journal of Pharmacology, 222, 157–162. Watanabe, Y., Gould, E., & McEwen, B. S. (1992). Stress induces atrophy of apical dendrites of hippocampal CA3 pyramidal neurons. Brain Research, 588, 341– 345. Wechsler, D. (1945). A standardized memory scale for clinical use. Journal of Psychology, 19, 87–95. Wolkowitz, O. M., Reus, V. I., & Weingartner, H. (1990). Cognitive effects of corticosteroids. American Journal of Psychiatry, 147, 1297–1303. Wolkowitz, O. M., Weingartner, H., Rubinow, D. R., Jimerson, D., Kling, M., Berretini, W., et al. (1993). Steroid modulation of human memory: Biochemical correlates. Biological Psychiatry, 33, 744–746. Wooley, C. S., Gould, E., & McEwen, B. S. (1990). Exposure to excess glucocorticoids alters dendritic morphology of adult hippocampal pyramidal neurons. Brain Research, 531, 225–231. Yehuda, R. (1997). Stress and glucocorticoid. Science, 275, 1662–1663. Yehuda, R., Keefe, R. S. E., Harvey, P. D., Levengood, R. A., Gerber, D. K., Geni, J., et al. (1995). Learning and memory in combat veterans with posttraumatic stress disorder. American Journal of Psychiatry, 152, 137–139. Yehuda, R., Southwick, S. M., Nussbaum, E. L., Giller, E. L., & Mason, J. W. (1991). Low urinary cortisol in PTSD. Journal of Nervous and Mental Disorders, 178, 366–369. Zoellner, L. A., Foa, E. B., Brigidi, B. D., & Przeworski, A. (2000). Are trauma victims susceptible to “false memories”? Journal of Abnormal Psychology, 109, 517– 524.
128
4
. , . ,
.
M
ost experts agree that people exposed to traumatic events remember them all too well. (For reviews, see McNally, 2003c; Pope, Oliva, & Hudson, 1999.) Indeed, survivors of trauma often report intrusive memories despite their efforts to forget. But debate persists regarding whether a substantial subset of survivors repress, dissociate, or otherwise forget having been traumatized—and then remember it all later (McNally, 2003a). Consider the views of Brown, Scheflin, and Hammond (1998), who proclaim “overwhelming scientific support for the existence of repressed or dissociated memory” (pp. 538–539). Summarizing across studies, they conclude that “generally speaking, approximately a third of sexually abused victims report some period of their lives where they did not remember anything about the abuse and later recovered the memory of the abuse” (p. 196). Yet after considering much of the same evidence, Loftus and Ketcham (1994) published their different conclusions in The Myth of Repressed Memory. How can scholars scrutinize the same evidence yet arrive at such diametrically opposed conclusions? To address this question, we examine the evidence adduced in support of repressed and recovered memory for traumatic experiences. To provide a fresh perspective on the debate, we also summarize our recent laboratory research on cognitive functioning in people who report repressed, recovered, or continuous memories of trauma.
Common Assumptions About
Forgetting Trauma
Some theorists distinguish between repression and dissociation, defining the former as inhibition of taboo urges and the latter as inhibition of disturbing 129
memories. But in the recovered memory debate, this is a distinction without a difference. In fact, as Brown et al. indicate, “repressed memory” and “dissociated memory” are used interchangeably. Regardless of whether the theorist’s favored mechanism is repression or dissociation, he or she usually endorses the following assumptions about forgetting and trauma. First, people are motivated to forget unpleasant experiences. Second, they are especially likely to forget atrocious trauma: “The ordinary response to atrocities is to banish them from consciousness” (Herman, 1992, p. 1). Third, because sexual abuse, in particular, is so traumatic, if someone fails to think about the abuse for many years, then he or she must have repressed or dissociated it from awareness. That is, an active inhibitory force must be keeping it out of consciousness. Otherwise, why would someone not think about it for years? Fourth, the more frequently a person (especially a child) is traumatized, the more difficult it will be for him or her to remember having been traumatized (Terr, 1991). That is, people exposed to repetitive trauma develop dissociative skills to cope with inescapable situations, making it difficult to recall these experiences many years later. Fifth, children are more likely to forget having been traumatized if they were abused by their parents than if they were abused by strangers (Freyd, 1996). Sixth, forgotten trauma is neither inert nor benign; it is the silent source of diverse psychological problems (Blume, 1990). Accordingly, remembering forgotten trauma, emotionally processing it, and integrating it into one’s autobiography are important steps toward healing (Courtois, 1992). All assumptions except the first have been flashpoints for intense controversy.
Evidence Adduced for
Forgetting of Trauma
The trauma therapists Brown et al. (1998) have assembled evidence in support of the six assumptions. Many of the studies they cite were modeled on an influential survey conducted by Briere and Conte (1993). Recruiting subjects through a network of therapists specializing in the treatment of sexual abuse, these authors obtained questionnaire data on 450 patients who reported having been sexually abused as children. In response to the question, “Was there ever a time when you could not remember the forced sexual experience,” 59% of the subjects responded in the affirmative. Many psychologists, such as Brown et al., interpreted these findings as evidence that many sexual abuse survivors experience amnesia for their traumatic experiences, only to remember them later in life. Scholars soon drew attention to the methodological limitations of this survey and others like it. First, the subjects were patients potentially exposed to therapeutic techniques likely to foster illusory memories of abuse (Poole, Lindsay, Memon, & Bull, 1995). As in many abuse recollections, external corrobo-
130
ration was apparently unavailable, and the duration of “amnesia” for trauma was unspecified (a week? decades?). The most important issue concerned the key survey question about forgetting. Interpreted literally, this question seems to make little sense. An affirmative response implies that the subject had repeatedly attempted to remember the abuse but failed to do so (i.e., could not remember). But if subjects were unaware of having been abused, on what basis would they attempt to recall it in the first place? Indeed, the most sensible way to regard affirmative responses to this question is to assume that subjects interpreted it to mean, “Was there ever a time when you did not think about your abuse?” A person who answered “yes” to this question might have experienced sexual abuse as a child, managed not to think about it for many years, but was reminded of the abuse in adulthood. But the long time when the person did not think about his or her abuse cannot be equated with “amnesia,” or an inability to recall something from memory when furnished with adequate retrieval cues. Just because a person did not remember (did not think about it) does not mean that the person could not remember. Moreover, as Schooler and others have shown, some people who believe they have not thought about their (documented) traumatic experiences for many years are surprised to learn that they had, in fact, discussed these events with family members when they thought the memories never came to mind (Schooler, Bendiksen, & Ambadar, 1997). That is, a person can forget having remembered (thought about) a traumatic event, producing an “illusion of amnesia.” A person might be especially likely to forget a previous instance of recalling abuse if the recollection was not accompanied by strong emotion. For example, in a conversation with her husband, a woman might mention that she was once molested by a distant relative. If this conversation occurs during an especially positive period in her life, she might not experience much emotion while recounting her abuse and therefore not recall the conversation with her husband. A key point in examining the literature on forgetting trauma is that one cannot equate amnesia for an event with merely not having thought about the event for a time. To confirm amnesia, one must show that the person had encoded the event in the first place and is now incapable of recalling it despite the adequate retrieval cues (e.g., such as being asked whether one had ever been abused). Failure to think about something for a time must not be confused with an inability to remember it. Moreover, as we have learned in our ongoing studies, many individuals report that the abuse was not experienced as traumatic when it occurred. Many of our subjects have told us that they felt it must be “wrong”—that something was not “right” about it. At the time, they often felt confused or embarrassed by the experience. Only years later did they grasp the full significance of what had happened (i.e., that they were sexually abused). In fact, many of them rate the abuse as more traumatic now than it was at the time.
131
These points also apply to the growing number of case reports concerning people who recall long-forgotten—and often corroborated—episodes of abuse (for reviews, see Cheit, 1998, 1999; McNally, 2003c, chapter 7). Some children are exposed to one (or sometimes more) episodes of molestation, often nonpenetrative (e.g., fondling) but abusive nevertheless. They find the experiences confusing or upsetting but not necessarily terrifying. They avoid dwelling on the experience and may succeed in forgetting it, especially if reminders are no longer present (e.g., the perpetrator moves away). Then, many years later, they encounter retrieval cues (e.g., someone mentions the perpetrator, or they hear news reports on television regarding sexual abuse), and the memory suddenly pops back into awareness. Such cases clearly qualify as “recovered memories of sexual abuse,” but they do not count as amnesia (i.e., an inability to recall when provided adequate retrieval cues). In fact, sudden recollections of seemingly forgotten experiences are fairly common in the general population and are not confined to adverse events like sexual abuse (Read, 1997). Most important, a failure to think about something for many years does not mean that active forces (repression, dissociation) are preventing these memories from entering awareness. An important study by Williams (1994) further illustrates these interpretive pitfalls. Her research team interviewed 129 women who had been assessed for suspected sexual abuse approximately 17 years earlier, when they were children. Medical evidence confirmed the abuse in many cases. The main purpose of the study was to survey women about their health and experiences with the health care system. Embedded within the interview were questions about sexual abuse. Women who acknowledged having been abused were questioned about the event. Their narratives were compared with the hospital records that described the index event. Of the 129 women, 49 of them did not mention the index event (although 33 of them did describe other incidents of sexual abuse). Strikingly, 16 of the 49 women denied ever having been sexually abused. Some authors have interpreted these data as showing that victims can repress or dissociate all memory of their abuse rather than merely forget it (i.e., not think about it for a long time), but other interpretations are possible. Many of the subjects were younger than 5 when assessed for abuse. Because few memories from very early childhood survive into adulthood, memories of the index event may have undergone ordinary forgetting. Moreover, the younger the child was when abused, the more likely she might have failed to understand what the perpetrator was doing, thereby undermining encoding and later retrieval of the memory. Other subjects may have remembered the index event but may have been too embarrassed to disclose such personal matters to the interviewer. A study by Goodman et al. (2003) provides further data on failure to disclose abuse. They assessed 175 subjects (81% female) who had been involved in legal proceedings concerning sexual abuse. The proceedings occurred when the subjects were approximately 9 years old (range: 3 years to 16 years). Questions about sexual abuse were embedded in a longer survey concerning attitudes about the
132
law and experiences with the legal system. The survey was administered 13 years after the subjects had been involved in the legal proceedings. The results revealed that 88% of the subjects mentioned the index event that led to the legal proceedings. Like some subjects in Williams’s study, 17 subjects denied ever having been abused. However, further analysis of the data undermines many popular explanations for why these nondisclosing individuals may have “forgotten” their abuse. Contrary to the notion that people prone to dissociation are most likely to forget their abuse, the higher a subject’s score on the Dissociative Experiences Scale (DES; Bernstein & Putnam, 1986), the more likely the subject was to disclose the abuse. Contrary to Freyd’s (1996) conjecture that children exposed to parental abuse are those most likely to be unable to remember it, there was no correlation between failure to disclose and relationship to the perpetrator. The studies by Williams (1994) and Goodman et al. (2003) do not provide clear support for any special repression or dissociation mechanism that actively inhibits recollection of the index event. The findings from these studies can be most parsimoniously interpreted as showing that a minority of adults with childhood abuse histories will deny having been abused when questioned by a survey interviewer. However, in neither study did researchers conduct subsequent clarification interviews to ascertain the basis for the discrepancy between documents concerning the abuse and the subject’s denial of it. In a study on young adults with documented histories of physical abuse, Femina, Yeager, and Lewis (1990) found that all subjects who had denied (or minimized) their abuse during an interview later acknowledged that they had not forgotten it when originally questioned by the interviewer. Subjects who had earlier denied their documented abuse later said that they did not wish to discuss such upsetting experiences or did not like the interviewer. None had repressed or forgotten the abuse. The findings of this study suggest that some of the nondisclosers in the studies of Williams and Goodman et al. may have been disinclined to discuss abuse they had, in fact, remembered. In any event, survey studies have not been the only source of evidence adduced in support of repressed memories of trauma (Brown et al., 1998; van der Kolk & Fisler, 1995). For example, Brown et al. wrote that “Dollinger (1985) found that two of the 38 children studied after watching lightning strike and kill a playmate had no memory of the event” (pp. 609–610). Although this trauma was well documented, the amnesia experienced by these two children for this horror has nothing to do with dissociation or repression. Brown et al. forgot to mention that both children had themselves been struck by side flashes from the main lightning bolt, had been knocked unconscious, and had nearly died (Dollinger, 1985). Given the severe effects on the central nervous system of a lightning strike (Kotagal, Rawlings, Chen, Burris, & Nouri, 1982), it is little wonder that these two children had amnesia for the episode. Obviously, their amnesia was entirely organic. Indeed, none of the other children who had not been struck by lightning forgot this horrific event.
133
Interpreting research on memory for trauma among former inmates of concentration Camp Erika (Wagenaar & Groeneweg, 1990), Brown et al. (1998) claimed that “amnesia for Nazi Holocaust camp experiences has also been reported” (p. 156). Wagenaar and Groeneweg assessed the memories of camp survivors 40 years after the survivors had provided testimony about their internment shortly after their release. Impressed by the strikingly accurate recollections of the former inmates, Wagenaar and Groeneweg emphasized that “there is no doubt that almost all witnesses remember Camp Erika in great detail, even after 40 years” (p. 84). Why, then, did Brown et al. cite this study as providing evidence for traumatic amnesia? As it turns out, several subjects claimed to have forgotten the name of an especially brutal guard or failed to mention several violent incidents that they had mentioned 40 years earlier. But even these isolated incidents of forgetting do not indicate amnesia. With one exception, subjects immediately recollected the forgotten details once they examined their original testimony. Given the “remarkable degree of remembering” (Wagenaar & Groeneweg, 1990, p. 80) exhibited by the former inmates, Brown et al.’s claim about amnesia regarding the Holocaust seems far-fetched. Because memory does not operate like a video recorder, it is hardly surprising that not every detail mentioned was instantly recalled 40 years later. [See chapter 11—Eds.] One of the most common blunders made by theorists of traumatic amnesia is to confuse general memory impairment in the wake of trauma with an inability to remember that one has been traumatized. Listing studies that supposedly support the notion of traumatic amnesia, Brown et al. (1998) mention Wilkinson’s (1983) study of individuals who had witnessed the horrific collapse of the Hyatt Regency skywalks in Kansas City. Of those assessed, 27% endorsed the DSM-III symptom of “memory difficulties” while affirming their intensely vivid recollections of the disaster itself. Indeed, intrusive memories of the disaster may have contributed to problems with forgetfulness in everyday life. In another often misinterpreted study (Brown et al., 1998; van der Kolk & Fisler, 1995), Archibald and Tuddenham (1965) reported that 65% of their World War II combat fatigue cases complained of “difficulty in memory” versus only 5% of psychiatrically healthy combat veterans. However, 60% of veterans who were not in combat and suffered from other psychiatric disorders endorsed the same complaint. Moreover, the authors themselves emphasized how unforgettable combat was for the traumatized patients, noting that these veterans “cannot blot out their painful memories” (p. 480). Obviously, the “memory problems” reported by these men concern everyday forgetfulness, not amnesia for their combat trauma. However, other studies on combat veterans have uncovered evidence of apparent amnesia for combat itself among groups of men referred for psychiatric evaluation and treatment. Eder (1917) reported that 2% of his psychiatric battle
134
casualties could not recall certain combat events occurring in World War I. According to Sargant and Slater (1941), 14% of psychiatric casualties who had escaped from Dunkirk reported amnesia for events occurring during the chaotic retreat. World War II psychiatrists reported rates of amnesia for combat in 5% of patients in the Pacific Theater (Henderson & Moore, 1944) and 8.6% in North Africa (Torrie, 1944). But as Pope et al. (1999) pointed out, these psychiatrists were often unable to rule out organic causes of amnesia (e.g., exhaustion, head injury) or malingering. Other more recent cases of amnesia for traumatic events provide an interesting contrast with the typical case of repressed and recovered memory of childhood sexual abuse. Swihart, Yuille, and Porter (1999) have described cases of “red-out” in which a person murders a loved one in a fit of rage but experiences amnesia for the murder itself. Swihart et al. suspect that these cases may be genuine, not malingered. Indeed, the murderer often phones the police and immediately admits to committing the crime despite claiming not to have remembered the actual attack. There have been other cases of psychogenic amnesia in which an event, often shocking, triggers retrograde amnesia (see McNally, 2003c, chapter 7, for a review of this case study literature). Reviewing this literature, Arrigo and Pezdek (1997) connected it to the furor over recovered memories of sexual abuse. However, the phenomenon of psychogenic amnesia is drastically different from the controversial cases in the child abuse field. In cases of psychogenic amnesia, say, after the death of a loved one, the person develops complete retrograde amnesia shortly after the shocking event, accompanied by loss of personal identity. Psychogenic amnesia seldom lasts for more than a few hours to a few weeks. Typically, memory and personal identity return abruptly without any psychotherapy. In contrast, cases of traumatic amnesia, say, of repressed and recovered memories of childhood sexual abuse, have no clear onset, never involve identity loss, may last for years or decades, and often gradually return piece by piece during the course of psychotherapy. Moreover, the amnesia selectively blocks retrieval of trauma; it does not blot out the entire person’s life and identity. Accordingly, classic psychogenic amnesia has little to do with the controversy over repressed and recovered memories of childhood abuse. Finally, some theorists note that “inability to recall an important aspect of the trauma (psychogenic amnesia)” has been a diagnostic criterion of PTSD ever since DSM-III-R (American Psychiatric Association [APA], 1987, p. 250) and adduce this fact to support the notion that trauma survivors can repress or dissociate their memory for trauma. Unfortunately, this diagnostic criterion is fatally ambiguous because it does not distinguish between an inability to recall an aspect of the trauma that was never encoded into memory in the first place and an inability to recall an aspect of the trauma because an inhibitory mechanism (e.g., repression) blocks its access to awareness. Because the mind is not a
135
video recorder, there will inevitably be aspects of the traumatic event that never make it into memory. Encoding failure must not be confused with amnesia, let alone repression. In summary, extant data do not support the claim that unconscious repression or dissociation mechanisms expel traumatic memories from awareness and prevent them from entering consciousness. People often try not to think about their sexual abuse or other unpleasant experiences, and sometimes they succeed. But not thinking about something for a long time does not imply an inability to remember it (i.e., amnesia).
A View From the Cognitive
Psychology Laboratory
Cognitive psychologists have conducted considerable research on mechanisms of inhibition in memory retrieval (e.g., Anderson & Neely, 1996). And they have drawn on basic research to speculate about how false memories of trauma might be inadvertently fostered in clinical settings (e.g., Ceci & Loftus, 1994; Lindsay & Read, 1994). But this work is swiftly dismissed by repressed memory theorists, who assert that generalizations from laboratory studies with college students have questionable relevance for how abuse survivors might remember or forget their trauma (e.g., Brown et al., 1998, p. 98). In fact, one of the most striking features of the recovered memory debate has been the near absence of any research on cognitive functioning in people reporting repressed or recovered memories of abuse. This situation may have arisen because few clinicians possess expertise in laboratory research, and few cognitive psychologists have access to trauma populations. In any event, during the past few years our research group has been conducting laboratory studies designed to test hypotheses relevant to (1) mechanisms implicated in either the ability to repress and recover traumatic memories or (2) mechanisms relevant to proneness to forming false memories of trauma (McNally, 2001, 2003b). This line of research began quite by accident. The first author was conducting psychiatric diagnostic interviews for a positron emission tomography (PET) study designed to map the functional neuroanatomy of autobiographical recollection of traumatic memories in women who had been sexually abused as children (Shin et al., 1999). We had posted advertisements seeking female adult survivors of childhood sexual abuse, and we planned to distinguish those with and without PTSD prior to the study. During the diagnostic interview, three women who expressed interest in the project turned out not to have any explicit autobiographical memories of molestation, despite considering themselves survivors of abuse. Further questioning revealed that they assumed that they harbored repressed memories of abuse because they were inexplicably tense in the presence of certain relatives, experienced diverse symptoms (e.g., depressed mood), and so forth; lacking a suit-
136
able explanation, they assumed that their problems must be attributable to repressed memories of early abuse. Because they lacked autobiographical memories of trauma, these women did not qualify for the neuroimaging study. But their reports of repressed memories inspired us to recruit and study individuals who reported diverse memory manifestations of trauma. In our first series of studies, we recruited four groups of women from the community. The repressed memory group included subjects who suspected they had been sexually abused as children but had no autobiographical memories of abuse. These subjects inferred their abuse history from a diverse range of symptoms (e.g., nightmares, depressed mood, bulimia). About a third of this group mentioned having acquired the belief in their repressed memories of abuse during psychotherapy. The recovered memory group included subjects who reported recollecting abuse after long periods of not having thought about it. Nearly half of them remembered their abuse during the course of psychotherapy, but only one mentioned recovering a memory in session. The continuous memory group reported never having forgotten their abuse. The control group included individuals who denied ever having been sexually abused. We were unable to corroborate reports of the abuse (although we are actively attempting to do so in our current set of studies). Accordingly, we do not know whether the recovered (or continuous) memories of abuse are genuine or whether those who suspected they harbored repressed memories of abuse were really abused. Of course, the absence of corroboration does not mean that the memories are false. To characterize our groups, we conducted a study on personality profiles and clinical symptoms (McNally, Clancy, Schacter, & Pitman, 2000b). Our subjects completed the Multidimensional Personality Questionnaire (MPQ; Tellegen, 1982), an inventory designed to characterize normal personality variation that includes an Absorption scale (Tellegen & Atkinson, 1974) that taps fantasy proneness. In addition to these personality measures, they also completed three psychiatric questionnaires. The Beck Depression Inventory (BDI; Beck & Steer, 1987) measured symptoms of depression, the Dissociative Experiences Scale (DES; Bernstein & Putnam, 1986) measured alterations in consciousness (e.g., memory lapses, depersonalization, episodes of “spacing out”), and the civilian version (Vreven, Gudanowski, King, & King, 1995) of the Mississippi Scale for Combat-Related Posttraumatic Stress Disorder (CMISS; Keane, Caddell, & Taylor, 1988) measured PTSD symptoms, as well as other problems (e.g., occupational difficulties, suicidal ideation) sometimes linked to PTSD. The MPQ uncovered striking similarities and differences in personality profile among the groups. The continuous memory group and the control group were indistinguishable on every measure of personality. The same was true for the repressed and recovered memory groups, who did not differ on any scale. Moreover, the four groups did not differ on positive affectivity, or proneness to experience joy and enthusiasm. In contrast, the groups did differ in negative affectivity, or proneness to experience sadness, anxiety, anger, and guilt. Re-
137
pressed memory subjects scored higher than did either the continuous memory or control groups, whereas recovered memory subjects scored midway between the repressed subjects and subjects in the other two groups. Consistent with patterns of negative affectivity, repressed memory subjects had more depressive, dissociative, and PTSD symptoms than did continuous memory and control subjects. Repressed memory subjects also had more depressive and PTSD symptoms than recovered memory subjects, who, in turn, had more dissociative and PTSD symptoms than did control subjects. Strikingly, continuous memory subjects were indistinguishable from control subjects in PTSD, dissociative, and depressive symptoms. Finally, the repressed and recovered memory groups did not differ on the absorption scale, but both groups scored higher than the control group. The repressed memory group also scored significantly higher than the continuous memory group. Table 4.1 summarizes these findings. These data indicate that people who believe they harbor repressed memories of sexual abuse are more psychologically distressed than those who have never forgotten their abuse. There are at least two explanations for this pattern. Heightened levels of distress among repressed memory subjects may reflect the psychic toll of blocking out memories of abuse. On the other hand, their distress may have arisen from diverse and poorly understood causes, motivating an “effort after meaning” that led them to attribute their problems to repressed memories of abuse. As Bass and Davis (1988) emphasized in The Courage to Heal, “When you first remember your abuse or acknowledge its effects, you may feel tremendous relief. Finally there is a reason for your problems. There is someone, and something to blame” (p. 173). When confronted with a choice between being miserable and not knowing why and being miserable and knowing why, many people will choose the latter, especially if they can also blame others for their difficulties. To test Bass and Davis’s (1988) hypothesis, we recontacted 11 of our recovered memory subjects and asked them to complete a Child Abuse Survivor Questionnaire (Clancy, 2000). In this pilot study, subjects wrote brief responses to
table 4.1 Summary of Psychometric Results Variable Positive affectivity Negative affectivity Depression symptoms PTSD symptoms Dissociative symptoms Absorption
Repress a a a a a a
Group Recover Continuous a ab b b ab ab
a b b bc bc bc
Control a b b c c c
Note. Repress = repressed memory group; recover = recovered memory group. Means of groups sharing a letter do not differ significantly (p > .05). A mean value represented by a is greater than a mean value represented by b, which, in turn, is greater than a mean value represented by c (i.e., a > b > c).
138
each question and then rated their strength of endorsement of each item on a 9-point Likert scale. The questions and anchors for each scale follow. 1. Since you recovered the memory, do you feel your self-esteem has changed? The anchors were 1 (“I feel much worse about myself ”) and 8 (“I feel much better about myself”). 2. Since you recovered the memory, do you find that other people are more or less supportive of you? The anchors were 1 (“I have many less social supports since I recovered the memory”) and 8 (“I have many more social supports since I recovered the memory”). 3. Since you recovered the memory, has your level of happiness changed? The anchors were 1 (“I feel much less happy”) and 8 (“I feel much happier”). 4. Overall, since you recovered the memory, have things been better or worse for you? The anchors were 1 (“Things have been much worse”) and 8 (“Things have been much better”). 5. If you had to do it all over again, would you choose NOT to remember that you were abused? The anchors were 1 (“I would definitely choose not to remember”) and 8 (“I would definitely choose to remember”). 6. Do you feel that recovering the memory changed how you understand yourself? The anchors were 1 (“I understand myself much less now”) and 8 (“I understand myself much better now”).
The results revealed that 100% of the subjects reported at least some benefits from recovering their memories of abuse, and 73% reported benefits in response to all questions. After recovering their memories, 100% of the subjects reported increased self-esteem (M = 6.7), 82% reported increased social support (M = 5.4), 91% reported increased happiness (M = 5.8), 82% reported overall life improvement (M = 6.2), 91% said they would choose to remember their abuse if they could do it all over again (M = 7.2), and 100% reported increased self-understanding (M = 7.5). These pilot data suggest that Bass and Davis (1988) may be correct. Identifying oneself as an abuse survivor by recovering memories of abuse may yield psychological benefits to already distressed individuals even if the effects of the abuse itself have been clearly harmful.
Forgetting Trauma-Related Material Some trauma theorists assert that sexually abused children develop an avoidant (or dissociative) encoding style that enables them to disengage attention from terrifying stimuli during abuse episodes and redirect it elsewhere (e.g., Terr, 1991). Unable to escape physically from the perpetrator, abused children may endeavor to escape mentally. Although adaptive in the short run, a dissociative style of coping with threats presumably fosters long-term psychological problems.
139
This hypothesized cognitive style ought to be evident in the laboratory and detectable through directed forgetting methods. There are two main directedforgetting paradigms, and each engages different mechanisms (Golding, in press; Johnson, 1994). In the list method, the experimenter presents the subject with a list of words, halfway through the procedure, he or she tells the subject that the first set of words was “only for practice” and hence can be forgotten. After this instruction to forget, the experimenter resumes presentation of the words to be remembered. In the item method, an instruction to remember or to forget follows the presentation of each word. For both paradigms, the experimenter asks the subject to recall as many words as possible, regardless of original instructions. Generally, subjects tend to recall more remember-words than forget-words, but the mechanisms underlying this effect differ for each paradigm. In the list paradigm, forget-words appear to have been encoded but areinhibited at retrieval. In the item paradigm, subjects appear to attenuate their encoding of a word as soon as the forget instruction occurs; hence, later recall failures seem chiefly attributable to poor encoding of the forget-words. In our first experiment, we tested adult female survivors of childhood sexual abuse with PTSD, psychiatrically healthy survivors of childhood sexual abuse, and nonabused control subjects (McNally, Metzger, Lasko, Clancy, & Pitman, 1998). Subjects had experienced penetrative abuse, and most had always remembered it. The subject viewed a series of words on a computer screen that were either trauma-related (e.g., incest), positive (e.g., cheerful), or neutral (e.g., cupboard). After each word’s presentation, instructions appeared that told the subject either to remember or to forget the immediately preceding word. Immediately after this encoding phase, we asked subjects to write down all the words they could remember, regardless of whether a word had been followed by remember or forget instructions. The standard directed forgetting effect emerging from basic research indicates that subjects abort encoding of forget-words and, accordingly, exhibit better recall for remember-words than for forget-words (Johnson, 1994). However, if psychologically traumatized survivors are characterized by a superior ability to disengage attention from trauma cues, then the PTSD group, compared to healthy survivors and controls, should exhibit superior forgetting of trauma words relative to other words. The results contradicted this hypothesis: childhood sexual abuse survivors with PTSD exhibited memory deficits for positive and neutral words they had been instructed to remember, and they exhibited excellent memory for trauma words, including words they had been instructed to forget. Exhibiting the standard directed forgetting effect, healthy survivors and control subjects recalled remember-words more often than forget-words, regardless of the word’s valence. Contrary to our original hypothesis, survivors with PTSD seemed to exhibit an impairment in the ability to banish trauma words from awareness. One limitation of the previous study was that our sexual abuse survivors were suffering from intrusive thoughts as part of their current PTSD. Perhaps indi-
140
viduals reporting repressed or recovered memories of abuse would be most likely to exhibit superior forgetting of trauma words in the laboratory. To test this hypothesis, we used the same directed-forgetting protocol to compare the relative ability to forget trauma words in recovered memory subjects, repressed memory subjects, and nonabused control subjects (McNally, Clancy, & Schacter, 2001). If anyone should exhibit heightened ability to forget trauma words, it should be individuals who once repressed (or still have repressed) their memories of abuse. Contrary to this hypothesis, the memory performances of the repressed and recovered memory groups were entirely normal. Both groups recalled remember-words better than forget-words, regardless of valence. (Control subjects, however, exhibited difficulty forgetting trauma words.) The directed-forgetting paradigm involves the forgetting of trauma words, mere pale proxies for autobiographical memories of abuse. If these individuals developed a skill for repressing or dissociating genuine memories of trauma, it should easily be brought into play in the laboratory when they are asked to forget words linked to abuse. Yet we found no evidence of any superior ability to forget trauma words among our childhood sexual abuse groups, and those with PTSD appear to experience a breakdown in the ability to banish trauma words from awareness. The very fact that abuse survivors with PTSD have difficulty forgetting is fully consistent with the cardinal symptom of the disorder: intrusive recollection of traumatic memories. On the other hand, preoccupation with the possibility that one might have been sexually abused might heighten the salience of trauma words, making it difficult for subjects to disengage attention from them and forget them in this paradigm.
Intrusion of Traumatic Material PTSD is characterized by intrusive recollection of traumatic memories. Assessment of this symptom has traditionally relied on introspective self-reports disclosed in interviews or questionnaires. The emotional Stroop color-naming paradigm provides a quantitative measure of intrusive cognition that does not rely on introspection (for reviews, see McNally, 1998; Williams, Mathews, & MacLeod, 1996). In this laboratory paradigm, subjects view words that vary in emotional significance and are asked to name the colors of the words while ignoring their meanings. When the meaning of a word captures the subject’s attention, the subject exhibits a delay in naming its color. The intrusiveness of meaning is reflected in delayed color naming (Stroop interference). Psychiatrically disturbed trauma survivors who qualify for PTSD take longer to name the colors of words related to their trauma than to name the colors of neutral words or emotional words unrelated to their trauma (for a review, see McNally, 1998).
141
Trauma survivors who do not suffer from PTSD seldom show much Stroop interference. Psychologically traumatized survivors of combat (e.g., McNally, Kaspi, Riemann, & Zeitlin, 1990), rape (e.g., Foa, Feske, Murdock, Kozak, & McCarthy, 1991), and sexual abuse (Dubner & Motta, 1999) exhibit Stroop interference for words related to their traumatic memories. Indeed, color naming of trauma words is more strongly associated with intrusive symptoms than with avoidance/numbing symptoms (Cassiday, McNally, & Zeitlin, 1992). We used the emotional Stroop paradigm to test whether subjects reporting either continuous, repressed, or recovered memories of sexual abuse, compared to nonabused control subjects, would exhibit interference for trauma words (McNally, Clancy, Schacter, & Pitman, 2000a). Although we did not conduct diagnostic interviews to assess for PTSD, the repressed memory group exhibited elevations on the CMISS. Moreover, if severity of trauma is what causes PTSD and what motivates repression of traumatic memories, subjects who cannot recall their presumably repressed memories may nevertheless exhibit interference on the emotional Stroop task—a measure of automatic (obligatory) emotional processing (McNally, 1995). Subjects named the colors of a series of trauma-related (e.g., molested), positive (e.g., elation), and neutral (e.g., carpet) words on a computer screen as quickly as possible. Unlike patients with PTSD, none of the groups exhibited delayed color naming of trauma words relative to neutral or positive ones. These data suggest that interference effects for trauma-related material may be confined to abuse survivors who qualify for PTSD (Dubner & Motta, 1999). Believing that one harbors repressed memories of abuse is not associated with patterns of interference that characterize survivors with PTSD.
Memory Distortion Many psychologists have warned that certain therapeutic techniques may inadvertently foster false memories of abuse in distressed patients (e.g., Poole et al., 1995). One such technique is guided imagery. Misguided therapists who believe that patients with certain symptom patterns harbor repressed memories of abuse will ask these patients to visualize abuse scenarios that might have happened, hoping that the content of guided imagery might trigger recollection of the blocked trauma. Unfortunately, repeated visualization of imagined events may increase confidence that the events actually occurred (Garry, Manning, Loftus, & Sherman, 1996; Heaps & Nash, 1999). Using an imagination inflation paradigm developed by Garry et al. (1996), we tested whether recovered memory subjects are more susceptible than control subjects to this form of memory distortion (Clancy, McNally, & Schacter, 1999). Subjects rated their confidence regarding whether they had experienced unusual, but
142
nontraumatic, childhood events (e.g., getting stuck in a tree, finding a $10 bill in parking lot). During a subsequent visit to our laboratory, they performed a brief guided imagery task requiring them to visualize certain events but not others. Immediately thereafter, they rated their confidence in whether any of these events had actually occurred during their childhoods. Perhaps because of modest statistical power, the increase in confidence that imagined events had in fact happened fell short of significance. However, the effect size for this imagination inflation phenomenon was more than twice as large in the control group than in the recovered memory group. Several subjects in the recovered memory group later said that they thought the experiment was about creating false memories in the laboratory, perhaps indicating that the procedure was transparent. Continuing to explore whether recovered memory subjects are especially vulnerable to false memory effects in the laboratory, we next applied a variant of the Deese/Roediger-McDermott (DRM) paradigm (Deese, 1959; Roediger & McDermott, 1995). This paradigm is less transparent than the guided imagery one. During the encoding phase, subjects hear a series of word lists, each consisting of semantically related items (e.g., sour, bitter, candy, sugar) that converge on a nonpresented word—the “false target”—that captures the theme of the list (e.g., sweet). The false memory effect occurs when subjects “remember” having heard the false target on subsequent recognition tests. Research suggests that false memory effects in this paradigm occur when people rely on their memory for the general semantic aspects (or gist) of the items they studied. We tested subjects who reported either repressed, recovered, or continuous memories of childhood sexual abuse and nonabused control subjects (Clancy, Schacter, McNally, & Pitman, 2000). None of the lists was trauma-related. The results revealed that the recovered memory group was more prone to “remember” false targets than were the other groups. This group exhibited no general memory impairment; material that had been presented was remembered as well by the recovered memory subjects as by the other subjects. Moreover, the higher a subject’s DES score, the more likely she was to exhibit the false memory effect. (The DES also predicts this effect among college students in the DRM paradigm; Winograd, Peluso, & Glover, 1998.) Thus, self-reported dissociation in everyday life was linked to remembering words never presented. This tendency for false memory formation notwithstanding, we do not know whether the effect would be more or less pronounced for material directly related to abuse. What implications do these findings have for false memories of trauma? The fact that recovered memory subjects apparently relied on gist rather than memory for specific words suggests a processing style that might contribute to memory confusion later in life. Affirming that something occurred based on general resemblance to events that did occur might lead one to suspect they were sexually abused if they did, in fact, experience other similar adverse events in childhood (e.g., emotional abuse, neglect).
143
False Memory Effects in
“Space Alien Abductees”
The DRM experiment suggested that individuals who report recovered memories of sexual abuse are prone to exhibit false memory effects in the laboratory. But this does not mean that their recovered memories of abuse are false. We were unable to obtain independent evidence bearing on the veracity of their abuse reports, but external corroboration of abuse is often difficult to obtain. In our next experiment, we extended our work to a group of subjects whose memories of trauma were likely inaccurate: people who report recovering memories of abduction by space aliens. Applying Robinson and Roediger’s (1997) DRM paradigm, we tested the false recognition and false recall propensities in three groups of subjects (Clancy, McNally, Schacter, Lenzenweger, & Pitman, 2002). One group reported recollections of alien abduction. A second group, similar to our repressed memory subjects, included individuals who believed they had been abducted but had no explicit memories of the trauma. They inferred a history of abduction from sources such as unexplained marks on their bodies, interpreted as evidence of alien medical probes; a passion for reading science fiction books; panic attacks triggered by seeing drawings of aliens on book covers; and so forth. They assumed that the aliens had control of their memories or that the abduction occurred in another dimension. A third (control) group included individuals who denied a history of alien abduction. Like subjects who report having recovered memories of sexual abuse, those reporting abduction by space aliens exhibited greater false memory effects in the DRM paradigm than did either the control group or the group that had suspected (but had no memory) they had been abducted. The group that believed they had been abducted, but who had no conscious memories of abduction, exhibited greater false memory effects than the control group but not as great effects as the group that had recovered “memories” of abduction. Therefore, subjects who had recovered “memories” of abduction relied most on gist memory in the DRM paradigm, a propensity that may explain why they recall these experiences after having undergone quasi-hypnotic regression therapies, read books about abduction, and so forth. Accordingly, this pattern may explain why the recovered memory group went on to develop full-blown false memories of abduction, whereas others did not. People who have recovered memories of sexual abuse often experience intense emotion while remembering these long-forgotten events, tempting some therapists to credit the veracity of the accounts. Surely the memories must be genuine, some therapists believe; otherwise, how could these recollections trigger such extreme affect? In fact, considerable research indicates that when people with PTSD recollect their traumatic experiences—more specifically, when they listen to
144
audiotaped descriptions of them—they exhibit heightened psychophysiologic reactivity (for a review, see Orr & Roth, 2000). Indeed, Orr et al. (1998) found not only that female survivors of sexual abuse with PTSD exhibited greater physiological responding while listening to autobiographical trauma scripts than did abuse survivors without PTSD but also that those with PTSD who had recently recovered their memories tended to be more physiologically responsive than were those who had always remembered their abuse. Using this script-driven imagery procedure, we tested whether people reporting abduction by space aliens exhibit heightened psychophysiologic responding while listening to scripts describing their purported abduction experiences (McNally et al., 2002). If they did, then this finding would imply that belief that one has been traumatized can generate a physiologic profile similar to that of PTSD patients who have endured traumatic conditioning experiences. Each of our 10 “abductees” traced contact with space aliens to an episode of apparent sleep paralysis accompanied by hyponopompic (upon-awakening) hallucinations. These episodes occur as the person is awakening from rapid eye movement (REM) sleep (Cheyne, Rueffer, & Newby-Clark, 1999). Dreams occur during REM sleep, and this stage of sleep is associated with full–body paralysis. However, during an episode of sleep paralysis, the cognitive and motoric aspects of REM become desynchronized for several seconds to several minutes. The sleeper awakens before paralysis has waned and becomes aware of an inability to move. No more pathological than a hiccup, sleep paralysis occurs at least once to approximately 30% of the general population (Cheyne, Newby-Clark, & Rueffer, 1999). About 5% of the population, however, has also experienced hallucinations during these episodes, including flashing lights, buzzing sounds, feelings of levitating off the bed, electrical tingling sensations shooting through one’s body, and seeing threatening figures hovering near one’s bed. Within a few seconds or minutes, the person is entirely awake, can move again, and the hallucinations cease. None of our subjects realized that he or she was experiencing sleep paralysis, and all eventually interpreted the episode as an encounter with alien beings. Frightened by such experiences, some sought hypnosis to recover presumably repressed aspects of these episodes. During these sessions, they recalled additional “memories” of having been transported into spacecraft, where they endured medical and sexual experimentation at the hands of alien beings. For the script-driven imagery protocol, each abductee furnished material for five individualized, autobiographical scripts: two scripts related to abduction trauma; a script related to a different, highly stressful event; a script related to a positive experience; and one related to an emotionally neutral experience. Our control group consisted of individuals who denied a history of alien abduction. Each control subject heard the scripts of one of the abductees. Compared with control subjects, abductees exhibited greater heart rate and skin conductance responses during imagery of abduction scripts than to imagery of the other scripts (ps < .06; effect size rs = .40). Although none of the
145
abductees ever met full criteria for PTSD, the magnitude of their heart rate and skin conductance responses to their abduction scripts matched or exceeded the reactivity exhibited by PTSD subjects to their trauma scripts in previous research. For example, abductees had a mean skin conductance response of 1.82 æS, whereas Vietnam combat veterans with PTSD in Keane et al.’s (1998) definitive study had a mean response of 0.8 æS. Likewise, the mean heart rate response of abductees was 7.8 bpm, whereas for Vietnam veterans with PTSD it was 3.2 bpm (Keane et al., 1998). Believing that one has been traumatized by space aliens produces heightened physiologic responding similar to that of people exposed to combat and other traumatic events. Therefore, emotional responding during recollection provides no guarantee that the memory is veridical.
Are Memories of Trauma Special? Some clinical theorists believe that memories of traumatic events differ qualitatively from memories of other emotionally intense experiences. Consider the claim of Brown et al. (1998): Thus, across all studies, a robust finding is that narrative memory for the gist or central action of a negative emotionally arousing event is extremely well retained. The more involved the person is, the more likely the memory will be retained. Traumatic events are a significant exception to this general rule. A significant portion of traumatized individuals suffer from traumatic amnesia; even if their narrative memory for the traumatic event is retained, it may not be readily accessible. (p. 368)
Intensity of emotion, Brown et al. believe, is a key variable. “[W]hen emotional material reaches the point of being traumatic in intensity—something that cannot be replicated in artificial laboratories—in a certain subpopulation of individuals, material that is too intense may not be able to be consciously processed and so may become unconscious and amnesic” (p. 97). Others have made similar assertions (e.g., Joseph, 1999; van der Kolk, 1994), suggesting that memory for traumatic experience follows an inverted-U function (Yerkes & Dodson, 1908). That is, emotional arousal enhances memory, but only up to a point. After that point is reached, these theorists believe that intense emotion somehow undermines hippocampal function and impairs explicit memory for the trauma. This theory is riddled by many conceptual and empirical problems (see McNally, 2003c, chapter 6, for detailed critique). First, intense arousal does not block the formation of explicit narrative memory for traumatic experiences. As Langer’s (1991) work has shown, survivors of the Nazi concentration camps are haunted by detailed, explicit, vivid memories of the horrors they experienced. Indeed, the evidence overwhelmingly shows that traumatic experiences are remembered all too well (Pope et al., 1999). People tend not to forget the central aspects of trau-
146
matic events, even if they fail to encode or remember details of these experiences. [as with nontraumatic emotional memories; see chapter 1—Eds.]. Second, abundant laboratory research shows that amygdala activation promotes encoding and hippocampus-mediated explicit memory for emotionally intense experiences (for a review, see McGaugh, Ferry, Vazdarjanova, & Roozendaal, 2000). To be sure, release of stress hormones can impair performance on certain explicit memory tasks mediated by the hippocampus (for a review, see Lupien & McEwen, 1997). But stress does not impair memory for the stress-producing experience itself. For example, Kirschbaum, Wolf, May, Wippich, & Hellhammer (1996) exposed subjects to a social stressor (giving a speech prior to trying to solve math problems out loud) before having them learn and recall a list of words. The higher the levels of stress hormone (cortisol) triggered by this stressor, the fewer words subjects recalled. Hence, stress can impair memory for an incidental activity, such as memorizing words. But it does not abolish memory for the stressor itself. Third, as Christianson’s (1992) review shows, studies on emotional stress and memory fail to support a Yerkes-Dodson–like (1908) notion that moderately high stress enhances memory and extreme stress impairs it. (Moreover, as some trauma theorists seem to have forgotten, Yerkes and Dodson studied visual discrimination learning in mice, not memory for trauma. Hence, the work of these two scientists has scant relevance for the recovered memory debate.) As Easterbrook’s (1959) theory implies, extreme stress enhances memory for the central aspects of the traumatic experience, sometimes at the expense of the peripheral details (McNally, 2003c, chapter 5). Fourth, some theorists have suggested that extreme stress establishes implicit memories of the traumatic experience while sometimes undermining explicit memories of it (e.g., van der Kolk, 1994). As Brown et al. (1998) asserted, “Much of the memory for trauma is retained as an implicit rather than explicit memory” (p. 483). In contrast to memory for ordinary events, dissociated trauma memory supposedly remains frozen and relatively impervious to distortion during the passage of time. These allegedly dissociated implicit memories of trauma are nevertheless manifested “in the form of body memories [conditioned emotional responses?], flashbacks, fragments, sudden intense feelings, avoidant behaviors, images, sensory processes, and dreams” (p. 187). There are several problems with this formulation (McNally, 2003c, chapter 6). In the absence of explicit memory, one cannot reason backward and infer a repressed or dissociated memory from dreams, sudden feelings, and so forth. These alleged implicit markers of buried memories can arise from many causes. Moreover, implicit memories are just as subject to change and distortion as explicit ones are (Lustig & Hasher, 2001). Finally, the notion that people can undergo trauma yet retain only implicit traces (e.g., conditioned emotional responses) contradicts recent critiques of the literature on conditioning without awareness (Lovibond & Shanks, 2001; Shanks & Lovibond, 2001). The
147
clinical implications are clear: if people are sufficiently traumatized to acquire conditioned emotional responses and other implicit memories of the experience, they will retain explicit memories as well.
Dispelling Semantic Confusion The debate about repressed and recovered memories has been partly fueled by confusion concerning the meaning of basic concepts such as remembering and forgetting. At times, the debate has been inflamed by proponents on each side who use the same words for different issues. Consider the verb forget. In accordance with everyday parlance, many of our research subjects say, “I was abused as a child, but then I forgot about it.” Further questioning clarifies what they mean. Many say they deliberately tried to push these experiences out of their minds after they occurred. Engaged by other matters, they often succeeded in their attempts not to think about their abuse, sometimes for long periods. Prior to their enrolling in our research program, they encountered cues that reminded them of their “forgotten” abuse—abuse that had not entered their mind for some time. Other subjects say, “I never forgot that I was abused as a child.” Obviously, memories of their abuse were often absent from awareness for hours, days, or perhaps even weeks at a time, only to pop into mind either spontaneously or when triggered by some reminder. Some traumatic amnesia theorists seem to believe that memories of sexual abuse ought to come to mind with reasonable frequency, and if someone does not think about the abuse for months or years at a time, some mechanism must be actively inhibiting its retrieval during these periods of “forgetting.” Because these theorists assume that sexual abuse memories are highly emotionally charged, they assume that a failure of these memories to enter the mind during long periods must be attributable to an active force keeping them out of awareness. The problem with this assumption is that the evidence for this postulated mechanism—repression or dissociation—is inferred from the very phenomena it is designed to explain. In these cases, there is no independent evidence of inhibitory mechanisms other than the mere absence of certain memories from awareness for periods of time. To be sure, cognitive psychologists have devised ingenious methods for confirming active inhibitory processes in memory, at least for neutral words (e.g., Anderson & Green, 2001). But to defend a claim of active inhibition, one should show that people are unable to recall their abuse in the presence of adequate retrieval cues, such as the question: “Were you ever sexually abused as child?” An inability to recall abuse when it should be readily accessible provides a stronger basis for inferring inhibition than does the mere failure to think about it for periods of time (everyday forgetting), which does not require postulation of any inhibitory mechanism.
148
Thus, merely because someone has not thought about something for a long time—has “forgotten” it—does not mean that the person has been unable to remember it. One need not postulate any special mechanisms to explain why someone tried not to think about something unpleasant and managed not to think about it for long stretches of time. Only an inability to remember when exposed to adequate retrieval cues can provide a reasonable basis for labeling the phenomenon as amnesia (assuming, of course, that the event was encoded in the first place). Until such evidence is adduced, the most parsimonious explanation for not thinking about trauma for long periods of time—for forgetting it— must lie with the conventional memory and forgetting mechanisms of cognitive psychology, not repression or traumatic dissociation. Finally, some recovered memories may not correspond to genuine events. Some recovered memories are likely false. In our research, we have noted two strikingly different types of recovered memory experience. In one type, subjects are suddenly reminded of events that they had not thought about in many years. They are surprised at their recollection but not at the content of the memory per se. For example, in one of our cases, the subject suddenly recalled her own abuse on learning that her daughter had just been molested. These individuals often remark, “My God! I can’t believe I forgot that.” In the other type, subjects realize that they are abuse survivors, sometimes gradually recalling new memories over a time. For example, one of our cases recovered memories of being involved in cult abuse and cannibalism. This kind of recollection differs from those in which the person is surprised not by the content of the recollection but by not having thought about it for so long. The second type of experience, we suspect, may be more likely to involve false memory than the first type.
Note Preparation of this chapter was supported in part by NIMH grant MH61268 awarded to the first author.
References American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders. 3rd ed., rev. Washington, DC: APA. Anderson, M. C., & Green, C. (2001). Suppressing unwanted memories by executive control. Nature, 410, 366–369. Anderson, M. C., & Neely, J. H. (1996). Interference and inhibition in memory retrieval. In E. L. Bjork & R. A. Bjork (Eds.), Memory (pp. 237–313). San Diego, CA: Academic Press. Archibald, H. C., & Tuddenham, R. D. (1965). Persistent stress reaction after combat: A 20-year follow-up. Archives of General Psychiatry, 12, 475–481. Arrigo, J. M., & Pezdek, K. (1997). Lessons from the study of psychogenic amnesia. Current Directions in Psychological Science, 6, 148–152.
149
Bass, E., & Davis, L. (1988). The courage to heal: A guide for women survivors of child sexual abuse. New York: Harper and Row. Beck, A. T., & Steer, R. A. (1987). Beck Depression Inventory manual. San Antonio, TX: The Psychological Corporation. Bernstein, E. M., & Putnam, F. W. (1986). Development, reliability, and validity of a dissociation scale. Journal of Nervous and Mental Disease, 174, 727–735. Blume, E. S. (1990). Secret survivors: Uncovering incest and its aftereffects in women. New York: Wiley. Briere, J., & Conte, J. (1993). Self-reported amnesia for abuse in adults molested as children. Journal of Traumatic Stress, 6, 21–31. Brown, D., Scheflin, A. W., & Hammond, D. C. (1998). Memory, trauma treatment, and the law. New York: Norton. Cassiday, K. L., McNally, R. J., & Zeitlin, S. B. (1992). Cognitive processing of trauma cues in rape victims with post-traumatic stress disorder. Cognitive Therapy and Research, 16, 282–295. Ceci, S. J., & Loftus, E. F. (1994). “Memory work”: A royal road to false memories? Applied Cognitive Psychology, 8, 351–364. Cheit, R. E. (1998). Consider this, skeptics of recovered memory. Ethics and Behavior, 8, 141–160. Cheit, R. E. (1999). Junk skepticism and recovered memory: A reply to Piper. Ethics and Behavior, 9, 295–318. Cheyne, J. A., Newby-Clark, I. R., & Rueffer, S. D. (1999). Relations among hypnagogic and hypnopompic experiences associated with sleep paralysis. Journal of Sleep Research, 8, 313–317. Cheyne, J. A., Rueffer, S. D., & Newby-Clark, I. R. (1999). Hypnagogic and hypnopompic hallucinations during sleep paralysis: Neurological and cultural construction of the night-mare. Consciousness and Cognition, 8, 319–337. Christianson, S.-Å. (1992). Emotional stress and eyewitness memory: A critical review. Psychological Bulletin, 112, 284–309. Clancy, S. A. (2000). Child Abuse Survivor Questionnaire. Unpublished questionnaire, Department of Psychology, Harvard University, Cambridge, MA. Clancy, S. A., McNally, R. J., & Schacter, D. L. (1999). Effects of guided imagery on memory distortion in women reporting recovered memories of childhood sexual abuse. Journal of Traumatic Stress, 12, 559–569. Clancy, S. A., McNally, R. J., Schacter, D. L., Lenzenweger, M. F., & Pitman, R. K. (2002). Memory distortion in people reporting abduction by aliens. Journal of Abnormal Psychology, 111, 455–461. Clancy, S. A., Schacter, D. L., McNally, R. J., & Pitman, R. K. (2000). False recognition in women reporting recovered memories of sexual abuse. Psychological Science, 11, 26–31. Courtois, C. A. (1992). The memory retrieval process in incest survivor therapy. Journal of Child Sexual Abuse, 1, 15–31. Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal of Experimental Psychology, 58, 17–22. Dollinger, S. J. (1985). Lightning-strike disaster among children. British Journal of Medical Psychology, 58, 375–383.
150
Dubner, A. E., & Motta, R. W. (1999). Sexually and physically abused foster care children and posttraumatic stress disorder. Journal of Consulting and Clinical Psychology, 67, 367–373. Easterbrook, J. A. (1959). The effect of emotion on cue utilization and the organization of behavior. Psychological Review, 66, 183–201. Eder, M. D. (1917). War-shock. London: William Heinemann. Femina, D. D., Yeager, C. A., & Lewis, D. O. (1990). Child abuse: Adolescent records vs. adult recall. Child Abuse and Neglect, 14, 227–231. Foa, E. B., Feske, U., Murdock, T. B., Kozak, M. J., & McCarthy, P. R. (1991). Processing of threat-related information in rape victims. Journal of Abnormal Psychology, 100, 156–162. Freyd, J. J. (1996). Betrayal trauma: The logic of forgetting childhood abuse. Cambridge, MA: Harvard University Press. Garry, M., Manning, C. G., Loftus, E. F., & Sherman, S. J. (1996). Imagination inflation: Imagining a childhood event inflates confidence that it occurred. Psychonomic Bulletin & Review, 3, 208–214. Golding, J. M. (in press). Understanding directed forgetting. In A. Wenzel & D. C. Rubin (Eds.), Cognitive methods in psychopathology research. Washington, DC: American Psychological Association Press. Goodman, G. S., Ghetti, S., Quas, J. A., Edelstein, R. S., Alexander, K. W., Redlich, A. D., et al. (2003). A prospective study of memory for child sexual abuse: New findings relevant to the repressed-memory controversy. Psychological Science, 14, 113–118. Heaps, C., & Nash, M. (1999). Individual differences in imagination inflation. Psychonomic Bulletin and Review, 6, 313–318. Henderson, J. L., & Moore, M. (1944). The psychoneuroses of war. New England Journal of Medicine, 230, 273–278. Herman, J. L. (1992). Trauma and recovery. New York: Basic Books. Johnson, H. M. (1994). Processes of successful intentional forgetting. Psychological Bulletin, 116, 274–292. Joseph, R. (1999). The neurology of traumatic “dissociative” amnesia: Commentary and literature review. Child Abuse & Neglect, 23, 715–727. Keane, T. M., Caddell, J. M., & Taylor, K. L. (1988). Mississippi Scale for CombatRelated Posttraumatic Stress Disorder: Three studies in reliability and validity. Journal of Consulting and Clinical Psychology, 56, 85–90. Keane, T. M., Kolb, L. C., Kaloupek, D. G., Orr, S. P., Blanchard, E. B., Thomas, R. G., et al. (1998). Utility of psychophysiological measurement in the diagnosis of posttraumatic stress disorder: Results from a Department of Veterans Affairs Cooperative Study. Journal of Consulting and Clinical Psychology, 66, 914–923. Kirschbaum, C., Wolf, O. T., May, M., Wippich, W., & Hellhammer, D. H. (1996). Stress- and treatment-induced elevations of cortisol levels associated with impaired declarative memory in healthy adults. Life Sciences, 58, 1475–1483. Kotagal, S., Rawlings, C. A., Chen, S.-C., Burris, G., & Nouri, S. (1982). Neurologic, psychiatric, and cardiovascular complications in children struck by lightning. Pediatrics, 70, 190–192.
151
Langer, L. L. (1991). Holocaust testimonies: The ruins of memory. New Haven, CT: Yale University Press. Lindsay, D. S., & Read, J. D. (1994). Psychotherapy and memories of childhood sexual
abuse: A cognitive perspective. Applied Cognitive Psychology, 8, 281–338.
Loftus, E., & Ketcham, K. (1994). The myth of repressed memory. New York: St. Martin’s.
Lovibond, P. F., & Shanks, D. R. (2002). The role of awareness in Pavlovian conditioning: Empirical evidence and theoretical implications. Journal of Experimental Psychology: Animal Behavior Processes, 28, 3–26. Lupien, S. J., & McEwen, B. S. (1997). The acute effects of corticosteroids on cognition: Integration of animal and human model studies. Brain Research Reviews, 24, 1–27. Lustig, C., & Hasher, L. (2001). Implicit memory is not immune to interference. Psychological Bulletin, 127, 618–628. McGaugh, J. L., Ferry, B., Vazdarjanova, A., & Roozendaal, B. (2000). Amygdala: Role in modulation of memory storage. In J. P. Aggleton (Ed.), The amygdala: A functional analysis (2nd ed., pp. 391–423). Oxford: Oxford University Press. McNally, R. J. (1995). Automaticity and the anxiety disorders. Behaviour Research and Therapy, 33, 747–754. McNally, R. J. (1998). Experimental approaches to cognitive abnormality in posttraumatic stress disorder. Clinical Psychology Review, 18, 971–982. McNally, R. J. (2001). The cognitive psychology of repressed and recovered memories of childhood sexual abuse: Clinical implications. Psychiatric Annals, 31, 509–514. McNally, R. J. (2003a). Progress and controversy in the study of posttraumatic stress disorder. Annual Review of Psychology, 54, 229–252. McNally, R. J. (2003b). Recovering memories of trauma: A view from the laboratory. Current Directions in Psychological Science, 12, 32–35. McNally, R. J. (2003c). Remembering trauma. Cambridge, MA: Belknap Press/ Harvard University Press. McNally, R. J., Clancy, S. A., & Schacter, D. L. (2001). Directed forgetting of trauma cues in adults reporting repressed or recovered memories of childhood sexual abuse. Journal of Abnormal Psychology, 110, 151–156. McNally, R. J., Clancy, S. A., Schacter, D. L., & Pitman, R. K. (2000a). Cognitive processing of trauma cues in adults reporting repressed, recovered, or continuous memories of childhood sexual abuse. Journal of Abnormal Psychology, 109, 355–359. McNally, R. J., Clancy, S. A., Schacter, D. L., & Pitman, R. K. (2000b). Personality profiles, dissociation, and absorption in women reporting repressed, recovered, or continuous memories of childhood sexual abuse. Journal of Consulting and Clinical Psychology, 68, 1033–1037. McNally, R. J., Kaspi, S. P., Riemann, B. C., & Zeitlin, S. B. (1990). Selective processing of threat cues in posttraumatic stress disorder. Journal of Abnormal Psychology, 99, 398–402. McNally, R. J., Lasko, N. B., Clancy, S. A., Macklin, M. L., Pitman, R. K., & Orr, S. P. (2002). Psychophysiologic responding during script-driven imagery in people reporting abduction by space aliens. Manuscript submitted for publication. McNally, R. J., Metzger, L. J., Lasko, N. B., Clancy, S. A., & Pitman, R. K. (1998).
152
Directed forgetting of trauma cues in adult survivors of childhood sexual abuse with and without posttraumatic stress disorder. Journal of Abnormal Psychology, 107, 596–601. Orr, S. P., Lasko, N. B., Metzger, L. J., Berry, N. J., Ahern, C. E., & Pitman, R. K. (1998). Psychophysiologic assessment of women with posttraumatic stress disorder resulting from childhood sexual abuse. Journal of Consulting and Clinical Psychology, 66, 906–13. Orr, S. P., & Roth, W. T. (2000). Psychophysiological assessment: Clinical applications for PTSD. Journal of Affective Disorders, 61, 225–240. Poole, D. A., Lindsay, D. S., Memon, A., & Bull, R. (1995). Psychotherapy and the recovery of memories of childhood sexual abuse: U.S. and British practitioners’ opinions, practices, and experiences. Journal of Consulting and Clinical Psychology, 63, 426–437. Pope, H. G., Jr., Oliva, P. S., & Hudson, J. I. (1999). Repressed memories: The scientific status. In D. L. Faigman, D. H. Kaye, M. J. Saks, & J. Sanders (Eds.), Modern scientific evidence: The law and science of expert testimony (Vol. 1), Pocket part (pp. 115–155). St. Paul, MN: West. Read, J. D. (1997). Memory issues in the diagnosis of unreported trauma. In J. D. Read & D. S. Lindsay (Eds.), Recollections of trauma: Scientific evidence and clinical practice (pp. 79–108). New York: Plenum. Robinson, K. J., & Roediger, H. L., III. (1997). Associative processes in false recall and false recognition. Psychological Science, 8, 231–237. Roediger, H. L., III, & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 803–814. Sargant, W., & Slater, E. (1941). Amnesic syndromes in war. Proceedings of the Royal Society of Medicine, 34, 757–764. Schooler, J. W., Bendiksen, M., & Ambadar, Z. (1997). Taking the middle line: Can we accommodate both fabricated and recovered memories of sexual abuse? In M. A. Conway (Ed.), Recovered memories and false memories (pp. 251–292). Oxford: Oxford University Press. Shanks, D. R., & Lovibond, P. F. (2002). Autonomic and eyeblink conditioning are closely related to contingency awareness: Reply to Wiens and Öhman (2002) and Manns et al. (2002). Journal of ExperimentalPsychology: Animal Behavior Processes, 28, 38–42. Shin, L. M., McNally, R. J., Kosslyn, S. M., Thompson, W. L., Rauch, S. L., Alpert, N. M., et al. (1999). Regional cerebral blood flow during script-driven imagery in childhood sexual abuse-related PTSD: A PET investigation. American Journal of Psychiatry, 156, 575–584. Swihart, G., Yuille, J., & Porter, S. (1999). The role of state-dependent memory in “red-outs.” International Journal of Law and Psychiatry, 22, 199–212. Tellegen, A. (1982). Brief manual for the Differential Personality Questionnaire. Minneapolis: Department of Psychology, University of Minnesota. Tellegen, A., & Atkinson, G. (1974). Openness to absorbing and self-altering experiences (“absorption”), a trait related to hypnotic susceptibility. Journal of Abnormal Psychology, 83, 268–277.
153
Terr, L. C. (1991). Childhood traumas: An outline and overview. American Journal of Psychiatry, 148, 10–20. Torrie, A. (1944, January 29). Psychosomatic casualties in the Middle East. Lancet, 246, 139–143. van der Kolk, B. A. (1994). The body keeps the score: Memory and the evolving psychobiology of posttraumatic stress. Harvard Review of Psychiatry, 1, 253–265. van der Kolk, B. A., & Fisler, R. (1995). Dissociation and the fragmentary nature of traumatic memories: Overview and exploratory study. Journal of Traumatic Stress, 8, 505–525. Vreven, D. L., Gudanowski, D. M., King, L. A., & King, D. W. (1995). The civilian version of the Mississippi PTSD Scale: A psychometric evaluation. Journal of Traumatic Stress, 8, 91–109. Wagenaar, W. A., & Groeneweg, J. (1990). The memory of concentration camp survivors. Applied Cognitive Psychology, 4, 77–87. Wilkinson, C. B. (1983). Aftermath of a disaster: The collapse of the Hyatt Regency Hotel skywalks. American Journal of Psychiatry, 140, 1134–1139. Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology. Psychological Bulletin, 120, 3–24. Williams, L. M. (1994). Recall of childhood trauma: A prospective study of women’s memories of child sexual abuse. Journal of Consulting and Clinical Psychology, 62, 1167–1176. Winograd, E., Peluso, J. P., & Glover, T. A. (1998). Individual differences in susceptibility to memory illusions. Applied Cognitive Psychology, 12, S5–S27. Yerkes, R. M., & Dodson, J. D. (1908). The relation of strength of stimulus to rapidity of habit-formation. Journal of Comparative Neurology and Psychology, 18, 459–482.
154
5
An Overview of Research Findings and Their Implications
E
xperimental support for the idea that anxious individuals may display anomalous performance on cognitive tasks extends back many decades (Calvin, Koons, Bingham, & Fink, 1953). Early research focused primarily on investigating the patterns of cognitive impairment associated with anxiety. Now convincing evidence shows that anxious participants indeed demonstrate performance decrements across measures of cognitive functioning, especially difficult and demanding memory tests (cf. Eysenck, 1982). However, contemporary accounts of such performance deficits attribute them, in part, to anxious individuals’ tendency to process task-irrelevant information associated with their worries. Recently, many researchers have directly examined whether the emotional valence of stimulus materials can moderate anxiety-linked patterns of performance on cognitive tasks. One hypothesis suggests that anxious individuals may display memory biases that selectively favor the retrieval of emotionally threatening information. Following a brief review of research showing that anxiety can be associated with memory deficits, we provide an overview of the extensive experimental work testing whether anxious participants preferentially retrieve threat-related memories. Support for this position is by no means compelling, though favorable evidence has been more apparent in certain populations of anxious individuals and for certain types of memory tasks.
Memory Deficits in Anxiety Like other dysphoric states such as depression, anxiety is associated with cognitive performance deficits, including those involving memory (Eysenck, 1982; Humphreys & Revelle, 1984; Mueller, 1992). In a highly influential account, 155
Eysenck and Calvo (1992) proposed that anxiety affects information processing in two quite different ways. First, by eliciting task-irrelevant cognitive activity such as worry, anxiety consumes processing resources within working memory (Baddeley, 1986). At the same time, however, their concern to avoid poor performance motivates anxious individuals to invest greater effort in task performance to compensate for their capacity limitations. If the task is not too demanding, this additional effort may lead anxious individuals to perform as well as, or even better than, nonanxious individuals. In contrast, if a task places high demands on temporary storage capacity, performance of anxious participants likely suffers. In general, research findings have supported predictions derived from Eysenck and Calvo’s (1992) account. For example, Darke (1988b) demonstrated that highly anxious participants display a lower digit span than do less anxious individuals and show poorer ability to intentionally recall words presented at the end of sentences. In another study, Darke (1988a) observed that, though highly anxious readers demonstrate a normal ability to draw the elementary inferences necessary to comprehend text, their performance is impaired on inference tasks that plausibly demand greater processing resources. Consistent with the hypothesis that anxiety is associated with restricted cognitive capacity, MacLeod and Donellan (1993) found that increases in mental load disproportionately impaired reasoning task performance in highly anxious individuals. One important but unresolved issue concerns whether performance deficits in anxious individuals result from mood state at the time of testing or represent an enduring characteristic of individuals with high trait vulnerability to anxiety. Eysenck and Calvo’s (1992) hypothesis causally implicates anxious mood state in such cognitive impairment. Under many circumstances, the high correlation between trait and state anxiety can make dissociating their respective roles difficult. However, state anxiety elevated by quite specific worries still causes impaired cognitive performance. For example, anxiety about mathematics is associated with temporary impairment of the ability to hold numbers in memory and with lesser ability to avoid distracting thoughts when one performs calculations (Ashcroft & Kirk, 2001). Although little direct evidence is available concerning longer-term memory consequences, Ashcroft and Kirk suggest that these temporary problems likely contribute to impaired long-term knowledge about mathematical operations. Such performance deficits clearly need to be a part of discussion on memory in anxiety disorders. However, the bulk of research and theory to be reviewed here has focused on whether anxiety is associated with the relative enhancement of memory for threatening material. There are several reasons why this pattern of memory selectivity might be expected. One is that most people tend to remember the gist of emotional events, such as the content of threatening pictures, better over time (cf. chapter 1 here). Furthermore, people remember emotional events better than matched unemotional events perhaps because of individual
156
differences in emotional arousal (see Andrews, cited in Reisberg & Heuer, 1992). Various explanations for this effect have been proposed (Christianson, 1992; Wessel, van der Kooy, & Merckelbach, 2000). Selective attention to the emotional focus of a scene may lead to better encoding of information in that location, rather than in other less well–attended locations. Also, people may elaborate disproportionately on the meaning of emotional scenes, leading to more extensive memorial representations and consequently to improved retrieval. Additionally, emotional arousal at the time of processing the scene may directly enhance the durability of the memory trace (Revelle & Loftus, 1990; Cahill & McGaugh, 1995). Though these are not mutually exclusive possibilities, particularly strong evidence supports the first suggestion, that the elevation of anxiety may be associated with an attentional focus on threat during encoding. People with heightened levels of anxiety are more likely than nonanxious individuals to let threatening words or pictures hold their attention (Fox, Russo, Bowles, & Dutton, 2001; Mathews & MacLeod, 1994; Yiend & Mathews, 2001). Even if anxiety exerts no direct effect on the retrieval process itself, greater attention to threatening aspects of stimuli during encoding likely enhances later memory for such information. However, whether more attention to a threatening stimulus event results in better or worse memory for that event also may depend on the type of encoding, as well as the task used to assess memory. If attention to threat during encoding is accompanied by subsequent processing of the distinctive meaning of this emotional information, then its later recall is likely enhanced. On the other hand, if attention to threat during encoding results, for example, principally in registration only of the aversive quality of the feelings evoked, then there may be fewer distinctive features within the memory representation to help in its later recall. A memory test requiring retrieval of certain aspects of an event could thus reveal either good or poor recall, despite the greater allocation of attention resources to this event at the time of presentation, depending on how the event has been encoded. Therefore, an encoding bias for threatening information may result in preferential memory for such material, even when the retrieval process itself is not biased to favor this type of information. Conversely, if there is a memory advantage for threatening information or not, such information nevertheless may have recruited disproportionate attentional resources during encoding. The evidence we review here suggests that increased selective attention to threatening stimuli, demonstrated so reliably by highly anxious individuals, is not consistently associated with better subsequent memory for this information. In contrast to research on depressed participants, who demonstrate recall advantages for negative information that are easy to replicate, the bulk of research on memory bias in anxious individuals has proven surprisingly negative. Anxious individuals commonly display no greater tendency to recall emotionally threatening material than less anxious individuals exposed to the same material do. Exceptions to this general rule will be highlighted during our review and
157
will be discussed at the end of this chapter. To anticipate, despite variable results across studies, greater evidence of a recall bias favoring threatening information has been obtained in studies examining panic disorder patients than in studies of other anxiety disorders or of nonclinical individuals with elevated levels of trait or state anxiety. Also, measures of implicit memory, especially those involving perceptually driven processes, may detect an anxiety-linked memory advantage for threatening information more often than measures of explicit memory, such as recall and recognition. In drawing conclusions, we distinguish between effects that could arise solely from selective encoding and those attributable to biases affecting subsequent access to, or retrieval of, emotionally valenced information. We suggest that an anxiety-linked recall advantage for threatening material is more likely when participants are free to initially encode emotional material without restrictions. Under such free encoding conditions, highly anxious groups are disproportionately likely to encode personally relevant meanings of threatening stimulus items, and this facilitates memory performance for such information, even without anxiety-linked bias in the retrieval process itself. The review that follows is organized into sections that separately describe research investigating the association between anxiety and tests of autobiographical memory, recall, recognition, and implicit memory. Within these main sections, we review studies carried out on nonclinical populations selected according to their scores on anxiety questionnaires, as well as studies that have assessed patients suffering from anxiety disorders, including generalized anxiety disorder (GAD), phobias, obsessive compulsive disorder (OCD), posttraumatic stress disorder, and panic disorder.
Memory for Emotional Information
in Anxiety Disorders
Recall of Autobiographical Memories A number of researchers have sought to establish whether autobiographical memory functioning is biased within anxious individuals. For example, Richards and Whittaker (1990) asked high and low trait-anxious individuals to recall a specific personal memory evoked by either happy cue words (e.g., pleased, lucky) or anxiety-related cue words (e.g., anger, pain). High trait-anxious participants, unlike low trait anxious individuals, were significantly faster in retrieving autobiographical memories elicited by the anxiety-related cue words, suggesting that anxiety-related memories may be disproportionately accessible for such participants. However, because Richards and Whittaker did not examine the emotional content of the memories produced, their data instead could reflect high trait-anxious participants producing more memories with objectively neutral
158
content in response to the anxiety-related cues. That is, high trait-anxious individuals may simply have associated more of their memories with the anxietyrelated cues, regardless of the emotional tone of the memories themselves. Subsequent studies by Burke and Mathews (1992), carried out on generalized anxiety disorder (GAD) patients and nonanxious control participants, employed independent raters to assess the emotional tone of memory content. GAD patients and controls were presented with emotionally neutral cue words and were instructed to recall the first personal memory brought to mind by each item. Independent blind ratings revealed a nonsignificant trend for the GAD patients to report memories that were more threatening than those reported by controls. In a second study, GAD patients, unlike controls, produced significantly more memories when directly instructed to recall anxious autobiographical memories to neutral cues. However, judges classified only 80% of these memories to be truly “anxious” in GAD patients, in comparison to 92% in the case of control participants. Thus, GAD patients may indeed use more liberal criteria when defining recollected events as anxiety–relevant. More important, it has proven difficult to replicate the early finding that anxiety is associated with biased autobiographical memory performance. Levy and Mineka (1998) obtained independent ratings of the emotional tone of memories elicited in response to neutral, positive, and threatening cue words in high and low trait-anxious participants. There were no group differences in the relative number of memories produced in response to each type of cue, in the emotional tone of these memories, or in the pattern of recall latencies across these three cue conditions. Rapee, McCallum, Melville, Ravenscroft, and Rodney (1994) presented social phobics and nonanxious participants with social-threat and neutral cue words and instructed them to retrieve associated memories of events either they or a close friend or relative experienced. Across all conditions, there was no evidence that the social phobics differed from the nonanxious participants in their tendency to recall negative events. Wenzel, Jackson, and Holt (2002) also presented social phobics and nonanxious controls with social–threat and neutral cue words, instructing them to report the first specific personal memory evoked by each cue. The social phobics showed no enhanced ability to recall memories in response to the threatening retrieval cues. Using a similar autobiographical memory task, with both happy and anxious cue words, Wilhelm, McNally, Baer, and Florin (1997) found an equivalent pattern of retrieval latencies across obsessive-compulsive disorder patients and control participants. Even in posttraumatic stress disorder, where patients might be expected to show better recall of trauma-related events, the experimental evidence provides little support for such an autobiographical memory bias. McNally, Litz, Prassas, Shin, and Weathers (1994) presented Vietnam combat veterans, some who suffered from PTSD, with negative, positive, and neutral words as retrieval cues, again instructing them to recollect personal memories elicited by each cue word. All participants were faster to retrieve memories in response to
159
the negative cues than in response to the neutral or positive cues, but this effect did not differ across groups with or without PTSD. [For evidence concerning the forgetting of traumatic memories, see chapter 4—Eds.] Overall, the balance of evidence provides little support for the idea that anxiety disorders are associated with the facilitated recall of negative autobiographical material. In many ways, this finding is rather surprising, given the strong possibility that individuals who develop anxiety pathology may actually have experienced a disproportionate number of threatening events (Finlay-Jones & Brown, 1981). Indeed, the fact that one cannot assume equivalence of negative event frequency across anxious and control populations compromises interpretation of those few early studies that claimed to demonstrate evidence of an autobiographical memory bias. To overcome this methodological problem, many researchers have turned away from autobiographical memory tasks and instead have tested for memory bias following the presentation of to-be-recalled stimuli under controlled conditions. We now turn to these studies.
Recall of Experimentally
Presented Stimuli
The emotional stimuli presented in such controlled experiments usually have been single emotional words. Participants have either read these words or used them in an imagery or judgment task, before an unexpected recall test for the items. Following both reading and imagery tasks, Richards and French (1991) assessed recall of emotionally threatening stimulus words (e.g., trapped, humiliated) and neutral words in high and low trait-anxious participants. Regardless of whether these words had simply been read or had been employed to evoke imagined scenes, no differences in relative recall separated high and low traitanxious participants. Using a rather different encoding task, Nugent and Mineka (1994) directed high and low trait-anxious participants to rate how much they liked or disliked presented threatening and nonthreatening words, then unexpectedly tested ability to recall these words. Although there was some evidence from an initial study to suggest relatively better recall of threat words in high trait-anxious participants, this effect failed to emerge in a second experiment, leading these researchers to conclude that trait anxiety is not reliably associated with a recall advantage for threatening information. In contrast, three articles have reported finding that the recall of negative words is enhanced for high trait-anxious participants (Eysenck & Byrne, 1994; Reidy & Richards, 1997; Russo, Fox, Bellinger, & Nguyen-Van-Tam, 2001). Under certain combinations of encoding and recall tasks, Eysenck and Byrne observed relatively facilitated recall of negative stimulus words in their high traitanxious participants, though correlational analysis failed to disentangle the potential mediating roles of anxiety and depression, both elevated within this sample. Similarly, following a self-descriptive or other-descriptive encoding task,
160
Reidy and Richards found that recall of self-encoded negative words was disproportionately good for high trait-anxious participants. However, this effect proved to be correlated with depression rather than with anxiety level. Russo et al. (2001) observed no differences in the patterns of recall shown by high and low trait-anxious individuals following a semantic encoding task that required participants to rate the pleasantness of emotionally toned stimulus words. They suggest that this finding could reflect a ceiling effect, given that the high level of overall recall left little scope for stimulus emotionality to influence memory performance. Consistent with this account, Russo et al. found greater evidence of an anxiety-linked recall bias following an alternative encoding task that required participants to make structural judgments about the stimulus words (i.e., counting the number of syllables). Although recall was poorer following this structural encoding task than in the semantic encoding task, threat words now were recalled disproportionately by the high trait participants. However, as the authors themselves acknowledge, this effect might be attributed to differential encoding and so does not require the conclusion that anxiety is associated with a biased retrieval process. The structural encoding task instructed participants to attend to the syllabic format of each word while ignoring its semantic content. Previous research convincingly demonstrates that high traitanxious individuals display an impaired ability to direct attention away from the semantic content of threatening stimuli, even when explicitly instructed to do so (e.g., Williams, Mathews, & MacLeod, 1996; Wood, Mathews, & Dalgleish, 2001). Certainly, despite instructions to process syllabic structure, Russo et al.’s (2001) simple task permitted ample time for the processing of semantic content, as each word was exposed for 4 seconds. Under these circumstances, one would predict on the basis of established research findings that high trait-anxious participants would attend disproportionately to threat word content. Thus, although the net result was one of better memory for threat words in anxious participants, compared to nonanxious individuals, it seems more parsimonious to regard this result as further evidence of selective encoding, rather than of an anxiety-linked bias in the memory retrieval process. Those investigators who have examined memory for emotional stimulus words in clinical populations have not obtained consistent evidence of a recall advantage for threatening material in patients suffering from anxiety disorders. In an early study, Mogg, Mathews, and Weinman (1987) required patients with GAD, and nonanxious control participants, to judge whether positive and negative words were either self-descriptive or other-descriptive. In a subsequent unexpected memory test, there was no group difference in relative ability to recall these two classes of stimulus words. Indeed, a subsidiary analysis carried out on the subset of negative words judged to be most threatening revealed that they actually were recalled disproportionately poorly by the anxiety patients. In a virtually identical experiment, reported by Mogg and Mathews (1990), GAD patients did display a nonsignificant trend toward relatively better
161
recall of negative words, compared to nonanxious control participants, across both encoding conditions. However, Mogg and Mathews noted that their GAD patients also generated a disproportionate number of negative words as intrusion errors while performing the recall task, suggesting that the observed effects may result from a negative response bias, rather than from any enhanced ability to retrieve the negative material from memory. Becker, Roth, Andrich, and Margraf (1999) had GAD participants and nonanxious controls rate the ease with which they could imagine scenes evoked by emotionally valenced words, some closely related to the usual concerns of GAD patients, before giving them an unexpected recall test for these words. No group differences in pattern of recall performance were observed, leading Becker et al. to conclude that GAD is not associated with any memory advantage for threatening words, even when they are relevant to disorder-related concerns. In a study of cued recall in GAD patients and nonanxious controls, Mathews, Mogg, May, and Eysenck (1989) used a similar encoding method but then provided three-letter word stems that participants were directed to complete to yield one of the previously exposed words. There were no differences between groups in the observed pattern of cued recall accuracy. Otto, McNally, Pollack, Chen, and Rosenbaum (1994) replicated this null result. As noted, Russo et al. (2001) suggested that the failure to find an anxietylinked memory bias in certain studies may reflect ceiling effects, which render the task insensitive to group differences in the ability to recall valenced stimuli. However, in a study comparing patients suffering from GAD and those with clinical depression, Bradley, Mogg, and Williams (1995) also failed to obtain evidence of any anxiety-linked recall bias, while nevertheless confirming their task’s sensitivity to individual differences in the selective recall of emotional materials. Participants first rated depression-relevant words (e.g., discouraged, despair), anxiety-relevant words (e.g., embarrassed, assaulted), positive words (e.g., paradise, bliss), and neutral words (e.g., bench, cereal) for their relevance to personal concerns. In an unexpected subsequent memory test, the depressed patients showed evidence of a recall bias favoring depression-related words. However, once again there was no difference between the recall profiles of GAD patients and nonanxious controls. In contrast, Friedman, Thayer, and Borkovec (2000) did observe better recall of threat words in GAD patients than controls when they used a less directive approach to initial stimulus processing. Specifically, threat and neutral words were presented for 8 seconds each, during which time the participant was required simply to read each item aloud. When memory for these words then was tested, GAD patients recalled more threat than nonthreat words and more threat words than did control participants. The initial word processing task used by Friedman et al. was similar, in potentially important ways, to that employed by Russo et al. This task permitted, but did not require, the processing of semantic content, and the exposure time was long. These conditions likely allowed participants ample time and freedom to encode the words in quite
162
different ways, so an anxiety-linked tendency to selectively encode the aversive semantic content of the threat words may underlie the enhanced recall of such materials shown by the anxious patients. Studies that have investigated the memorial characteristics of social anxiety have failed to yield evidence that it is associated with any recall advantage for threatening information. Sanz (1996) found no differences between socially anxious students and nonanxious controls in their recall of neutral, positive, and negative trait words previously encountered in a self-descriptiveness rating task, even when these words were closely related to social anxiety. When comparing memory for words appraised earlier by participants for emotional tone, Rapee et al. (1994) likewise found no difference between clinical social phobics and nonanxious controls in the relative recall of neutral words, positive words, and negative words related either to general threat or specifically to social threat. In a second study, Rapee et al. modified their encoding task, now requiring socially phobic patients and control participants to imagine a scene involving themselves and the referent of each stimulus word and then to rate this scene for pleasantness. Afterward, Rapee et al. tested cued recall, rather than free recall, by instructing participants to complete three-letter word stems to produce previously encountered words. Once again, however, no group difference in the relative recall of neutral, positive, and negative words emerged. Finally, in a third experiment designed to increase ecological validity, Rapee et al. provided positive and negative feedback concerning participant performance on an initial task (e.g., your material was poorly organized) and then unexpectedly tested recall of this emotional feedback. Compared to nonanxious control participants, social phobic patients displayed no facilitated recall of the negative information. Indeed, these patients actually recalled a disproportionately small number of the negative feedback statements. Cloitre, Cancienne, Heimberg, Holt, and Liebowitz (1995) and Lundh and Ost (1997) also examined the ability of socially phobic patients to recall emotional stimuli, using free and cued recall procedures, respectively, and found no evidence of any anxiety-linked memory bias. Becker et al.’s (1999) study, cited earlier, included a group of social phobic as well as GAD patients. These social phobic patients, like the GAD patients, did not differ from nonanxious control participants in terms of their relative ability to recall threat words associated with social concerns, general threat words, neutral words, and positive words. Amir, Coles, Brigidi, and Foa (2001) investigated whether practicing the recall of neutral, positive, or social threat words would reveal differences between socially phobic and nonanxious participants. Words initially were exposed together with a preceding category label, such as Fruit—Mango or Job—Stress. Subsequently, half of the participants were given cued retrieval practice, which involved presenting the category labels together with the initial fragment of the associated test word (e.g., Fruit—Ma- -), which had to be completed to produce the original item. Finally, all participants were given a cued recall test, within which they were shown parent category words and told to recall the words paired with them.
163
Without retrieval practice, the groups did not differ in their relative ability to recall the emotional categories of stimulus words. Retrieval practice did not elicit any relative recall advantage for negative words in the socially phobic patients. Indeed, the socially phobic patients gained less than controls from retrieval practice with socially threatening words. Collectively, therefore, these experimental studies provide no empirical support for the proposal that the recall of emotional information is enhanced in social phobia. Fewer experimental studies have investigated whether patients suffering from OCD or PTSD demonstrate preferential recall of threatening information. One problem that arises quite commonly in these studies is that supposedly emotional material may be neutral for control participants. For example, Radomsky and Rachman (1999) found better recall of “contaminated” than “clean” objects in OCD patients, relative to nonanxious control participants. However, contamination was manipulated by varying whether objects were touched by the experimenter, a distinction unlikely to affect the control group emotionally. Within a directed forgetting paradigm, Wilhelm et al. (1996) exposed OCD patients and control participants to neutral words (e.g., spoon, curtain), positive words, (e.g., laugh, confident), and negative words often associated with obsessive concerns (e.g., danger, disease). In this experimental approach, after each word has been presented, a signal directs the participant to either remember or forget that item. Wilhelm et al. found that when participants were instructed to remember words, the two groups demonstrated equivalent subsequent recall across all three classes of stimuli. However, for those items they were directed to forget, the OCD patients subsequent recalled the negative words more often than the controls did. Using a similar paradigm, McNally, Metzger, Lakso, Clancy, and Pitman (1998), exposed survivors of childhood abuse, with or without PTSD, to neutral words, positive words, and negative words related to their traumatic experience (e.g., molested, abuse). Their results were similar to those obtained by Wilhelm et al. using OCD patients. Participants did not differ in their patterns of recall for the words they had been instructed to remember, but for those they had been directed to forget, the PTSD patients recalled a disproportionate number of the negative items. Anxiety is associated with a reduced capacity to inhibit attention to threatening meanings during the encoding of stimulus words (Wood et al., 2001). Directing participants to forget a word, at the time of its initial exposure, means that they must try not to process the meaning of that stimulus, which is disproportionately difficult for anxious individuals when this meaning is emotionally threatening. Thus, the relative difficulty PTSD and OCD patients appear to experience, when trying not to process threat words under such “forget” instructions, may reflect this well-established encoding effect, rather than any anxiety-linked bias in the retrieval process itself. A related issue arises in a study by Vrana, Roodman, and Beckham (1995), who exposed Vietnam veterans with and without PTSD to neutral and threat-
164
ening words (e.g., firefight, death), within an emotional Stroop task (Williams et al., 1996). Consistent with much previous research, PTSD participants were disproportionately slow to color-name the threatening words during their initial exposure, an effect commonly attributed to anxious individuals’ difficulty in suppressing the processing of these words’ negative semantic content. Vrana et al. also found that the PTSD patients subsequently recalled a disproportionate number of the threatening words. However, because of the observed pattern of color-naming interference, these researchers quite reasonably ascribe this group difference to the PTSD patients’ tendency to selectively encode threat word meaning during the color-naming task, despite instructions to ignore word content. Thus, the few available studies of OCD or PTSD patients are either rendered problematic by the use of stimuli unlikely to be threatening for control participants or they invite explanation in terms of an anxiety-linked failure to inhibit the encoding of negative meanings, not in terms of an anxiety-linked bias in the retrieval of information from memory. In contrast, there is much more empirical support for the existence of a memory bias that favors the retrieval of threatening information in panic disorder patients. Of course, some of the relevant studies suffer from the methodological limitations already described, and null results sometimes have been obtained even in sound experimental designs. For example, in an early study of agoraphobia (now considered to be closely associated with panic disorder), Nunn, Stevenson, and Whalan (1984) assessed recall of words and short passages. Agoraphobic patients recalled more words and propositions from threatening passages than did nonanxious control participants. However, the threatening material described activities such as shopping and included words such as street, travel and cinema, which control participants probably would not consider emotionally negative. Furthermore, Pickles and van den Broek (1988) failed to replicate Nunn et al.’s original findings, despite employing a very similar experimental design. Otto et al. (1994) found no differences between the patterns of recall shown by panic disorder patients and nonanxious control participants for neutral, positive, general threat, and panic-related words that earlier had been rated for personal emotional significance. Likewise, Rapee (1994) did not observe any difference in the patterns of recall shown by panic disorder patients and nonclinical controls for differing classes of emotional words previously exposed within a free association task. However, these few negative reports are outweighed by a greater number of studies that have found that panic disorder patients demonstrate enhanced recall of threatening information. McNally, Foa, and Donnell (1989) required panic disorder patients and nonanxious control participants to rate anxietyrelated words (e.g., apprehensive, uptight) and other words unrelated to anxiety (e.g., charming, outgoing) for self-descriptiveness. In an unexpected subsequent memory test, control participants recalled fewer anxiety-related words than anxiety-unrelated words, whereas this pattern reversed in the panic disorder
165
patients. Contrary to the findings of Mogg and Mathews (1990), the emotional valence of intrusion errors did not differ between patients and controls, suggesting that these results reflect a genuine group difference in recall, rather than in response bias. Similar findings have been reported in studies that have employed different encoding tasks. For example, Lundh, Czyzykow, and Ost (1997) required participants to imagine scenes evoked by each of their stimulus words and then tested memory using a cued recall test that required three-letter word stems to be completed to make previously exposed words. Control participants recalled similar numbers of threat and neutral words, whereas panic disorder patients recalled a disproportionate number of physically threatening words (but not socially threatening words), with the magnitude of this recall advantage for threat predicted by a measure of anxiety sensitivity. Using the same encoding task with panic disorder patients and control group participants, Becker, Rinck, and Margraf (1994) observed no group difference in the subsequent recall of positive and generally negative words (e.g., lonely, brutal) but found that panic disorder patients did recall more of the negative words specifically related to panic (e.g., fainting, madness). Once again, such words did not appear more commonly among the intrusion errors made by panic patients, mitigating against a response bias explanation of the effect. In their subsequent research, Becker et al. (1999) included three types of panic-related words: somatic (e.g., palpitations, sweating), cognitive (e.g., helplessness, dying), and situational (e.g., tunnel, airplane). Panic disorder patients demonstrated enhanced recall of somatic and situational threat words but not of cognitive threat words, without showing increased representation of such items among their intrusion errors. Cloitre and Liebowitz (1991) employed two quite different encoding tasks to expose panic disorder patients and nonclinical controls to neutral, positive, and threat words. One task required participants to decide whether each item described a feeling state, whereas the other task required them to decide only whether the item was a legitimate English word. The nonanxious group subsequently recalled a similar number of words from all three emotional categories, whereas the panic disorder patients recalled more threat words than either positive or neutral words, regardless of encoding task. Again, the patterns of intrusion errors were equivalent across groups, suggesting that the observed recall effects could not be attributed to a group difference in response bias. Cloitre, Shear, Cancienne, and Zeitlin (1994) used another encoding task variant, in which panic disorder and control participants assessed the semantic association between members of word pairs, which could be neutral (e.g., gauge-metric), positive (e.g., smiles-elated), or panic-related (e.g., dizzy-faint). An unexpected cued recall test then was given, with participants shown the first word of each pair followed by the first three letters of the second word, which they were required to complete to yield the original item. Control participants displayed
166
equivalent recall of the classes of emotional words, but panic disorder patients recalled a disproportionate number of the panic-related threat words. In summary, then, although recall studies have yielded little compelling experimental evidence to support a negative retrieval bias among individuals who report high levels of trait anxiety, or in patients suffering from most clinical anxiety disorders, they have provided firmer grounds for believing that such a retrieval bias may characterize panic disorder. The few studies that have demonstrated a recall advantage for threat stimuli among other anxious populations typically have tested memory following encoding tasks that have either instructed participants to ignore word content or permitted individuals considerable freedom to choose how they process stimulus content during encoding. When, instead, participants have been directed to process the semantic content of stimuli in a specified manner during encoding, typically no anxiety-linked bias in recall performance has been observed. This pattern of findings suggests that the occasionally observed superior recall of threatening words by such anxious participants likely results from their established tendency to selectively process threat stimulus content during encoding rather than from any bias in the retrieval process itself. The same is not true, however, for panic disorder patients, for whom encoding instructions appear to have little impact on observed patterns of selective recall. Across a wide range of studies, varying in both encoding and recall methodology, panic disorder patients repeatedly have displayed a recall bias favoring threatening stimuli, suggesting that the retrieval of such information may indeed be facilitated within this particular anxious population. We now will turn to the consideration of experiments that have employed measures of recognition, rather than recall, to test for selective memory bias in anxious participants. Such studies continue to provide little evidence that the retrieval of threatening information is generally enhanced across differing populations of anxious participants.
Recognition Memory Recognition memory differs from recall; for example, it does not need to engage the search component of memory retrieval that characterizes recall, which may make recognition measures less sensitive to emotional biases that exert their primary influence on this search process. However, recognition tasks have the advantage of permitting assessment of memory for types of stimuli that do not readily fit within recall tasks, such as pictures of emotional faces. Furthermore, recognition data can be subjected to signal detection analysis, allowing a distinction between measures of memory sensitivity (d’) and response bias (B). Across two experiments, Nugent and Mineka (1994) assessed high and low trait-anxious participants’ recognition memory for threat and nonthreat words, which previously had been rated for likeability, and subjected their data to signal detection analysis. In one of these studies, high trait individuals were found
167
to demonstrate a response bias favoring threat words. However, in neither of the experiments was any difference observed between the groups in their relative memory sensitivity for the two classes of stimulus words. Similarly, Dalgleish (1994) found no evidence from signal detection analysis to indicate that high and low trait-anxious participants differed in relative ability to accurately recognize neutral, positive, or anxiety-related words that earlier had been encountered as solutions to an anagram task. A number of researchers have compared the ability of GAD patients and nonanxious controls to recognize threat and nonthreat words previously presented within an emotional Stroop task. Using signal detection analysis, both Mathews and MacLeod (1985) and Mogg, Mathews, and Weinman (1989) failed to find any group difference in the sensitivity of recognition memory for these two word types. However, recognition performance was generally poor in these studies, probably because the encoding task invited little processing of word content. To counteract this limitation, MacLeod and McLaughlin (1995) modified the initial emotional Stroop task, by requiring participants to read each word aloud after it had been color-named. This did indeed improve recognition memory for the stimulus words, but still GAD patients and nonanxious controls did not differ in relative recognition memory for the threat and nonthreat words. This failure to observe an anxiety-linked recognition memory advantage for threat words extends to other studies that have employed quite different encoding tasks. Mogg et al. (1987) examined recognition memory for negative and positive trait adjectives that previously had been rated by participants for selfdescriptiveness or other-descriptiveness. Patients suffering from GAD showed no relative enhancement of recognition memory for threat words, compared to nonanxious control participants. Indeed, the anxiety patients tended to recognize fewer threat words, and more positive words, than the nonanxious participants did, although this group difference in recognition sensitivity was not statistically significant. Rapee et al. (1994) included recognition measures to assess memory for words that participants previously had rated for likeability but observed no differences between social phobics and controls in ability to recognize threat and nonthreat words. Cloitre et al. (1995) also found equivalent patterns of recognition memory for emotional words in social phobics and nonanxious control participants. Lundh and Ost (1996a) assessed recognition memory for images of faces rather than word stimuli. These researchers required socially phobic patients and nonanxious controls initially to classify faces according to whether they would anticipate a high- or low-quality interaction with the person shown. Subsequently, both groups of participants demonstrated an equivalent ability to recognize the faces, regardless of the classification imposed. However, in a second study, Lundh and Ost (1996b) had participants rate the faces as either critical or accepting. In a later memory task, the social phobics, unlike the control participants, recognized a disproportionate number of those faces they earlier had rated
168
as critical. Although this may suggest that social phobics display a recognition memory advantage for emotionally negative faces, there are other possible interpretations. Because the experimenters did not directly manipulate the emotional valence of the faces, their findings may reflect anxiety-linked differences in rating task performance, rather than in recognition memory performance. For example, social phobics may have been disproportionately inclined to judge the more perceptually distinctive faces as critical. Even if all participants then find these distinctive faces easier to subsequently recognize, to the same degree, this rating bias would lead to the social phobics recognizing a disproportionate number of the faces they deemed critical, though to construe this difference as a group difference in selective memory would be misleading. This problem was circumvented by Mansell, Clark, Ehlers, and Chen (1999), who presented social phobics and nonanxious individuals with equivalent arrays of differing faces, some displaying neutral, some happy, and some negative emotional expressions (anger, disgust, fear, or sadness). Signal detection analysis revealed no subsequent group differences in relative recognition memory sensitivity for faces with differing emotional expressions. Thus, to conclude that social phobics display a recognition memory advantage for negative faces would be premature. Similarly, measures of recognition memory for neutral and negative words do not reveal differences between nonanxious controls and participants suffering from OCD, PTSD, or phobic states. Using intentional memory instructions, Wilhelm et al. (1996) found that OCD patients and controls display an equivalent pattern of recognition memory for neutral, positive, and negative stimulus words. Even when textual stimuli have been employed, tailored to describe events closely related to typical OCD concerns, these anxiety patients do not demonstrate enhanced recognition memory for such emotionally negative information. Foa, Amir, Gershuny, Molnar, and Kozak (1997) presented OCD patients and nonanxious individuals with neutral sentences and others related to contamination concerns (“they soiled their clothes with blood”). Both groups subsequently displayed an equivalent pattern of recognition memory, with all participants recognizing the neutral sentences better than the contamination sentences. Vrana et al. (1995) exposed Vietnam veterans with and without PTSD to neutral, negative, and trauma-related words within an emotional Stroop task. Subsequent recognition memory was better for the negative than for the neutral words, but the observed pattern did not differ between the anxiety patients and the control participants. Even when anxious participants are characterized by specific fears, they still do not generally show enhanced recognition memory for stimuli associated with these fears. Watts, Trezise, and Sharrock (1986) had spider-phobics and nonanxious controls look at a variety of dead spiders and then tested their capacity to subsequently recognize them. The phobics showed no overall superiority in recognition memory performance. Indeed, for the larger spider, which presumably represented the more highly feared stimuli, the phobics actually demonstrated poorer recognition memory than control participants did.
169
We concluded earlier that good evidence indicates that panic disorder patients display relatively enhanced recall of emotionally negative material, yet such patients appear to show no parallel recognition memory advantage for negative information. Ehlers, Margraf, Davies, and Roth (1988) gave panic disorder patients and control participants a recognition memory test for words that had been presented within an earlier emotional Stroop task. Signal detection analysis revealed that threat words were recognized more accurately than neutral words, but this effect was equivalent across both patient and control groups. Beck, Stanley, Averill, Baldwin, and Deagle (1992) assessed recognition memory, in panic disorder patients and control participants, for previously exposed neutral and emotional words. Overall, they found greater recognition memory sensitivity for socially threatening words than for neutral words, but once again this pattern did not distinguish the two groups. Across two experiments, Lundh, Thulin, Czyzykow, and Ost (1998) assessed recognition memory in panic disorder patients and nonanxious controls, using the face recognition methodology previously employed by Lundh and Ost (1996b) to test social phobics. There were no group differences in relative recognition memory for faces participants previously had classified as accepting or critical or for faces they had classified as high or low in degree of desired contact. Only when participants initially had classified faces according to whether they would consider the displayed person to be safe or unsafe were different patterns of subsequent recognition memory observed between the patients and the controls. However, the nature of this effect was that the panic disorder patients displayed disproportionately poor recognition memory for those faces that they previously rated unsafe. Clearly, therefore, these experiments provide no support for the proposal that recognition memory for negative information is enhanced in panic disorder patients. Only one study has found evidence to support the existence of such a bias, and it employed a rather unusual measure of recognition memory. Cloitre and Liebowitz (1991) first exposed neutral, positive, or threatening words within a task that required panic disorder patients and control participants to judge whether the item described a feeling state or simply whether it was a legitimate English word. Participants subsequently were given a high-speed recognition memory task adapted from Jacoby and Dallas (1981). Previously seen and new words were displayed for only 35 ms each, and participants made a forced–choice decision concerning whether the word had been encountered earlier in the experiment. Although such a brief exposure likely restricted awareness of these test stimuli, Cloitre and Liebowitz nevertheless contend that their task assessed explicit memory, because participants were directed to identify which words they had previously seen. There was a trend toward greater recognition memory sensitivity for threat words in the panic disorder group alone, but this effect reached significance only when performance on negative and positive words was directly compared, after removing neutral word data from the analysis.
170
Overall, research on recognition memory has provided little evidence that anxiety is associated with enhanced recognition memory for negative information. Most well-designed experiments have yielded null results. Only the unusual recognition task Cloitre and Liebowitz (1991) used has offered limited support for the hypothesis, though the effect was significant only when an experimental condition was excluded within a post-hoc subsidiary analysis, so whether the finding will prove replicable remains uncertain. Nevertheless, the best evidence for an anxiety-linked recognition memory advantage for negative information comes from a task that imposed a substantial restriction on awareness. We turn now to the consideration of experiments designed specifically to assess implicit memory performance. The results of such studies have been very variable, and although they have yielded somewhat greater evidence for such an anxietylinked bias than explicit memory tasks, the phenomenon nevertheless remains elusive.
Implicit Memory Whereas recall and recognition tests require participants to consciously recollect previously encountered information, implicit tests infer memory indirectly, by examining how prior exposure to stimulus items influences later task performance. For example, when asked to complete word stems with the first items that come to mind, people are disproportionately likely to generate words they have seen recently (Richardson-Klavehn & Bjork, 1988). Similarly, people are more likely to accurately identify a briefly exposed word if this word has been recently encountered (Jacoby & Dallas, 1981). Evidence that measures of implicit memory, yielded by such indirect tests, can usefully be distinguished from the more common recall and recognition measures of explicit memory has come from work on amnesic patients, who typically show impaired performance on explicit memory tasks without demonstrating corresponding deficits on implicit memory tasks (Jacoby & Witherspoon, 1982). In a study that compared the performance of high and low trait-anxious participants on several implicit tests, Russo, Fox, and Bowles (1999) failed to find evidence that anxiety is associated with enhanced implicit memory for threatening words. No group differences were observed in the impact exerted by previously exposed emotional words (which had been simply read or used within an imagery task) on either subsequent word fragment completion or on the perceptual identification of threat and nonthreat words. However, though null results of this type have not been uncommon, the pattern of findings has been highly inconsistent across studies, and a number of researchers also have presented evidence to support anxiety-linked bias on implicit memory tasks.1 Nugent and Mineka (1994) initially presented high and low trait-anxious participants with emotional words they were to rate for likeability and then assessed implicit memory for this information using a word stem completion task.
171
As evidence of implicit memory, more of these stems were completed to yield previously exposed (primed) words than alternative (nonprimed) words. However, the relative magnitude of this effect, across threat and nonthreat words, did not differ between the groups. In contrast, using this same word stem completion task, Richards and French (1991) did find a different pattern of implicit memory for threat words and nonthreat words in high and low trait-anxious individuals, though only when these words initially had been processed in a selfreferential manner during their initial exposure. For high trait-anxious participants, but not for low trait-anxious individuals, the priming measure of implicit memory was significantly greater for threat words than for non-threat words. Eysenck and Byrne (1994) also used a stem word completion task to compare high and low trait-anxious participants’ implicit memory for emotional words, encountered earlier in an encoding task that required either simple reading or the generation of the words from their definitions. Regardless of encoding task, Eysenck and Byrne observed that the priming index of implicit memory was greater for threat than for neutral words in high trait-anxious individuals but not in either medium for low trait-anxious participants. Mathews et al. (1989) found evidence that GAD patients also display disproportionately good implicit memory for threatening words. They presented neutral, positive, and threatening words to these anxiety patients, and to nonanxious controls, within an initial encoding task that required participants to generate a self-referent image using each stimulus item. Following this, implicit memory was assessed by measuring the magnitude of the priming effects exerted by these stimuli on a subsequent word stem completion task. Control participants evidenced significantly smaller priming effects for the threat words than for the nonthreatening words, whereas GAD patients showed slightly greater priming effects for the threat words than for the nonthreatening words. Although this pattern of findings is consistent with the presence of an anxiety-linked implicit memory advantage for threatening information, the effect has not replicated consistently across other studies of GAD patients, possibly because of variations in the encoding tasks employed. Mathews, Mogg, Kentish, and Eysenck (1995) exposed GAD patients and controls to word sets similar to those used by Mathews et al. (1989), but instead of instructing participants to generate images from these words, they asked them to count occurrences of the letter e in each word. Under these encoding conditions, there was no tendency for the GAD and control participants to differ in the relative degrees to which the threatening and nonthreatening words primed a later word stem completion, suggesting no differential pattern of implicit memory between the groups. Bradley et al. (1995) also employed fairly superficial encoding tasks, requiring participants to judge personal usage frequency of words or simply exposing them very briefly (14 ms), before inferring implicit memory for these word from the magnitude of repetition priming effects observed on a subsequent lexical decision task. These researchers, too, failed to observe any difference between GAD patients and control participants
172
in relative implicit memory for neutral, positive, and anxiety- or depressionrelevant words. Nevertheless, MacLeod and McLaughlin (1995) did find evidence of an implicit memory advantage for threat words in GAD patients, without using a self-referential encoding task. GAD patients and controls were presented with threat and nonthreat words in an initial emotional Stroop task and were instructed first to name the ink color and then to pronounce each word aloud. Subsequently, these words, together with matched new words, were exposed for a duration calibrated to permit around 70% accurate identification, and implicit memory was inferred from participants’ increased ability to identify the old words relative to the new words. GAD patients, unlike nonanxious control participants, showed evidence of greater implicit memory for the threat words than for the neutral words. MacLeod and McLaughlin concluded that enhanced implicit memory for negative information is a replicable, even if not entirely reliable, characteristic of GAD patients. The results of studies designed to investigate implicit memory bias in social phobics also have been inconsistent. Using a word stem completion task to assess implicit memory for neutral and social threat words, previously encountered in a self-referential imagery task, Rapee et al. (1994) found no differences between the patterns of performance by social phobics and by nonclinical controls. However, using this same word stem completion procedure to assess implicit memory for emotionally toned stimuli following a similar encoding task, Lundh and Ost (1997) obtained more encouraging findings. Despite the absence of a significant interaction of group by word type, hypothesis-driven tests revealed a trend toward disproportionately strong priming effects for social threat words in the social phobics, compared to the control participants, consistent with an implicit memory advantage for such material. Amir, Foa, and Coles (2000) used a different procedure to assess implicit memory, based on Jacoby, Allan, Collins, and Larwill’s (1988) finding that background noise is judged to be quieter when familiar material is embedded within it. Amir et al. first had social phobics and nonanxious controls listen to, and repeat aloud, emotionally toned sentences. Later, the participants heard sentences, some previously presented and some new, embedded within white noise. Implicit memory was revealed by their tendency to rate the white noise intensity as lower when this embedded sentence had been exposed earlier. For social phobics, but not for control participants, the perceived intensity of this white noise was disproportionately low when it accompanied previously presented sentences with threatening emotional content. This finding was taken as evidence that implicit memory for such threatening sentences was enhanced in these anxious patients. An insufficient number of studies have examined implicit memory for emotional information in either OCD or PTSD patients to permit clear conclusions. Only one such experiment has been carried out on each of these populations, and each used the white noise paradigm to assess implicit memory. Foa et al. (1997) exposed contamination-fearful OCD patients to contamination-related and neu-
173
tral sentences before presenting these same sentences and matched new sentences embedded in white noise. Perceived noise intensity was lower when the noise accompanied old sentences, confirming implicit memory for this material. However, the relative magnitude of this effect across both emotional categories of sentences was equivalent for OCD patients and nonanxious individuals. In contrast, when Amir, McNally, and Wiegartz (1996) used the same approach to assess implicit memory for neutral and for combat-related sentences in Vietnam veterans with and without PTSD, evidence of a group difference was obtained, though only when white noise volume was high. Veterans with PTSD, unlike those without, judged subjective noise intensity as lower when it accompanied the old combat-related sentences. Thus, though only a few studies have addressed the issue, no evidence of an implicit memory advantage for negative information yet has been found in OCD patients, though some evidence indicates that such a bias may characterize PTSD patients. Amir, McNally, Riemann, and Clements (1996) used this same white-noise methodology to assess implicit memory for emotional sentences in panic disorder patients and control participants. Old and new sentences were embedded in a low level of white noise, and panic disorder patients demonstrated disproportionate attenuation of subjective noise intensity when it accompanied old sentences describing panic-related threat scenarios. Although this finding is consistent with an implicit memory advantage for threatening information in panic disorder, some studies using other measures of implicit memory to assess this same population of clinical patients have failed to support this conclusion. Rapee (1994) used the word stem completion task to measure implicit memory for neutral and threat words, initially exposed within a word association task, and found no difference in the patterns of performance shown by panic disorder patients and nonanxious control participants. Similarly, Lundh et al. (1997) failed to find any difference between panic disorder patients and controls on a word stem completion measure of implicit memory for such emotional words, initially encoded in a self-referential imagery task. In a later experiment, Lundh, Wikstrom, Westerlund, and Ost (1999) employed a perceptual-identification measure to assess implicit memory for these emotional words, following their initial exposure in an emotional Stroop task. Perceptual identification was facilitated by prior exposure, confirming the presence of implicit memory, but the magnitude of this effect was equivalent across all three classes of words for both panic disorder patients and control participants. Nevertheless, Amir et al.’s claim that implicit memory for threat is enhanced in panic disorder patients gains some support from a study by Cloitre et al. (1994). In this experiment, emotional words first were presented in pairs, and participants rated the strength of semantic association between items. They later were shown the first word from each pair, together with the first three letters of the second word, and were instructed to complete these stems to make the first words that came to mind. A measure of memory was provided by degree to which the previously exposed words,
174
rather than alternatives, appeared in these completions. Cloitre et al. observed that, among panic disorder patients, this priming effect was disproportionately great for threat words and took this result as evidence of an implicit memory advantage for threatening information in patients with this anxiety disorder. As shown in this review, the overall pattern of implicit memory findings is not yet sufficiently consistent to sustain firm conclusions. There are many possible reasons for this inconsistency, including the potential impurity of the memory measures. Though implicit memory tasks are designed to be sensitive to the influence of implicit memory, performance on most implicit memory tasks also can be influenced by explicit memory (Perruchet & Baveux, 1989), and the degree to which observed effects reflect the impact of each type of memory may depend on subtle aspects of the experimental procedure. For example, the fact that word stems in Cloitre et al.’s (1994) study were accompanied by another word, which earlier had been paired with an item that could complete this stem, might encourage the use of explicit memory to identify this candidate completion. Thus, one might plausibly contend that Cloitre et al.’s finding could reflect the influence of enhanced explicit memory, rather than enhanced implicit memory, for threatening information in panic disorder. However, with the exception of panic disorder, evidence of an anxiety-linked explicit memory advantage for threat is conspicuously lacking. Therefore, it would seem unreasonable to argue that the more common appearance of such an anxiety-linked bias on implicit memory tasks should be attributed to the contaminating influence of explicit memory. Indeed, contamination from explicit memory might reduce the capacity of these tasks to detect an anxiety-linked bias in implicit memory. If so, perhaps task refinements that increase the purity of the resulting implicit memory measures may result in more consistent evidence of such a bias. For the moment, however, we must concede that, whereas a greater proportion of implicit memory tasks than explicit memory have supported the idea that memory for threatening information may be enhanced in anxious individuals, neither approach has provided sufficiently reliable evidence to confirm the validity of this hypothesis.
Theoretical Implications The most obvious conclusion we draw from this review is that, from the wide range of studies that now have examined memory performance in high traitanxious groups and in patients suffering from most anxiety disorders, not enough evidence has been obtained to support the existence of a memory bias that consistently favors the retrieval of threatening information. Despite the wellestablished fact that both clinically anxious patients and high trait-anxious members of the normal population selectively direct attentional resources toward such
175
information during encoding (cf. Mathews & MacLeod, 1994), they do not reliably display an enhanced ability to subsequently access this type of material from memory. Such findings contrast markedly with those observed in depressed participants, who do indeed commonly display disproportionately good recall of negative self-related information (e.g., Bradley et al., 1995). [See chapter 6— Eds.] Within the anxiety disorders, such a pattern of selective recall appears only in panic disorder. For this condition alone, the majority of experimental studies have supported the existence of an explicit recall bias favoring the retrieval of threatening or panic-related information.
Why Is There So Little Evidence
of a General Anxiety-Related
Memory Bias?
Elsewhere, we have suggested two possible reasons why anxiety-related memory advantages for threatening information typically are not observed (e.g., Mathews & MacLeod, 1994). First, although anxious individuals selectively attend to threat cues during encoding, this information may not be processed in a manner that produces representations that can readily be accessed during memory search. The success of such a search process likely depends highly on the degree to which stimulus information has undergone elaborative processing, thereby establishing associations between the new to-be-remembered information and prior knowledge (Graf & Mandler, 1984). If the aversive nature of threatening stimuli motivates highly anxious individuals to rapidly detect them, but also to limit their further processing, such elaboration will be restricted, countering any subsequent retrieval advantage that otherwise would result from increased attention (Eysenck & Mogg, 1992; Williams, Watts, MacLeod, & Mathews, 1997). Although many studies have instructed participants to process stimulus information during the encoding task, in ways that might counter such avoidance (such as relating word meanings to the self), highly anxious individuals may fail to comply fully with such instructions, perhaps to avoid a sustained aversive experience. There is little direct evidence to support the idea that anxious individuals avoid elaboratively processing threatening stimuli. Indeed, the widely recognized tendency for anxious individuals—particularly those suffering from GAD—to worry excessively suggests that extensive elaborative processing of threat occurs in some anxiety disorders, especially in the form of verbal propositions (Borkovec & Inz, 1990). One could argue that this type of elaboration would aid the recall of such information. However, elaborative processing in the form of worry may exert no beneficial impact on recall performance, particularly if the worry involves stereotypical themes. Instead, if different threat cues elicit similar worry content in anxious individuals, the distinctiveness of the representations resulting from this type of elaboration could be reduced, compromising retrieval of the original cues.
176
The robust tendency to display relatively facilitated recall of negative information in depression often has been attributed to processing that establishes novel linkages with other self-related negative information, already represented within depressives’ memory. Depressive individuals typically demonstrate such a negative memory bias only for words encoded in tasks that encourage selfreferential processing (Bradley & Mathews, 1983). If depression alone is associated with particular negative ideation about the self, the apparent difference in the patterns of memory selectivity demonstrated by depressed and by anxious participants might be explained. For depressed individuals, the elaborative processing of new negative information links it to existing negative self-schema, which then functions as a retrieval aid during memory search. In contrast, although anxious individuals worry about future events, there is little evidence to suggest that they habitually ruminate about their own negative qualities (as do depressives), so there is no compelling reason to expect anxious individuals to display a negative recall bias following such commonly used self-referential encoding procedures (e.g., Mogg et al., 1989). Another reason that could explain the lack of an anxiety-linked recall bias is that anxious individuals may tend to represent threatening information perceptually rather than in a conceptual form based on verbal propositions. This contention is consistent with the supposition that threat can be processed within a primitive fear system (or module) that evolved prior to the development of language (Öhman & Mineka, 2001). If anxious participants are particularly inclined to represent threat using perceptual codes, perhaps because this primitive fear system is overly active, then according to the transfer-appropriate processing accounts of memory (Roediger & Blaxton, 1987), retrieving this information when performing memory tasks that encourage conceptual processing, such as free recall, would prove difficult for them. In contrast, memory tasks that encourage perceptual processing should more effectively tap anxious participants’ representations of previously encountered threat stimuli. This may explain why conventional implicit memory tasks, such as rapid word identification, which characteristically encourage perceptual rather than conceptual processing, have more often revealed evidence of an anxiety-related memory advantage for threat. According to this account, the widespread tendency to use self-referential or other semantic encoding procedures may reduce the likelihood of finding such biases on this type of implicit memory task. Certainly, as we have shown, the evidence in support of an anxiety-linked implicit memory advantage for threat is very mixed, but it may not be fully explained by variations in encoding tasks. Several studies that have used structural encoding methods have failed to reveal any implicit memory advantage for threat in anxious participants, though others have yielded more supportive findings (e.g., MacLeod & McLaughlin, 1995). Perhaps the prevailing use of words as stimuli also may undermine the likelihood of finding enhanced perceptual memory for threat in anxious individuals, as the perceptual encoding of such stimuli seems less likely than would be ex-
177
pected with pictorial materials. Though such a proposal is speculative at this stage, both the restricted elaboration hypothesis and the perceptual encoding hypothesis remain viable explanations for the absence of any general memory advantage for threat stimuli in anxious individuals, despite their tendency to selectively allocate attention to such information. Neither account, however, has been directly supported by experimental evidence; indeed, few experiments have been designed to test these hypothetical accounts.
Why Is There a Memory Bias
in Panic Disorder?
Despite the sparse evidence that anxiety is associated with a memory bias favoring threat, panic disorder patients tend to recall more panic-related threat words than nonanxious controls do. We cannot attribute this effect to anxiety per se, given the absence of parallel effects in other anxiety disorders, so it seems necessary to explain why panic disorder is an exception to the rule. One obvious possibility, already noted in our comments on the study by Nunn et al., (1984), is that the supposedly negative words in these studies often may be threatening only for the panic patients. For example, words describing common cues for the panic attacks experienced by such patients (e.g., street, crowd) frequently have been used in these experiments, even though such words are unlikely to be emotionally charged for nonanxious control participants. As we reported in the introduction, emotional information is more likely to be recalled than neutral information, regardless of anxiety level, and the finding that panic patients recall these words disproportionately well may reflect only this general effect. The confound between clinical status and stimulus valence makes it difficult to determine whether the panic disorder patients and the controls would have recalled differently stimulus materials that both groups had considered threatening. This problem is less evident within studies that have used stimulus materials likely to be emotionally threatening for all participants. Words such as torture and death presumably would have similar threat values for all readers, so any general recall advantage for emotional words would not produce group differences in recall. Nevertheless, as we have pointed out, likely group differences in the perceived threat value of stimulus materials have compromised at least one study of OCD patients (Radomsky & Rachman, 1999). In this study, discrepant patterns of memory performance arose perhaps because a contamination manipulation increased stimulus threat value for patients, but not for nonanxious controls. The more common finding is that OCD patients do not differ from nonanxious controls in memory for threatening stimuli, even when these are related to their concerns, although they do often report less confidence in their memory (e.g., Tolin et al., 2001). Thus, the anomalous memory advantage for threat-related words shown by panic disorder patients may reflect the fact that panic-related word cues possess more emotionally negative meaning for panic
178
patients than for nonanxious controls. The use of stimulus words that have special emotional significance only for one group of participants prevents the attribution of observed group differences in the recall of such material to a tendency to selectively remember emotional information.
Are Discrepancies in Memory Effects
Due to Differing Encoding Tasks?
We have suggested that some of the memory effects shown by anxious participants can readily be attributed to anxiety-linked patterns of selective encoding, without the need to implicate a bias in the retrieval process itself. For example, in the experiments carried out by Friedman et al. (2000) and by Russo et al. (2001), words were exposed for durations that far exceeded the minimum time required to register their identity. Anxious individuals characteristically direct attention toward, and find it disproportionately difficult to disengage attention from, emotionally threatening words during such encoding procedures. Consequently, their enhanced subsequent ability to recall the threatening words can be attributed to the superior encoding of the threatening information, resulting from this reduced ability to inhibit attention to such affective meanings (Williams et al., 1996; Wood et al., 2001). Further evidence to support this account comes from a series of experiments reported by Pury and Mineka (2001). In a typical study, participants were required to judge, as quickly as possible, whether a presented word (e.g., bleeding, confidence) described something dangerous or safe or whether it described a physical or psychological state. When participants were not informed which decision was required until immediately prior to word presentation, highly anxious participants were slower than low anxious controls to make the physical versus psychological judgment but not to make the dangerous versus safe judgment. This finding, together with the results of other experiments from this same series, provides evidence that high anxious individuals are more likely than their low anxious counterparts to encode stimuli in terms of their affective meaning. If anxious individuals’ infrequently observed threat-related bias in recall results from their heightened inclination to encode the emotionally threatening content of affective stimulus words, then this recall bias should be particularly likely following encoding tasks that permit ample opportunities for participants to select how to process these stimuli. Under such encoding conditions, we suppose that increased attention paid to threatening aspects of meaning may help high anxious individuals to retain such words better. In contrast—again assuming that the enhanced memory for threat occasionally demonstrated by anxious participants reflects an anxiety-linked tendency to process the affective meaning of such stimuli during encoding—this effect should not be observed following encoding tasks that require all participants to process stimuli in terms of their affective meaning. In general, as this review has indicated, anxious individuals more often show en-
179
hanced relative recall of threatening information following encoding tasks that permit but do not require the encoding of affective meaning. Technically, of course, the term “memory bias” is appropriate to describe the recall advantage for threat sometimes shown by anxious participants following free encoding conditions that employ long exposure durations (e.g., Friedman et al., 2000; Russo et al., 2001). However, we contend that it is not parsimonious to ascribe the effect to a bias in the process of retrieving information from memory, when it can be explained by the well-established tendency for highly anxious individuals to attend to and selectively encode threatening meanings. We believe that the effect may better be construed as an indirect consequence of the bias that favors the selective encoding of threatening stimulus content in anxiety. Nevertheless, the possibility that the patterns of selective attention and encoding associated with vulnerability to anxiety may, under certain circumstances, give rise to long-term memory effects is clearly important. These patterns of encoding selectivity may affect not only how old events are recalled but also how new events are understood. For example, our work on the experimental manipulation of such biases has shown that people trained to attend selectively to threatening meanings of stimulus information become more likely to store negatively disambiguated representations of newly presented ambiguous events in memory (Mathews & MacLeod, 2002). Distinctions between attention, interpretation, and memory thus can become blurred when life events are processed, given that initial encoding biases so readily can translate into differential patterns of interpretation and recall. In closing, however, we emphasize that, despite this potential capacity for encoding biases to influence memory performance, most experimental studies have failed to demonstrate that anxious individuals (with the possible exception of panic disorder patients) consistently display enhanced recall or recognition memory for emotionally threatening information. Though an implicit memory bias favoring threat stimuli sometimes has been observed in anxious participants, at best it is a fragile effect that often has proven difficult to replicate. It remains to be seen whether the future refinement of experimental methodologies will identify particular conditions under which anxious participants will more reliably demonstrate a memory advantage for negative information. For the moment, the most appropriate conclusion is that such a memory bias does not represent a robust general characteristic of anxiety.
Note 1. Whereas methodological rigor has varied across these studies, and it sometimes is possible to criticize experimental decisions such as choice of appropriate baselines, the pattern of observed inconsistences cannot readily be explained in terms of such design variations.
180
References Amir, N., Coles, M. E., Brigidi, B., & Foa, E. B. (2001). The effect of practice on recall of emotional information in individuals with generalized social phobia. Journal of Abnormal Psychology, 110, 76–82. Amir, N., Foa, E.B., & Coles, M. (2000). Implicit memory bias for threat-relevant information in individuals with generalized social phobia. Journal of Abnormal Psychology, 109, 713–720. Amir, N., McNally, R. J., Riemann, B. C., & Clements, C. (1996). Implicit memory bias for threat in panic disorder: Application of the “white noise” paradigm. Behaviour Research and Therapy, 34, 157–162. Amir, N., McNally, R. J., & Wiegartz, P. S. (1996b). Implicit memory bias for threat in posttraumatic stress disorder. Cognitive Therapy and Research, 20, 625–635. Ashcraft, M. H., & Kirk, E. P. (2001). The relationship among working memory, math anxiety, and performance. Journal of Experimental Psychology: General, 130, 224–237. Baddeley, A. D. (1986). Working memory. Oxford: Oxford University Press. Beck, J. G., Stanley, M. A., Averill, P. M., Baldwin, L. E., & Deagle, E. A. (1992). Attention and memory for threat in panic disorder. Behaviour Research and Therapy, 30, 619–629. Becker, E., Rinck, M., & Margraf, J. (1994). Memory bias in panic disorder. Journal of Abnormal Psychology, 103, 396–399. Becker, E. S., Roth, W. T., Andrich, M., & Margraf, J. (1999). Explicit memory in anxiety disorders. Journal of Abnormal Psychology, 108, 153–163. Borkovec, T. B., & Inz, J. (1990). The nature of worry in generalized anxiety disorder: A predominance of thought activity. Behaviour Research and Therapy, 28, 153–158. Bradley, B. P., & Mathews, A. (1983). Negative self-schemata in clinical depression. British Journal of Clinical Psychology, 22, 173–181. Bradley, B. P., Mogg, K., & Williams, R. (1994). Implicit and explicit memory for emotional information in non-clinical subjects. Behaviour Research and Therapy, 32, 65–78. Bradley, B. P., Mogg, K., & Williams, R. (1995). Implicit and explicit memory for emotion-congruent information in clinical depression and anxiety. Behaviour Research and Therapy, 33, 755–770. Burke, M., & Mathews, A. (1992). Autobiographical memory and clinical anxiety. Cognition & Emotion, 6, 23–35. Cahill, L., & McGaugh, J. L. (1995). A novel demonstration of enhanced memory associated with emotional arousal. Consciousness and Cognition, 4, 410–421. Calvin, A. D., Koons, P. B., Bingam, J. L., & Fink, H. H. (1953). A further investigation of the relationship between manifest anxiety and intelligence. Journal of Consulting Psychology, 19, 280–282. Christianson, S-Å. (1992). Emotional stress and eyewitness memory: A critical review. Psychological Bulletin, 112, 284–309. Cloitre, M., Cancienne, J., Heimberg, R. G., Holt, C. S., & Liebowitz, M. R. (1995). Case histories and shorter communications: Memory bias does not generalize across anxiety disorders. Behaviour Research and Therapy, 33, 305–307.
181
Cloitre, M., & Liebowitz, M. R. (1991). Memory bias in panic disorder: An investigation of the cognitive avoidance hypothesis. Cognitive Therapy and Research, 15, 371–386. Cloitre, M., Shear, M. K., Cancienne, J., & Zeitlin, S. B. (1994). Implicit and explicit memory for catastrophic associations to bodily sensation words in panic disorder. Cognitive Therapy and Research, 18, 225–240. Dalgleish, T. (1994). The relationship between anxiety and memory biases for material that has been selectively processed in a prior task. Behaviour Research and Therapy, 32, 227–231. Darke, S. (1988a). Anxiety and working memory capacity. Cognition & Emotion, 2, 145–154. Darke, S. (1988b). Effects of anxiety on inferential reasoning task performance. Journal of Personality and Social Psychology, 55, 499–505. Ehlers, A., Margraf, J., Davies, S., & Roth, W. T. (1988). Selective processing of threat cues in subjects with panic attacks. Cognition & Emotion, 2, 201–220. Eysenck, M. W. (1982). Attention and arousal: Cognition and performance. Berlin: Springer. Eysenck, M. W., & Byrne, A. (1994). Implicit memory bias, explicit memory bias, and anxiety. Cognition & Emotion, 8, 415–431. Eysenck, M. W., & Calvo, M. G. (1992). Anxiety and performance: The processing efficiency theory. Cognition & Emotion, 6, 409–434. Eysenck, M. W., & Mogg, K. (1992). Clinical anxiety, trait anxiety, and memory bias. In S-Å. Christianson (Ed.), Handbook of emotion and memory (pp. 429–450). Hillsdale, NJ: Erlbaum. Finlay-Jones, R., & Brown, G. W. (1981). Types of stressful life event and the onset of anxiety and depressive disorders. Psychological Medicine, 11, 803–815. Foa, E. B., Amir, N., Gershuny, B., Molnar, C., & Kozak, M. J. (1997). Implicit and explicit memory in obsessive-compulsive disorder. Journal of Anxiety Disorders, 11, 119–129. Fox, E., Russo, R., Bowles, R., & Dutton, K. (2001). Do threatening stimuli draw or hold visual attention in subclinical anxiety? Journal of Experimental Psychology: General, 130, 681–700. Friedman, B. H., Thayer, J. F., & Borkovec, T. D. (2000). Explicit memory bias for threat words in generalized anxiety disorder. Behavior Therapy, 31, 745–756. Graf, P., & Mandler, G. (1984). Activation makes words more accessible but not necessarily more retrievable. Journal of Verbal Learning and Verbal Behaviour, 23, 553–568. Humphreys, M. S., & Revelle, W. (1984). Personality, motivation, and performance: A theory of the relationship between individual differences and information processing. Psychological Review, 91, 153–184. Jacoby, L. L., Allan, L. G., Collins, J. C., & Larwill, L. K. (1988). Memory influences subjective experience: Noise judgments. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 240–247. Jacoby, L. L., & Dallas, M. (1981). On the relationship between autobiographical memory and perceptual learning. Journal of Experimental Psychology: General, 110, 303–340.
182
Jacoby, L. L., & Witherspoon, D. (1982). Remembering without awareness. Canadian Journal of Psychology, 36, 300–324. Levy, E. A., & Mineka, S. (1998). Anxiety and mood-congruent autobiographical memory: A conceptual failure to replicate. Cognition & Emotion, 12, 625–634. Lundh, L.-G., Czyzykow, S., & Ost, L.-G. (1997). Explicit and implicit memory bias in panic disorder with agoraphobia. Behaviour Research and Therapy, 35, 1003– 1014. Lundh, L.-G., & Ost, L.-G. (1996a). Face recognition in patients with social phobia. Scandinavian Journal of Behaviour Therapy, 25, 139–148. Lundh, L.-G., & Ost, L.-G. (1996b). Recognition bias for critical faces in social phobics. Behaviour Research and Therapy, 34, 787–794. Lundh, L.-G., & Ost, L.-G. (1997). Explicit and implicit memory bias in social phobia. The role of subdiagnostic type. Behaviour Research and Therapy, 35, 305–317. Lundh, L.-G., Thulin, U., Czyzykow, S., & Ost, L.-G. (1998). Recognition bias for safe faces in panic disorder with agoraphobia. Behaviour Research and Therapy, 36, 323–337. Lundh, L.-G., Wikstrom, J., Westerlund, J., & Ost, L.-G. (1999). Preattentive bias for emotional information in panic disorder with agoraphobia. Journal of Abnormal Psychology, 108, 222–232. MacLeod, C., & Donellan, A. M. (1993). Individual differences in anxiety and the restriction of working memory capacity. Personality and Individual Differences, 15, 163–173. MacLeod, C., & McLaughlin, K. (1995). Implicit and explicit memory bias in anxiety: A conceptual replication. Behaviour Research and Therapy, 33, 1–14. Mansell, W., Clark, D. M., Ehlers, A., & Chen, Y.-P. (1999). Social anxiety and attention away from emotional faces. Cognition & Emotion, 13, 673–690. Mathews, A., & MacLeod, C. (1985). Selective processing of threat cues in anxiety states. Behaviour Research and Therapy, 23, 563–569. Mathews, A., & MacLeod, C. (1994). Cognitive approaches to emotion and emotional disorders. Annual Review of Psychology, 45, 25–50. Mathews A., & MacLeod, C. (2002). Induced processing biases have causal effects on anxiety. Cognition & Emotion, 16, 310–315. Mathews, A., Mogg, K., Kentish, J., & Eysenck, M. (1995). Effect of psychological treatment on cognitive bias in generalized anxiety disorder. Behaviour Research and Therapy, 33, 293–303. Mathews, A., Mogg, K., May, J., & Eysenck, M. (1989). Implicit and explicit memory bias in anxiety. Journal of Abnormal Psychology, 98, 236–240. McNally, R. J., Foa, E. B., & Donnell, C. D. (1989). Memory bias for anxiety information in patients with panic disorder. Cognition & Emotion, 3, 27–44. McNally, R. J., Litz, B. T., Prassas, A., Shin, L. M., & Weathers, F. W. (1994). Emotional priming of autobiographical memory in post-traumatic stress disorder. Cognition & Emotion, 8, 351–367. McNally, R. J., Metzger, L. J., Lasko, N. B., Clancy, S. A., & Pitman, R. K. (1998). Directed forgetting of trauma cues in adult survivors of childhood sexual abuse with and without posttraumatic stress disorder. Journal of Abnormal Psychology, 107, 596–601.
183
Mogg, K., & Mathews, A. (1990). Is there a self-referent mood-congruent recall bias in anxiety? Behaviour Research and Therapy, 28, 91–92. Mogg, K., Mathews, A., & Weinman, J. (1987). Memory bias in clinical anxiety. Journal of Abnormal Psychology, 96, 94–98. Mogg, K., Mathews, A., & Weinman, J. (1989). Selective processing of threat cues in anxiety states: A replication. Behaviour Research and Therapy, 27, 317–323. Mueller, J. H. (1992). Anxiety and performance. In A. P. Smith & D. M. Jones (Eds.), Factors affecting human performance (Vol. 3, pp. 127–160). London: Academic Press. Nugent, K., & Mineka, S. (1994). The effect of high and low trait anxiety on implicit and explicit memory tasks. Cognition & Emotion, 8, 147–163. Nunn, J. D., Stevenson, R., & Whalan, G. (1984). Selective memory effects in agoraphobic patients. British Journal of Clinical Psychology, 23, 195–201. Öhman, A., & Mineka, S. (2001). Fears, phobias, and preparedness: Toward an evolved module of fear and fear learning. Psychological Review, 108, 483–522. Otto, M. W., McNally, R. J., Pollack, M. H., Chen, E., & Rosenbaum, J. F. (1994). Hemispheric laterality and memory bias for threat in anxiety disorder. Journal of Abnormal Psychology, 103, 828–831. Perruchet, P., & Baveux, P. (1989). Correlational analyses of explicit and implicit memory performance. Memory & Cognition, 17, 77–86. Pickles, A. J., & van den Broek, M. D. (1988). Failure to replicate evidence for phobic schemata in agoraphobic patients. British Journal of Clinical Psychology, 27, 271–272. Pury, C. L. S., & Mineka, S. (2001). Differential encoding of affective and nonaffective content information in trait anxiety. Cognition & Emotion, 15, 659–693. Radomsky, A. S., & Rachman, S. (1999). Memory bias in obsessive-compulsive disorder (OCD). Behaviour Research and Therapy, 37, 605–618. Rapee, R. M. (1994). Failure to replicate a memory bias in panic disorder. Journal of Anxiety Disorders, 8, 291–300. Rapee, R. M., McCallum, S. L., Melville, L. F., Ravenscroft, H., & Rodney, J. M. (1994). Memory bias in social phobia. Behaviour Research and Therapy, 32, 89–99. Reidy, J., & Richards, A. (1997). Anxiety and memory: A recall bias for threatening words in high anxiety. Behaviour Research and Therapy, 35, 531–542. Reisberg, D., & Heuer, F. (1992). Remembering the details of emotional scenes. In E. Wingrad & U. Neisser (Eds.), Affect and accuracy in recall: Studies of “flashbulb” memory (pp. 162–190). New York: Cambridge University Press. Revelle, W., & Loftus, D. A. (1990). Individual differences and arousal: Implications for the study of mood and memory. Cognition & Emotion, 3, 209–237. Richards, A., & French, C. C. (1991). Effects of encoding and anxiety on implicit and explicit memory performance. Personality and Individual Differences, 12, 131–139. Richards, A., & Whittaker, T. M. (1990). Effects of anxiety and mood manipulation in autobiographical memory. British Journal of Clinical Psychology, 29, 145–153. Richardson-Klavehn, A., & Bjork, R. A. (1988). Measures of memory. In M. R. Rosenzweig & L. W. Porter (Eds.), Annual Review of Psychology, 9, 475–543. Roediger, H. L., & Blaxton, T. A. (1987). Retrieval modes produce dissociations in memory for surface information. In D. Gorfein & R. R. Hoffman (Eds.), Memory
184
and cognitive processes: The Ebbinghaus Centennial Conference. Hillsdale, NJ: Erlbaum. Russo, R., Fox, E., Bellinger, L., & Nguyen-Van-Tam, D. P. (2001). Mood-congruent free recall bias in anxiety. Cognition & Emotion, 15, 419–433. Russo, R., Fox, E., & Bowles, R. J. (1999). On the status of implicit memory bias in anxiety. Cognition & Emotion, 13, 435–456. Sanz, J. (1996). Memory biases in social anxiety and depression. Cognition & Emotion, 10, 87–105. Tolin, D. F., Abramowitz, J. S., Brigidi, B. D., Amir, N., Street, G. P., & Foa, E. B. (2001). Memory and memory confidence in obsessive-compulsive disorder. Behaviour Research and Therapy, 39, 913–927. Vrana, S. R., Roodman, A., & Beckham, J. C. (1995). Selective processing of traumarelevant words in posttraumatic stress disorder. Journal of Anxiety Disorders, 9, 515–530. Watts, F. N., Trezise, L., & Sharrock, R. (1986). Processing of phobic stimuli. British Journal of Clinical Psychology, 25, 253–261. Wenzel, A., Jackson, L. C., & Holt, C. S. (2002). Social phobia and the recall of autobiographical memories. Depression & Anxiety, 15, 186–189. Wessel, I., van der Kooy, P., & Merckelbach, H. (2000). Differential recall of central and peripheral details of emotional slides is not a stable phenomenon. Memory, 8, 95–109. Wilhelm, S., McNally, R. J., Baer, L., & Florin, I. (1996). Directed forgetting in obsessive-compulsive disorder. Behaviour Research and Therapy, 34, 633–641. Wilhelm, S., McNally, R. J., Baer, L., & Florin, I. (1997). Autobiographical memory in obsessive-compulsive disorder. British Journal of Clinical Psychology, 36, 21– 31. Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology. Psychological Bulletin, 120, 3–24. Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1997). Cognitive psychology and emotional disorders. 2nd ed. Chichester, UK: Wiley. Wood, J., Mathews, A., & Dalgleish, T. (2001). Anxiety and cognitive inhibition. Emotion, 1, 166–181. Yiend, J., & Mathews, A. (2001). Anxiety and attention to threatening pictures. Quarterly Journal of Experimental Psychology, 54, 665–681.
185
6
A Question of Habit?
T
he truest claim that cognitive science can make might also be the least sophisticated: the mind tends to do what it has done before. In previous centuries philosophers and psychologists invented constructs such as associations, habit strength, and connectivity to formalize the truism, but others have known about it, too. In small towns in the Ozarks, for example, grandmothers have been overheard doling out warnings such as, “Don’t think those ugly thoughts; your mind will freeze that way.” Depressed persons, like most of us, usually don’t heed this advice. The thoughts frozen in their minds might not be “ugly,” but they often reflect disappointments, losses, failures, other unhappy events, and a generally negative interpretive stance toward ongoing experience. By considering these habits of thinking, we should better understand the nature of memory in depressed states. Deliberate attempts to remember are either impaired or facilitated in ways that appear related to habits of thinking. Even more commonly, memory is expressed indirectly through the content of current thoughts and interpretations. This chapter examines the relations among habitual thoughts—often called ruminations—and memory phenomena in depression. I use “depression” imprecisely, to refer to both diagnostic categories and the self-reported state of dysphoria. “Dysphoria” is used to denote undiagnosed negative affect, of the sort that produces moderate-to-high scores on the Beck Depression Inventory (BDI: Beck, Ward, Mendelson, Mock, & Erbaugh, 1961). Although the patterns of performance associated with self-reported measures of depressed mood—such as BDI scores—are often similar to those obtained from diagnosed samples, occasional evidence for differences should be noted (see Burt, Zembar, & Niederehe, 1995). We should note the imprecision associated with diagnosis as well (e.g., the boundary problems for mood disorders and generalized anxiety disorders; see
186
186
Brown, Di Nardo, Lehman, & Campbell, 2001). On the side of simplicity, however, many phenomena appear similar in form, if not extent, across studies measuring dysphoria and depression. (For a short review, see Wenzlaff, Meier, & Salas, 2002.) Further caution about the use of “depression” is appropriate when considering the many studies performed with experimental mood inductions (e.g., through the use of sad or happy music) in an effort to make causal statements about mood and memory. In some of these studies, the language of depression is used to describe the findings from negatively valenced inductions. (See Parrott & Hertel, 2000, for a description of these and other methodological issues.) Moodinduction studies sometimes reveal patterns similar to those in studies based on self-reports or diagnoses. But because habits of ruminating do not likely characterize students randomly assigned to listen to sad music, this chapter rarely addresses findings from mood inductions.
Habits of Thought Negative thinking has been such a prevalent feature of depression, as observed clinically, that Beck (1967) used “schema” to capture its habitual and interrelated qualities. These stable cognitive structures were proposed as one way of organizing and describing thought patterns in depression. “Rumination” is used as a process-oriented companion to the structural construct of a schema. Depressed people tend to ruminate. They ponder the episodes associated with sad feelings and imagine similar future occurrences. They focus on their feelings and wonder if they will ever change (see Ingram, 1990; Nolen-Hoeksema, 1991). Rumination can be an intensely attention-demanding process, yet the initiation of a ruminative episode is often thoughtless or automatic, even to the extent that the person can become so engaged without awareness. The draw toward rumination has a lot in common with the tendency to attend to external stimuli that are emotionally congruent with depressive concepts. For example, dysphoric students are slower to name the color of ink when it spells a word related to depression (e.g., Gotlib & McCann, 1984). On similar tasks nondepressed participants preferentially attend to positive members of word pairs and avoid the negative members, but clinically depressed and dysphoric participants lack this positive bias (McCabe & Gotlib, 1995; McCabe & Toman, 2000). Mood-related experience also encourages the tendency to find negative meaning in ambiguous information, relatively automatically. Lawson, MacLeod, and Hammond (2002) devised a clever indirect measure of such interpretive biases by measuring the blink reflex to noise occurring as participants imaged situations evoked by auditorially presented ambiguous and nonambiguous words. Participants who had scored high on the BDI produced particularly amplified blink reflexes during the imaging of ambiguous words that lend themselves to
187
negative interpretations, in a pattern similar to that for clearly negative words. Using a quite different paradigm, Wenzlaff and Bates (1998) presented strings of words that could be unscrambled to form either negative or non-negative sentences. Dysphoric students formed negative versions more often than others did. Even students who were formerly dysphoric displayed this bias under dual-task conditions. Although direct reports of beliefs and thoughts tend toward less negativity as depression lessens (Haaga, Dyck, & Ernst, 1991), findings such as those by Wenzlaff and Bates (1998) reveal the enduring nature of depressive habits of thought (also see Wenzlaff, Rude, Taylor, Stultz, & Sweatt, 2001). Depressive habits of thought seem, moreover, to exacerbate sad moods and predict future depressive episodes. In an extensive program of research by NolenHoeksema and her colleagues (e.g., Lyubomirsky, Caldwell, & Nolen-Hoeksema, 1998; Nolen-Hoeksema & Morrow, 1993), depressed and nondepressed participants have been asked to concentrate on either self-focused or distracting phrases. The self-focused phrases themselves do not suggest sad or depressed mood; in fact, the nondepressed participants typically report feeling no sadder after concentrating on them than they feel after entertaining thoughts about distracting phrases (e.g., geographical locations). But in study after study, the sad moods of depressed participants have increased following the self-focused statements and decreased following the distracting phrases. Similarly, Fennell, Teasdale, Jones, and Damle (1987) reported improvements in mood after depressed participants focused on distracting images of outdoor scenes. Therefore, habits of thought can affect mood, and recent discoveries have shown that they can also predict future mood impairment (e.g., Alloy, Abramson, & Francis, 1999; Nolen-Hoeksema, 2000; Rude, Wenzlaff, Gibbs, Vane, & Whitney, 2002). Measures of negative-thinking styles, demonstrated either through self-reports or by performance on laboratory tasks, predicted episodes of dysphoria and depression in weeks to come. So, just as depressed mood might establish habits of negative thinking, such habits at least portend, and perhaps help establish, not only temporary changes in mood but also future depressive episodes. Indeed, substantial empirical evidence now supports the concept of a vicious cycle of rumination and mood, described years ago by John Teasdale (1983).
Habits of Memory Separating habits of memory from habits of thinking is a somewhat arbitrary exercise. Obviously, ruminative episodes often include autobiographical memories that come to mind habitually. Less obviously, perhaps, even the acts of interpreting current experience in habitually negative ways are themselves instances of using memory implicitly or without intention. Jacoby and Kelley (1987, citing Polyani, 1958) distinguished the use of memory as an object for examination from the use of memory as a tool for perception and interpretation,
188
a helpful distinction in the context of considering cognitive habits. Both uses of memory—object and tool—can become biased out of habit. To refer to memory as an object (explicit memory) is to communicate awareness that the content of one’s thoughts derives from a past event. Episodes from one’s personal past can come to mind habitually, sometimes uninvited (but with full awareness of their temporal status) and sometimes deliberately and repeatedly sought. At the end of this section, I examine the well-documented tendency for depressed persons to remember negative episodes from their biographical past. Depressed persons habitually remember the bad. Habits of remembering negative events also show up in the laboratory. On tests of explicit memory, the tendency to think negatively can facilitate the deliberate remembering of experimental materials from the same conceptual neighborhood. On tests ostensibly unrelated to memory (implicit or indirect tests of memory), performance can also reflect habits of prior preferential processing. The experimental goal in both cases is to control the initial experience intended to operate as tool or object on the test (i.e., the experience to be “remembered”); however, the autobiographical method obviously lacks such control over the initial experience. In experimental studies, at least we know the characteristics of the immediate experience to be remembered. Yet it is important to keep in mind that the power of any mood-congruent effect—experimental or autobiographical—no doubt lies in uncontrolled, pre-experimental habits of thought.
Memory-as-Object in the Laboratory If we present positively and negatively affective words in a first experimental task and later ask the depressed and nondepressed participants to recall those words, chances are good that recall will in some way be congruent with mood. (See the meta-analysis by Matt, Vazquez, & Campbell, 1992.) Maybe the differences will be lopsided, especially in the case of dysphoria, with dysphoric participants recalling fewer positive words but the same number of negative words as nondysphoric participants do. Likely, the differences will obtain only for words encountered conceptually in the first task. In fact, some evidence suggests further restrictions to words considered in relation to the self (e.g., Bradley & Mathews, 1983; Derry & Kuiper, 1981; Dobson & Shaw, 1987). These limitations support the claim that habits of thought play a role in producing moodcongruent recall. Habits likely play a large role during the test itself. An illustration is the experiments performed by Murray, Whitehouse, and Alloy (1999). They found evidence of mood-congruent recall following a self-referential task, but only when participants were not required to guess in order to produce a criterion number of words. When recall was “forced,” the mood-related differences vanished. These results should encourage researchers to use a forcing procedure more often. A reasonable hypothesis is that mood-congruent biases influence the
189
fluency or ease of remembering, but that materials with other meaning can be brought to mind with sufficient persistence or external aid. Such an outcome, however, does not reduce our interest in the ubiquitous findings of moodcongruent recall and their interpretation. Most acts of recall are probably fluencydriven, without prolonged pursuit of additional material. Several theoretical frameworks have emphasized the greater degree of conceptual processing that depressed persons devote to negative materials. Some approaches propose that elaborative conceptual processing facilitates recall by establishing richer and more diverse retrieval routes (see Williams, Watts, MacLeod, & Mathews, 1988, and the network account by Bower, 1981). According to the framework of transfer-appropriate processing (Morris, Bransford, & Franks, 1977; see Roediger & McDermott, 1992), mood-congruent recall is established by the match between conceptual elaboration during the initial task and the conceptual basis of the attempt to remember. This account can be extended to include a third type of “occasion”: prior habits of thinking. Predated by such habits, negative self-referential material comes to mind with greater fluency, either during the initial episode to be remembered, during the test itself, or on both occasions. Probably more often than we realize, however, evidence for mood-congruent recall of conceptually processed material is not obtained. Can a consideration of cognitive habits suggest boundary conditions for the mood-congruent effect, at least on a post-hoc basis? Among others, Parrott and Spackman (2000) have written about the tendency for some people in negative moods to attempt mood repair by deliberately thinking positive thoughts. Indeed, evidence for moodincongruent recall might result from these attempts, as reconfirmed recently by Rusting and DeHart (2000). They instructed some participants to keep focusing on negative events (imagined vignettes for specified words or autobiographical episodes) used to induce negative mood, much like what one does in rumination. Other participants were instructed to engage in positive reappraisals of those events. Subsequent recall was mood-congruent for the former participants and mood-incongruent for the latter. These participants were not depressed; nevertheless, the results provide a model for what might happen naturally when depressed persons develop either set of cognitive habits.1 The literature on depression and mood-congruent recall is substantial. Many studies now conducted to examine depression-related biases in other types of cognitive tasks often include a demonstration of mood-congruent recall for comparison (e.g., Watkins, Mathews, Williamson, & Fuller, 1992). There are also published reports of mood-congruent recognition in depression (e.g., Wenzlaff et al., 2002), even one that reports mood-related differences in eventrelated potentials during the initial task of rating pleasantness, as well as during the recognition test (Deldin, Keller, Gergen, & Miller, 2001). Different levels of brain activity should indeed inform our understanding of mood-congruent memory, but they do not necessarily reveal the causal mechanisms underly-
190
ing the dispositions for negative interpretations. Habits might serve as instigators, outcomes, or both.
Memory as Tool In the last decade of the 20th century, a number of experiments on implicit memory in depression were published (e.g., Bazin, Perruchet, De Bonis, & Féline, 1994; Danion et al., 1991; Denny & Hunt, 1992; Elliott & Greene, 1992; Hertel, 1994; Hertel & Hardin, 1990; Watkins et al., 1992). Some of these experiments were not designed to address mood-congruence. But others varied the emotional valence of the materials in attempts to determine whether mood-congruent memory would be revealed on indirect tests, in situations in which people presumably are not trying to remember. In the same experiments that produced mood-congruent recall, there was at first a notable failure to find moodcongruent differences on indirect tests. Some researchers anticipated this outcome, predicting that prior conceptual processing, so advantageous for strategic retrieval tasks, is unimportant for nonstrategic indirect tests (e.g., Williams et al., 1988). If mood-congruent memory relies on differences during initial conceptual processing, mood congruence—like other conceptual manipulations, in this line of reasoning—should not characterize performance on indirect tests (also see Denny & Hunt, 1992). An exception to the initial rule was reported by Ruiz-Caballero and González (1994), who found evidence of mood congruence on a stem-completion task. In this indirect test, beginning letters of both previously read words and new words are provided, along with instruction to complete the stems with the first words that come to mind. Such a test can easily be turned into an explicit test of memory-as-object if participants begin to use the stems as cues for deliberate recall, even on occasional trials (see Watkins et al., 1992). Responding to these concerns about the process purity of the test, Ruiz-Caballero and González manipulated intention to learn in a second experiment. This manipulation affected levels of free recall (which followed stem completion), but not levels of “priming” on stem completion, although performance on both tests showed evidence of mood congruence. If the participants engaged in deliberate recall on the stemcompletion test, the authors argued, performance on that test should have shown effects of intention to learn. But, as Chapman and Chapman (1973) alerted us, the two memory tests might be differentially sensitive to the manipulation of intention to learn, while showing similar differences according to emotional valence. The provision of stems might compensate for the lack of intentionality in the unintentional condition and thereby overwhelm the effect of intentional learning, while nevertheless cuing recall in mood-congruent ways. Therefore, doubts about mood-congruent uses of memory as tool remained. Evidence for mood-congruent memory from tests of stem completion is surprising, particularly to a reader in a transfer-appropriate-processing frame of
191
mind. Among others, Roediger and McDermott (1992) discussed the importance of the match between types of processing at “study” and test. Moodcongruent thoughts during initial exposure should facilitate similar thoughts, but not perceptions, at the time of testing, regardless of whether memory is used as tool or object. Word-stem cues should function more perceptually than conceptually. Therefore, mood-congruent performance on stem completion should not occur. In hindsight, it now seems clear that prior habits of reading moodcongruent words can sometimes facilitate performance on so-called perceptually driven tests (e.g., unprimed word identification, as used by Von Hippel, Hawkins, & Narayan, 1994, and stem completion of both old and new items, as reported by Bazin et al., 1994). Nevertheless, it was rare to see or to expect to see evidence of mood-congruent memory on tests lacking a substantial reliance on conceptual processing. Taking Roediger and McDermott’s (1992) suggestion to use conceptual indirect tests, Watkins and his associates (Watkins, Martin, & Stern, 2000; Watkins, Vache, Verney, Muller, & Mathews, 1996) have found mixed evidence of mood-congruent performance by clinically depressed participants. Watkins et al. (1996) found it on a test of free association, but there was no accompanying assurance against contamination by explicit remembering. Using both perceptual and conceptual indirect tests, Watkins et al. (2000) found evidence of mood congruence following a conceptual orienting task, but only on one conceptually driven test, the test they called word retrieval (in which one produces words when cued by dictionary definitions). Again, we can’t be sure that participants were unaware of the memorial nature of the task or that they did not attempt to remember deliberately if they were aware. Contamination on indirect tests by deliberate remembering should not undermine interest in the many ways that habits of thought can influence memory. No doubt, all tests that reflect prior experience do so through a mixture of automatic and recollective processes (see Jacoby, 1991). Even so, it is important to know whether instances in which memory is used primarily as a tool for understanding current experience are affected by habits of negative thinking in depression, and in this regard more evidence is needed. Obtaining that evidence in the laboratory is likely made more difficult by problems in controlling the thoughts that should come to mind on indirect tests. Finding conceptual tests that do not invite deliberate uses of memory as object and, at the same time, produce effects that override other sources of fluency from the past has been difficult. Thinking of words from definitions might work because other words cannot be used, but tests of free associations may invite too many extra-experimental responses. In other words, memory-as-tool is hard to control experimentally. A general-purpose or prototypical tool might work better in experimental demonstrations of implicit bias. Past habits of thinking influence performance on current tasks not only because the exact thought keeps returning (the realworld analog for the indirect test of memory as tool) but also because past-
192
related thoughts guide current thoughts. Recent experiments in my lab illustrate this point (Hertel, Mathews, Peterson, & Kintner, in press). An initial “training” phase was designed to encourage either negative or nonnegative interpretations of homographs as participants judged semantic relatedness (e.g., pursue vs. celery, in relation to stalk). The subsequent task directed participants to form images of individual words, none of which had been used during training and many of which were homographs with both negative and nonnegative interpretations; performance in this task reflected the bias established during training. The training phase in this line of research is intended to model what happens naturally when prior habits of thought influence current interpretations in emotionally biased ways. The influence itself is an example of memory, put to use as tool, although we tend not to think about it that way. Our typical impressions about memorial influences more often occur in conjunction with being reminded about our personal past.
Memory as Autobiographical Experience Habits of thought and rumination often include memories of events from the personal past. Some of the earliest evidence for mood-congruent memory in depression was obtained in autobiographical studies (e.g., Clark & Teasdale, 1982), with accompanying interpretations that related mood congruence to habits of thinking. Now, direct evidence of this linkage is available. Using the same set of ruminative or distracting procedures previously used to affect the temporary mood of dysphoric students, Lyubomirsky et al. (1998) produced differentially negative biases in autobiographical recall. In experiment 1, Lyubormirsky et al. (1998) randomly assigned dysphoric and nondysphoric students to engage in 8 minutes of rumination or distraction and then requested free recall of events from their lives. Following recall, the students rated the hedonic tone of the memories they produced. Dysphoric students who had ruminated prior to recall rated their memories as less positive and more negative than students in the other three conditions did. The same pattern of results was obtained in experiment 2, in which the students were prompted to recall two happy and two unhappy events, and in experiment 3, in which the students judged frequency of occurrence in their lives for 10 positive and 10 negative events listed by the experimenter. In experiment 4, students were instructed to think aloud during the rumination or distraction phase, and their thoughts were recorded. The first six autobiographical memories mentioned in the tapes were rated for negativity and unhappiness by independent judges. The same pattern of mood congruence occurred: dysphoric students who ruminated produced more negative memories, compared to those who were distracted and compared to nondysphoric students in both conditions. All experiments also produced the pattern of effects on mood ratings described previously in this chapter: only the dysphoric students felt more or less sad and depressed as a function
193
of rumination or distraction, respectively. In summary, ruminative thoughts not only affected mood; they also invoked or contained more negatively toned memories from the dysphoric students’ personal pasts. The experiments by Lyubomirsky et al. (1998) could not address the precise nature of the mood-congruent finding. The authors acknowledge difficulties in concluding that the effects in the first three experiments pertained to memory instead of merely to ratings exaggerated by rumination. And, of course, in any autobiographical study, we cannot be sure that the memories are accurate (although accuracy issues do not trump all other interests in bias). So, with these qualifications, it is reasonable to conclude that ruminative habits encourage negative thoughts about the past. Especially those students with a reason for focusing on negative past events were led to do so by a brief ruminative episode. Making a similar point, Rothkopf and Blaney (1991) found clearer evidence of mood-congruent autobiographical recall when students filled out the BDI prior to the recall test rather than at other times. One interpretation of this type of result points to demand established by the statements on the BDI. Another interpretation, however, posits that a depressed person must become aware of her depressed state for the ruminative memories to be invoked. The self-schema must be “activated” (Beck, 1967). Caught unaware of self, he might sidestep habits of remembering the negative, and then other characteristics of the situation could more powerfully guide performance. The notion that a schema or prototype guides attempts to remember is consistent with a second major characteristic of autobiographical memory in depression, that is, its tendency to be overly general, as if attempts to remember events from one’s life stop short at the categorical level (see the review by Healy & Williams, 1999). Williams, Teasdale, Segal, and Soulsby (2000) provide the example of the cue word “kindness” eliciting the following memory: “My grandmother was always kind to me. She used to take me out when my father got cross” (p. 150). The response referred to a category of events instead of to a (requested) specific memory. Both positive and negative autobiographical memories tend to be described at this categorical level by depressed people (as well as by those who suffer from posttraumatic stress disorder; see chapter 4). This tendency might reflect an unconscious habit or a conscious strategy to avoid emotion associated with specific details (Williams et al., 2000).2 As with so many aspects of depressive cognition, the direction of cause is not at all clear. An overly categorical pattern of thinking and remembering might very well contribute to onset of depressive episodes, as suggested by Beck’s (1967) theory and by hopelessness formulations of depression (Abramson, Metalsky, & Alloy, 1989). Whatever the origin, the style is not so habitual that it cannot be modified (Williams et al., 2000). There is even evidence that a focus on specific experience (instead of abstract analysis) can at least temporarily reduce the categorical extent of autobiographical memory (Watkins & Teasdale, 2001). Building new habits to recruit specific instances can deflect a habitual dark-cloud schema and reveal a differ-
194
ent self to the rememberer. Of course, this assumption has long been incorporated into the techniques of cognitive-behavioral therapy.
Summary When past events serve as the objects of current thought, these thoughts are both more negative and more abstract for depressed people than for others. Habit’s role in explicit memory bias is depicted in results from manipulations of selfreferential processing and rumination. When memory is being used as a tool for other purposes, however, convincing evidence of similar bias has not yet been established for specific events, even though common sense tells us that these unintended effects are ubiquitous. After all, these cognitive habits are nothing other than the nondeliberate influences of past conceptual experiences during current, similar conceptual acts. Interpretive biases, surely established primarily by past experience, are well documented in depression. A primary question about the influence of cognitive habits on both forms of memory concerns the extent to which the influence relies on awareness of one’s mood state. Tasks that occur prior to tests for mood-congruent recall include selfreferential judgments of trait words, self-referential images, ruminative training, and a variety of mood-related forms and inventories. These tasks are guaranteed to make one consider one’s mood state. The distracting phrases used by Lyubomirsky et al. (1998) might sidestep mood congruence not merely because mood is temporarily improved but also because other habits of thinking have not been recently exercised. A continuing task for future research is to ask whether negative biases can be invoked without awareness of self. Interpretive and memory biases are not always symmetrical with mood. Depending on the type of “depression” and on the nature of the materials, the depressed participants do not always recall more negative words than positive words, but they do fail to show the positive bias in the nondepressed group. Sometimes they recall negative materials at the same level as the nondepressed group (but fewer positive words). In part, these asymmetries reflect other differences associated with mood and memory—differences in the degree to which controlled, strategic thoughts are initiated and used. This is the other side of the “habit” coin.
Impaired Control
in Nonhabitual Tasks
Deficient cognitive control sets the stage for habits to emerge. At the same time, thoughts that habitually occupy attention leave little mental room for thoughts about anything else. These assertions of a reciprocal relation between habits of thought and controlled attention characterize theoretical approaches to memory
195
in depressed states, particularly when the topic pertains to impaired performance. Memory impairment is one of the frequent complaints of people in depressed states (Beck, 1967). “Control” refers to the operation of cognitive procedures at the opposite end of the continuum from habit (see Jacoby, Jennings, & Hay, 1996). Controlled procedures are initiated with awareness or intent, and their components are not well integrated. From a phenomenological perspective, an individual must decide what to do or think next—quite a strain in depression. No wonder that effort metaphors have been used so often in descriptions of depressed participants’ difficulties in tasks that require or reflect control (see Hasher & Zacks, 1979; Weingartner, Cohen, Murphy, Martello, & Gerdt, 1981).
Effortful Construction
of Memory as Object
The idea that depressed people have trouble carrying out effortful mental procedures emerged first in early studies of intellectual functioning. (See the review by Hartlage, Alloy, Vazquez, & Dykman, 1993.) Only after attention theorists (e.g., Posner & Snyder, 1975) began to write about the control of attention, however, did researchers begin to connect effortful or controlled processing to performance on memory tests (Hasher & Zacks, 1979; Tyler, Hertel, McCallum, & Ellis, 1979). In a nutshell, the idea was that the more effortful or attentiondemanding the process, the more likely that the product would be remembered. Like the levels-of-processing framework popular among memory researchers at that time (Craik & Lockhart, 1972), the connection of cognitive effort to memory was based on assumptions about the strength of memory traces or links in an associative network. In the case of effort, however, strength was thought to directly reflect the amount of attention or effort expended during initial processing, instead of the type or level of processing. Orienting tasks vary in the degree of effortful, attention-demanding processes required to complete them; those requiring more effort supposedly produce a stronger memory. This idea was not unlike the idea of difficulty, but effort referred to a characteristic of processing, not the task itself. Effort-inducing tasks inspired or required more focused concentration, as revealed by longer latencies to perform a simple secondary task (e.g., Tyler et al., 1979). The applicability of the effort-memory connection to depression was a central feature of Hasher and Zacks’s (1979) often-cited work. Since then, literature reviews and a meta-analysis have supported the idea that depression-related deficits in memory occur primarily when the initial orientation task requires controlled concentration (Burt et al., 1995; Hartlage et al., 1993). Hartlage et al. also described the variety of theoretical accounts of effort’s role in depression-related impairment. Condensed and simplified, one account connects the concept of reduced attentional control to abnormal frontal function. (See
196
Davidson, 2000, for a more recent review.) Another claims that effortful processing is limited by the allocation of resources to personal concerns, presumably for the duration of the depressive episode (e.g., Ellis & Ashbrook, 1988). Implicit or explicit in all effort accounts is the assumption that greater effort produces a memory trace that is more easily found when needed. This assumption, rarely stated quite so simply, is clearly too simple in any form. The ability to focus attention and concentrate—either because the neurotransmitters are fully supplied to the appropriate areas of the brain or because other matters are not more compelling—likely has beneficial consequences for memory, but only when those particular procedures are subject to replication on the test. The intensity of concentration is important to memory only in the sense that it must be sufficient to carry out the procedures that will later, when replicated, benefit performance on a memory test. According to a transferappropriate-processing perspective, research should examine that sufficiency, as well as the match in attentional focus across the occasions of initial exposure and memory test. One obvious way to evaluate the transfer-appropriate claim is to examine the literature on memory as tool-versus-object in depressed states, because both the need for controlled attention and its focus varies across types of tests. If depressed people do not focus on aspects that will later guide performance on memory tests, and if they do not concentrate sufficiently for later replication, performance should suffer, regardless of the “intentionality” of the test or its difficulty.
Appropriate Use of Memory
as Tool or Object
Indirect tests of memory are designed to elicit automatic or habitual use of prior experience and often do not require concentrated effort during their performance. Spelling homophones is an example of such a test. It takes little concentration to spell common words read aloud by the experimenter at a fast pace (e.g., week/ weak). Similarly, the orienting tasks prior to the spelling test often require little concentration on one of the two alternative meanings (e.g., while listening to the question: Name the days of the week). Simple spelling and comprehension use habitual procedures. As expected, the dysphoric students’ spelling performance matched that of controls in effects of question-biased meaning; yet in the same experiment, the subsequent use of recognition strategies was impaired in the dysphoric group (Hertel & Hardin, 1990). On the other hand, memory can operate as a tool in performing somewhat more attention-demanding tasks. Even stem- and fragment-completion tests, when not performed habitually, might function as puzzles that require at least momentary effort and concentration, yet the available reports show no evidence of depression-related impairment (in the same experiments that revealed impaired free or cued recall of nonnegative words; e.g., Bazin et al., 1994; Denny
197
& Hunt, 1992; Ruiz-Caballero & González, 1994; Watkins et al. 1992).3 Why should we not see occasional evidence of impairment? One possible reason is that it is sufficient to have focused attention on the look or sound of the words in the orienting task in order that memory operates as a tool during perceptual indirect tests—tests that might well demand concentration on seeing or hearing them again. Sustained processing of meaning does not boost performance, but any task that discourages attention to words as units might decrease it. For example, consider an experiment that used a word-identification test, in which previously presented and nonpresented words were displayed very briefly and back-masked, and the task was merely to try to name the words. Anyone who has tried these tests knows that they require focused attention to “see” the words. To what extent should identification benefit from prior exposure (memory as tool)? In the orienting task of one such experiment (Hertel, 1994), I asked participants to evaluate some words according to the degree of roundness in their letters. The words evaluated in this task were identified more frequently on the subsequent perceptual test than were words not previously exposed, but this effect was smaller for the depressed participants! A reasonable postdiction is that during the orienting task the depressed participants attended less often to the words as integrated units and so were disadvantaged in naming them on the test. Word identification benefits from prior lexical processing, even as it requires concentration or effort. Conceptual processing facilitates use of the same concepts later on. Conceptual implicit tests require prior attention to meaning and therefore should reflect the extent to which meaning has been fully attended. For example, Jenkins and McDowall (2001) found a depression-related impairment on both free recall and the indirect test of category association, but only for the words that had been generated from antonyms (vs. merely read) during the orienting phase. No clear evidence of impaired fragment completion was found for either the previously generated or previously read words. These results are not as unambiguous as I describe them; for example, baseline performance on the category-association test was low for the depressed group. Nevertheless, the pattern illustrates the importance of transfer-appropriate conceptual processing. And the results also suggest that the depressed participants’ conceptual processing during antonymgeneration was insufficient to support the later production of those concepts during recall or category association. Whether category association functioned purely as a test of memory as a tool cannot be assured, of course. Like the freeassociation task used by Watkins et al. (1996), it might have been contaminated by attempts to remember, particularly in the nondepressed group. Because other category associates are available, the depressed participants could perform the test without concentrating on the past and, therefore, without showing the benefit of prior generation of category members, and a mood-related impairment would thereby result. Depressed people are less likely to devote effort to thinking about the past when the current task can be performed without doing so. This claim receives indirect
198
support from a series of problem-solving experiments in which memory for similar past problems operated as tool or object in facilitating solutions for current problems (Hertel & Knoedler, 1996). The logic-based word problems were themselves attention–demanding and quite difficult without prior experience in learning how to solve similar problems. During training, problems were presented for solution attempts, and each (almost always unsuccessful) attempt was followed by experimenter-provided instruction in how to solve that problem. Following the training phase, a series of structurally identical but superficially dissimilar target problems was presented, always without such instruction. Because we were curious about how memory operates as a tool in problem solving, we included a condition to measure spontaneous transfer; in this condition, the training problems were not mentioned during the target series. We also included a condition in which we provided very specific cues and instructions for using memory for the structurally similar training problem “as object” just prior to presenting each target. In two experiments, these hints for recalling the prior analogy actually disrupted performance by the nondysphoric students. Focusing attention on the details of the past problem while solving a new one seemed to be a transfer-inappropriate use of prior experience; in comparison, more targets were solved when memory for the analogy was allowed to operate as a tool. The dysphoric students performed similarly well, with or without hints to think back, which led us to suspect that they did not sustain attention to the relevant past problem and therefore were not led astray in transfer-inappropriate ways. Attending to important features of the target problem itself benefited from earlier attention to those features of the analog. In predicting performance in the problem-solving experiments, Knoedler and I initially reasoned that if the nondepressed students’ solution rates were similar with and without hints, they might have been using memory as object on their own initiative in the no-hint condition. If this pattern were obtained, we expected to find depressive impairments. When the hints turned out to be detrimental for nondepressed students, we inferred that memory for training problems in the no-hint condition operated spontaneously as a tool. In reasoning this way, we did not truly believe that either use was “all or none.” Similarly, most researchers understand that memory usually operates as both tool and object in a variety of memory tasks and that memory tasks are not “process-pure” (Jacoby, 1991). Yet when no attempt is made to determine the degree to which one use of memory contaminates the other use, reasoning about the role of controlled attention in depressive memory is made more difficult. Luckily, Jacoby’s process-dissociation procedure directly addressed the problem of how to examine the separate influences of tool and object memory. The central method used by Jacoby—and others, subsequently—entails two conditions of instructions: one in which automatic or habitual uses of memory operate in concert with controlled attention to the past and one in which they operate in opposition to each other in producing a response. In recognition tasks,
199
for example, participants are instructed to endorse items that either feel familiar or are recollected from a specific prior task on inclusion trials, whereas on exclusion trials, they are instructed to reject items from that specific task while endorsing other old items. In the inclusion case, recognition hits are produced by both controlled memory for specific prior occurrences and habitual procedures of reading or thinking words that were earlier read or thought. The uses of memory operate in concert. In the exclusion case, however, items from the specific prior task are (erroneously) endorsed only if they produce a feeling of familiarity in the absence of memory for source as object. Habit and control operate in opposition to each other. When independence of the two classes of influences can be assumed, their estimates can be computed from the proportion of positive recognition responses under the two instructional conditions. The relevance of these procedures to depression-related impairment should be obvious: Unless habits are disrupted during orientation, habitual use of memory as tool should remain intact, and only the estimates of controlled uses of memory as object should reveal the impairment. This indeed is what we found on recognition tests following a conceptual orienting task (Hertel & Milan, 1994). The dissociation was replicated through similar procedures designed for conceptually cued fragment-completion tests (e.g., building-s_o_e) following the reading of meaningful cue-target pairs in the orienting task (e.g., building-stone), (Hertel, 1998; also see Hertel & Meiser, 2000, for a multinomial model applied to those results). Again, the importance of transfer-appropriate attention across orienting task and test is illustrated. To the extent that participants attended to the connection between the cue and target initially, they would be able to attend to it as an object in the past, including the target or excluding it when required by instruction. A less careful focus on the pair initially would make it harder to both exclude and include deliberately on the test. The component of memory that represents the past as the object of attention is the one that shows a deficit associated with depression. However, the mere reading of the word pair during the orienting task makes the target available later, as a completion of the fragment, in the presence or absence of controlled recollection, and this habit-oriented component of memory was unimpaired. Thus, attentional control is indeed a factor in producing memory impairment in depression, when performance relies on the replication of focus across the two occasions.4
Habit as Detriment to Control Poor performance often reflects absence of habit in performing a cognitive act. This is why controlled attention is required—to do the nonhabitual and think about things one does not usually think about. Seemingly, all of the reported cases of depression-related impairment involved nonnegative materials—materials unrelated to the personal concerns thought to occupy attention habitually in depression. Habits of negative thinking benefit the controlled use of memory
200
for related stimuli, as is the case on tests of mood-congruent recall, but otherwise they can detract. Therefore, attentional control in depression should depend on the suppression of habitual thoughts in the service of focusing elsewhere. A line of research begun by Hasher and Zacks (1988) has shown evidence of inhibition difficulties associated with aging that might also characterize depression. Like effort-related difficulties, inhibitory difficulties in depression could logically arise from fundamental frontal dysfunction, from specific concerns associated with the depressed state, or from both (see Hertel, 1997). Reliance on an inhibition construct is not the only way to represent the problem of impaired control in the context of habitual thinking. We might slightly shift perspective to propose that control is made more difficult when other habits are strong. Habits of attending (e.g., to negatively toned events) might make the act of disengaging attention more difficult than it otherwise might be in the course of turning attention to a different event. At this stage of research, however, the aptness of the metaphor (inhibition vs. disengagement) is less important than gathering evidence for depression-related difficulties in turning away from habit-related stimuli in order to do something else. The “something else” in experimental studies has often been the task to respond to a dot or color patch in a screen location that differs from the one occupied by a mood-relevant word (e.g., Bradley, Mogg, & Lee, 1997; McCabe & Gotlib, 1995). Bradley et al. found dysphoria-related biases under conditions of suprathreshold but not subthreshold exposures to the words. They argued that the problem was therefore one of disengaging attention, because a bias in allocating attention to relevant stimuli would pertain for both exposure conditions. Recently, investigators have used the negative-priming paradigm to examine depression-related difficulties in disengagement or inhibition. In the standard version of this paradigm, items to be attended and ignored are presented concurrently on each trial, with the attended item on trial X sometimes having served as the ignored item on trial X-1. When the ignored is subsequently attended, or even when the same category of the ignored is now the category of the attended, the judgment for trial X is slowed, compared to control conditions. This slowing (called negative priming) is attributed sometimes to inhibition of the ignored item or category and sometimes to transfer-inappropriate attention (Neill & Mathis, 1998). If depressed people have trouble disengaging from moodrelated thoughts, negative-priming effects should be reduced or absent. Some suggestion of this difficulty has been demonstrated in a simple letter-identification task after students were induced by music to feel sad (von Hecker, Conway, Meiser, & Holm, 2002). However, Joormann (in press) has used a more conceptually relevant task. She asked nondysphoric and mildly dysphoric students to judge the self-relevance of negative and positive targets while ignoring distractors of the same or opposite valence. All participants took longer to judge the selfrelevance of positive words when positive words had been ignored on the previous trial, compared to previous trials of ignored negative words. This finding
201
roughly corresponds to the typical negative-priming effect. The same pattern of negative priming was obtained for nondysphoric participants in judging negative items after having ignored negative (vs. positive) items, but it was not pertain obtained for mildly dysphoric participants or for those who reported prior episodes of depressed mood. The difference in these patterns suggests that the participants who tended currently to be in more negative mood states (or reported feeling depressed in the past) ignored the negative adjectives less successfully.5 Difficulty in disengaging attention from negative material is more clearly documented in connection with anxiety disorders or high levels of trait anxiety (see Fox, Russo, Bowles, & Dutton, 2001). Again, when researchers use scores on inventories like the BDI or even when they use diagnostic labels to form groups for experimental purposes, their dysphoric or depressed participants are likely to be anxious. In the other direction, ruminative habits more typically attributed to depression also typify anxious states or traits (Nolen-Hoeksema, 2000). Regardless of their lack of specificity to depression, however, habits of dwelling on the negative have consequences for memory beyond those reported in relation to evidence for mood-congruent memory, and these consequences are not always desirable. One consequence might be trouble in forgetting. Not only do depressed people have trouble ignoring mood-congruent material when they focus on other material presented concomitantly (as in Joormann’s negative-priming studies), they also less successfully suppress mood-congruent material in more sustained attempts. Evidence for difficulties in sustained thought suppression has been documented by Wegner, Wenzlaff, and their colleagues (for review, see Wenzlaff & Bates, 1998). Thought suppression, whether accomplished directly or indirectly in the service of thinking about something else, plays an important role in forgetting (see Anderson & Neely, 1996). And forgetting can be a valuable outcome, particularly for the depressed person who might prefer not to ruminate about past events beyond one’s control. Intentional forgetting has been studied in the lab by directing participants to forget items or lists of items.6 Clearly, instructions to forget lists can have their intended effect, particularly when the participants are given something else to think about. To document effects of direct suppression on individual items, Anderson and Green (2001) developed the “think/no-think” procedure. After learning pairs of unrelated words to a fixed criterion, the participants were given a varied number of trials (0, 1, 8, or 16) to practice the retrieval of the second member of some pairs and to practice the suppression of the second member of other pairs. The first member of each pair was presented as a cue for either retrieval or suppression. This practice phase was followed by a final test of cued recall, in which participants were instructed to recall all of the response words, regardless of previous instruction. In several experiments, the main finding was increased forgetting as a function of practice in suppression, both when the original studied cues were used at test and when new cues related categorically to
202
the response words were used at test. The simple story is that people can successfully suppress to the point of incurring forgetting. Can depressed people do this as well as others? Melissa Gerstle and I have recently conducted an experiment to answer that question (Hertel & Gerstle, in press). To investigate the possibility of mood-congruent suppression, we replicated the procedures developed by Anderson and Green (2001), but with one major change: The cues we used throughout the experiment were adjectives related to the target nouns. On half of the pairs, the adjectives gave the essentially neutral nouns a negative meaning, and on the other half, the meaning was positive. (The materials were fully counterbalanced, e.g., gloomy cottage vs. splendid cottage.) To encourage initial attention to the emotional valence of the materials, in the learning task we instructed participants to construct a self-referential image for each pair and to rate the meaningfulness of the image. These materials and instructions had been used earlier in a recall experiment without suppression instructions, together with other materials in which a third set of adjectives produced neutral concepts when paired with the same nouns (e.g., plain cottage; Hertel & Parks, 2002). In that experiment, dysphoric and nondysphoric participants alike recalled the nouns when they had been given emotional meanings (both positive and negative) more often than when they had been given neutral meanings. In the suppression experiment, Gerstle and I also found that recall was not significantly associated with emotional valence for the dysphoric participants (although the controls showed more forgetting of nouns from positive pairs). More important, however, was our finding that the effect of practicing suppression on later recall was significantly reduced in the dysphoric group. The dysphoric participants recalled as many of the rehearsed items as did the nondepressed participant and significantly more of the targets that they had practiced suppressing, regardless of the cues’ valence (see fig. 6.1). Similarly, Power, Dalgleish, Claudio, Tata, and Kentish (2000) found evidence for depression-related difficulties in one of their three experiments on directed forgetting. Perhaps because their participants were clinically depressed (experiment 3), however, this difference was restricted to negative adjectives. Recall of positive adjectives—to be remembered or to be forgotten—was no better or worse than that of anxious participants and controls. At the same time that depressed people might have trouble deliberately forgetting self-referential thoughts, they also have trouble remembering what they had been thinking about prior to an episode of self-referential rumination. At least this is true for the controlled component of memory, as estimated through process-dissociation procedures on a test of cued fragment completion (Hertel, 1998). In the experiment previously mentioned, we inserted what might be seen as an interference phase between the initial orienting task of reading word pairs and the final memory phase. We assigned a third of the dysphoric and control participants to merely sit and wait for 7 minutes, another third to ruminate, and the final third to think about other matters, such as geographical locations. The
203
Nondys Respond
Nondys Suppress
Dysphoric Respond
Dysphoric Suppress
Baseline
Practice
Mean % Recalled
100
90
80
70
Intervening Task Figure 6.1. The mean percentage of targets recalled on the final test, after depressed and nondepressed participants had practiced retrieving or suppressing them in response to cues. (Adapted from “Depressive Deficits in Forgetting,” by P. T. Hertel and M. Gerstle, in press, Psychological Science. Copyright © 2003 by the American Psychological Society.)
latter two groups actually underwent the rumination or distraction procedures used by Nolen-Hoeksema and Morrow (1993) that I described elsewhere in this chapter. Following the period of self-focus in the rumination condition, the dysphoric participants showed impaired control on the memory test (relative to the other participants), much like those who simply waited during the 7-minute interval. However, the participants who thought about objects and locations showed no depression-related impairment in control (see fig. 6.2). In addition to improving mood under some conditions (e.g., Lyubormirsky et al., 1998), distraction also seems to improve controlled attention to the past when the rememberer is depressed. And because the distracted group showed no impairment, it is tempting to infer that even those who merely waited for 7 minutes focused on sufficiently compelling matters that thinking back during the memory test was made more difficult. Inherent in those results (Hertel, 1998) are the final two points of this chapter: situations that permit mind wandering invite habitual thinking and impair memory for other matters. Deficient performance is not, therefore, an inevitable outcome of depression, because habits can be opposed in ways that benefit control.
204
Mean Estimated Recollection
0.5
Nondysphoric Dysphoric
0.4 0.3 0.2 0.1 0 Wait
Ruminate
Distract
Intervening Task Figure 6.2. The mean estimate of controlled retrieval by dysphoric and nondysphoric participants, following an intervening period of waiting for 7 minutes, entertaining ruminative thoughts, or entertaining distracting thoughts. Adapted from “The Relationship between Rumination and Impaired Memory in Dysphoric Moods,” by P. T. Hertel, 1998, Journal of Abnormal Psychology, 107, p. 170. Copyright © 1998 by the American Psychological Association.
Opposing Habit Through Control Here is the prototype of the depressed rememberer as constructed by the findings described thus far: she (well, yes, she) is someone who habitually ruminates about personal concerns and other negative events and whose interpretations and perceptions reflect those habits; someone who tends to remember negatively affective events more often or remembers events as more negatively toned; and someone whose memory for neutral or positive events is occasionally impaired, at least when attention is allowed to stray to habitual patterns. This person seems to have little cognitive self-control. (One can almost hear her grandmother’s admonishments.) I use “self-control” to imply that the most important aspect of impaired control in depression is control initiated by the self. Controlled attention to any stimulus can be facilitated by the environment; the structure of the task, stimulus salience, and its interest value are some of the relevant dimensions. Cognitive self-control is merely a convenient term for referring to the focusing, switching, and sustaining of attention under conditions where stimulus and task dimensions do not play obvious roles. In structural terms, self-control is loosely analogous to the central executive component of working memory (Baddeley & Hitch, 1974). Its physical components are frontal functions; indeed, much of the evidence for depression-related differences in brain activity implicates activity in the prefrontal cortex (see Davidson, 2000; Heller & Nitschke, 1998). Similar
205
behavioral evidence is supplied by Channon (1996), who found impairments on a neuropsychological test of frontal function—the Wisconsin Card Sorting Test (WCST)—even in a sample of dysphoric college students. Furthermore, by statistically controlling for correlations with BDI scores, Davis and Nolen-Hoeksema (2000) found WCST differences associated with the tendency to ruminate. Selfreported ruminators made more perseverative errors on the WCST and failed to maintain set more often than did nonruminators. In a basic sense, then, cognitive self-control is a problem in depression, as well as a problem for people who tend to ruminate. A good simple example of the memory problems associated with depressive deficits in self-control or initiation is impaired prospective memory in the absence of reminding cues (Rude, Hertel, Jarrold, Covich, & Hedlund, 1999). Depressed and control participants were instructed at the start of the session to press a function key every 5 minutes during their performance of an ongoing test of general knowledge. The depressed participants checked the time less often and thereby made more errors on the prospective task. Reduced self-control affects performance in memory experiments under conditions of poor external control. For example, dysphoric students’ performance yielded lowerestimates of controlled recollection in my cued-fragment-completion study (1998) after a period of lax external control. Doing nothing for 7 minutes presented ample opportunity for mind wandering, because the dysphoric students had more trouble subsequently turning attention to the task of remembering. Although this depression-related impairment in controlled recollection was closely mimicked by data from participants in the rumination condition, it is not certain that the “waiting” students also ruminated during that period. For some, the break from the ongoing experiment might have invited mind blanking. As a learned response to stress, the habit to “tune out” might be just as reflective of poor self-control as the habit to ruminate. Watts, MacLeod, and Morris (1988) found that blanking tended to occur in tasks that involve planning (tapping frontal functions). Although we know too little about blanking in comparison to mind wandering or rumination, either type of habit should take over during periods of lax external control of the procedures to be performed. As I have argued previously (1997, 2000), lax external control often typifies real-world thinking, and it sometimes characterizes the trials that constitute the orienting task in laboratory experiments. The orienting task is, according to effort metaphors and other processing frameworks, the critical period in determining whether the material will be recalled later. Therefore, in allowing sufficient time on each trial for the elaborative processing that benefits recall, memory researchers also encourage habits of mind wandering or blanking. Years ago, Stephanie Rude and I used a simple manipulation of either allowing an orienting decision to be made at any time during an 8-second exposure or requiring the decision at the end of the trial, along with a repetition of the target word. (The decision concerned whether a target word fit sensibly into a sentence frame.) We called the
206
two conditions unfocused and focused, respectively (Hertel & Rude, 1991a). Inthe focused condition, the materials did not stay on the screen for the 8-second period, as they did in the unfocused condition. To perform each trial in the focused condition, the participants had to hold the word and decision in mind for the duration of the trial, but their minds could wander in the unfocused condition because decisions could be reported early on during the 8 seconds. On the subsequent test of free recall, depressed participants from the unfocused condition performed poorly, compared to their nondepressed counterparts. This deficit, however, was eliminated for participants who had been required to focus during the orienting task. The depressed participants’ ability to perform the task could not ever have been at issue. The task was not difficult in either condition (even though it had been used previously as an example of an effortful orienting task by Tyler et al., 1979, and Ellis, Thomas, & Rodriguez, 1984). Indeed, the more difficult condition clearly was the focused one, because it required sustained attention. Therefore, instead of making the task easier to accomplish, the focusing instruction structured the task transfer-appropriately. Participants were required to attend to the recent past (a few seconds earlier) in the same way that they were later required to attend to the past during the recall test. Moreover, external control compensated for poor self-control by not allowing habits of thought (or no-thought) to carry the day. There were a number of variations of this attention-focusing procedure (Hertel & Rude, 1991b). We also tried to stiffen the requirements at the time of the test by using a forced-recall procedure. Although “forcing” the participants to write a certain number of words on the recall test merely increased levels of recall for everyone in our experiments, others have found that this procedure eliminated the depression-related deficit (e.g., Murray et al., 1999). The forcing procedure counteracts a possibly conservative response tendency, but it does so by asking the participant to refocus attention on the past events, which might be at least as important as overcoming hesitancy or conservatism. Other methods of improving performance by providing external support have not always been successful. For example, in the recognition experiment with process-dissociation procedures, Milan and I reinstated the original context word in some test conditions. That procedure boosted estimates of controlled recollection for both dysphoric and nondysphoric participants (Hertel & Milan, 1994). Clearly, the nondysphoric students had done something that we had failed to capture by the manipulation. This was also true when recognition was much more strategically dependent (Hertel & Hardin, 1990). Hardin and I noticed that the nondysphoric students seemed to be using their performance on the prior indirect test of homophone-spelling to decide whether the recognition item had occurred in the original set of questions (e.g., Week/Weak? Did I just spell that word? How did I spell it? And was that word in a question? Oh, name the days of the week? Yes.). In a subsequent experiment, we asked participants to answer a similar set of questions before making each recognition judgment. By doing so,
207
we found a pattern of dependency in the dysphoric group that mimicked the pattern of nondysphoric judgments under unassisted conditions, but the nondysphoric participants who got the set of question cues outwitted us. Their performance improved even for those items that they had spelled in the opposite way (e.g., Week/Weak? Did I just spell that word? Weak? Is there another way to spell it?). These and similar differences illustrate the difficulty in predicting exactly what people do under conditions of lax external control, but the findings also show that performance in depressed states can be improved by at least some version of external support. And the main point is that this support works because it forestalls the emergence of cognitive habits in depression—habits that interfere under poorly controlled conditions and make it difficult for the depressed person to take control and show cognitive initiative (see Hertel, 2000). Habits of thought are difficult to oppose. The historically important construct of habit strength in Hull’s theory of learning (1943) was opposed by the construct of inhibition, but Hull’s rats experienced immediate consequences—missing reward—that led to the orderly development of inhibitory strength. The consequences of thought are often much farther removed, so suppressing a turn of the mind is perhaps much harder to learn than is suppressing a turn in the maze. Modern approaches to thought suppression acknowledge this difficulty. Wegner’s ironic-processes theory (1994), for example, includes a construct called a monitor that checks whether thoughts are being successfully suppressed. According to this point of view, the depressed person’s need to suppress self-concerns, particularly during tasks that require self-control and initiative, presumably “activates” the monitor and ironically brings those concerns back to mind. At the very least, any attention devoted to the task of suppression likely disrupts procedures that require sustained attention or planning. Instructions to suppress are therefore much harder to follow when not accompanied by a task that directs attention through external means. The short version of that argument is that simple self-controlled thought suppression likely does not work well for depressed people, both in the sense of potentiating later forgetting (Hertel & Gerstle, in press) or freeing attention to focus in ways that later benefit remembering. In the latter case, what works best is environmental support, but the world is not often arranged to be helpful. What also seems to work in the sense of improving both memory and momentary mood is a good distraction procedure (e.g., Hertel, 1998; Lyubormirsky et al., 1998). Distraction at least interrupts ongoing habitual thoughts so that the depressed person begins the recall task on even footing. There is even some reason to wonder if reappraisals of intrusive thoughts, under some circumstances, might aid recall of nonnegative events through temporary mood improvement or altered perspective (in the manner found by Rusting & DeHart, 2000; also see Parrott & Spackman, 2000). However, are these measures enough? According to Teasdale et al. (2000) and others, approximately 80% of people diagnosed with major depressive disorder experience recurrent depressive epi-
208
sodes. Practical considerations therefore encourage us to consider not merely the opposition of habitual thinking by controlled procedures but the training of new habits through practice in control. Virtually the only line of research with that general aim is the research on mindfulness training, adapted from the meditation work of Kabat-Zinn by Teasdale, Williams, and Segal. The training consists of a sustained focus on some simple act (typically breathing). When attention wanders, the individual acknowledges the new thought or feeling but then redirects attention to the simple act. It differs from simple thought suppression in two main respects: acknowledgment of the thought and practice in redirection. Do the benefits of such training extend to long-term habits of control? So far, Williams et al. (2000) have found that mindfulness training boosts the effects of cognitive-behavioral therapy in reducing overly general autobiographical memory by formerly depressed participants. Moreover, for people with at least three prior episodes of depression, the addition of mindfulness training to cognitive-behavioral therapy significantly reduced recurrence during the subsequent year (Teasdale et al., 2000). These studies, of course, are merely a beginning. In short, the best antidote to maladaptive habits is a new set of habits—not the opposite sort of habits recommended by Pollyanna and the teachings of Norman Vincent Peale (1956) and not the habits of suppression as recommended by certain grandmothers but the habits of thought control. Cognitive control, as conceived by cognitive psychologists, is a concept that opposes habit, in that habitual procedures require little attention. But if the necessary control procedures—awareness of the wandering stream of thought and subsequent disengagement, for example—are practiced “religiously,” they too can become habitual. In the absence of mind wandering and rumination, memory biases are less extreme, and memory impairment is repaired.
Notes 1. Induction procedures are particularly useful when it is difficult to determine the content of natural thought, although some procedures do not produce memorial effects that mimic patterns in natural depression. For a case in which they fail to show mood-congruent recall in the same study in which it was obtained for dysphoric students, see Kwiatkowski and Parkinson (1994; also see Eich & McCaulay, 2000; Perrig & Perrig, 1988). Inductions have also been used to study mood-dependent memory: better memory when mood at initial encounter matches mood at test. Mood-dependent memory has also been studied with participants diagnosed with bipolar disorder. When the “mood” during initial encounter (mania or depression) matched the mood at test, higher levels of recall or recognition were found (Eich, Macaulay, & Lam, 1997; Weingartner, Miller, & Murphy, 1977). See Blaney (1986, p. 237) for a discussion of how mood-dependent memory can be seen as a special case of mood-congruent memory. 2. Ironically, overgeneral tendencies have also been correlated with spontaneous
209
intrusions of stressful memories (Brewin, Reynolds, & Tata, 1999; Wessel, Merckelbach, & Dekkers, 2002). 3. In one exception to the early rule, Elliott and Greene (1992) did show impaired word-stem completion and homophone spelling but omitted unprimed items on the test, which might have encouraged participants’ awareness of the memorial goal and a concomitant use of memory as object. (See Bazin et al., 1994, and Roediger & McDermott, 1992, for full discussions of their design.) 4. Of course, attention might well be impaired during initial exposure. Consider a report by Rokke, Arnell, Koch, and Andrews (2002), who investigated the size of attentional blinks in a rapid-serial-visual-presentation paradigm (RSVP) administered to students with varying degrees of dysphoria. Moderately to severely dysphoric participants performed relatively well when the only task was to detect the presence of a single letter in the stream of rapidly presented letters. However, their performance suffered more, compared to that of nondysphoric or mildly dysphoric participants, when the task also involved attention to a letter occurring earlier in the stream (a deeper attentional blink). This outcome implicates fundamental deficits in redirecting attention. 5. Some of the results reported in this section should be viewed cautiously, given either unexpected differences on baseline trials (e.g., Joormann, in press) or lack of information about baseline (e.g., Bradley et al., 1997). 6. Studies using this directed forgetting paradigm have revealed interesting differences associated with clinical disorders. For example, Sonntag et al. (2003) have shown that schizophrenics fail to show effects of directed forgetting on measures of recollection (see chapter 7). McNally, Metzger, Lasko, Clancy, and Pitman (1998) found poor directed forgetting of trauma words by women with posttraumatic stress disorder, as well as deficits in remembering neutral and positive words on the lists to be remembered (see chapter 4).
References Abramson, L. Y., Metalsky, G. I., & Alloy, L. B. (1989). Hopelessness depression: A theory-based subtype of depression. Psychological Review, 96, 358–372. Alloy, L. B., Abramson, L. Y., & Francis, E. L. (1999). Do negative cognitive styles confer vulnerability to depression? Current Directions in Psychological Science, 8, 128–132. Anderson, M. C., & Green, C. (2001). Suppressing unwanted memories by executive control. Nature, 410, 366–369. Anderson, M. C., & Neely, J. H. (1996). Interference and inhibition in memory retrieval. In E. L. Bjork & R. A. Bjork (Eds.), Memory (pp. 237–313). New York: Academic Press. Baddeley, A., & Hitch, G. J. (1974). Working memory. In G. Bower (Ed.), The psychology of learning and motivation (Vol. 7, pp. 47–89). New York: Academic Press. Bazin, N., Perruchet, P., De Bonis, M., & Féline, A. (1994). The dissociation of explicit and implicit memory in depressed patients. Psychological Medicine, 24, 239–245.
210
Beck, A. T. (1967). Depression: Clinical, experimental and theoretical aspects. New York: Harper & Row. Beck, A. T., Ward, C., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561–571. Blaney, P. H. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229– 246. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129–148. Bradley, B., & Mathews, A. (1983). Negative self-schemata in clinical depression. British Journal of Clinical Psychology, 22, 173–181. Bradley, B. P., Mogg, K., & Lee, S. C. (1997). Attentional biases for negative information in induced and naturally occurring dysphoria. Behaviour Research and Therapy, 35, 911–927. Brewin, C. R., Reynolds, M., & Tata, P. (1999). Autobiographical memory processes and the course of depression. Journal of Abnormal Psychology, 108, 511–517. Brown, T. A., Di Nardo, P. A., Lehman, C. L., & Campbell, L. A. (2001). Reliability of DSM-IV anxiety and mood disorders: Implications for the classification of emotional disorders. Journal of Abnormal Psychology, 110, 49–58. Burt, D. B., Zembar, M. J., & Niederehe, G. (1995). Depression and memory impairment: A meta-analysis of the association, its pattern, and specificity. Psychological Bulletin, 117, 285–305. Channon, S. (1996). Executive dysfunction in depression: The Wisconsin Card Sorting Test. Journal of Affective Disorders, 39, 107–114. Chapman, L. J., & Chapman, J. P. (1973). Problems in the measurement of cognitive deficits. Psychological Bulletin, 79, 380–385. Clark, D. M., & Teasdale, J. D. (1982). Diurnal variation in clinical depression and accessibility of memories of positive and negative experiences. Journal of Abnormal Psychology, 91, 87–95. Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11, 671–684. Danion, J.-M., Willard-Schroeder, D., Zimmermann, M.-A., Grange, D., Schlienger, J.-L., & Singer, L. (1991). Explicit memory and repetition priming in depression. Archives of General Psychiatry, 48, 707–711. Davidson, R. J. (2000). Affective style, mood, and anxiety disorders: An affective neuroscience approach. In R. J. Davidson (Ed.), Anxiety, depression, and emotion (pp. 88–108). New York: Oxford University Press. Davis, R. N., & Nolen-Hoeksema, S. (2000). Cognitive inflexibility among ruminators and nonruminators. Cognitive Therapy and Research, 24, 699–711. Deldin, P. J., Keller, J., Gergen, J. A., & Miller, G. A. (2001). Cognitive bias and emotion in neuropsychological models of depression. Cognition & Emotion, 15, 787–802. Denny, E. B., & Hunt, R. R. (1992). Affective valence and meory in depression: Dissociation of recall and fragment completion. Journal of Abnormal Psychology, 101, 575–580. Derry, P. A., & Kuiper, N. A. (1981). Schematic processing and self-reference in clinical depression. Journal of Abnormal Psychology, 90, 286–297. Dobson, K. S., & Shaw, B. F. (1987). Specificity and stability of self-referent encoding in clinical depression. Journal of Abnormal Psychology, 96, 36–40.
211
Eich, E., & Macaulay, D. (2000). Are real moods required to reveal mood-congruent and mood-dependent memory? Psychological Science, 11, 244–248 Eich, E., Macaulay, D., & Lam, R. W. (1997). Mania, depression, and mood dependent memory. Cognition & Emotion, 11, 607–618. Elliott, C. L., & Greene, R. L. (1992). Clinical depression and implicit memory. Journal of Abnormal Psychology, 101, 572–574. Ellis, H. C., & Ashbrook, P. W. (1988). Resource allocation model of the effects of depressed mood states on memory. In K. Fiedler & J. Forgas (Eds.), Affect, cognition and social behavior (pp. 25–43). Toronto: Hogrefe. Ellis, H. C., Thomas, R. L., & Rodriguez, I. A. (1984). Emotional mood states and memory: Elaborative encoding, semantic processing, and cognitive effort. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 470– 482. Fennell, M. J. V., Teasdale, J. D., Jones, S., & Damle, A. (1987). Distraction in neurotic and endogenous depression: An investigation of negative thinking in major depressive disorder. Psychological Medicine, 17, 441–452. Fox, E., Russo, R., Bowles, R., & Dutton, K. (2001). Do threatening stimuli draw or hold attention in subclinical anxiety? Journal of Experimental Psychology: General, 130, 681–700. Gotlib, I. H., & McCann, C. D. (1984). Construct accessibility and depression: An examination of cognitive and affective factors. Journal of Personality and Social Psychology, 47, 427–439. Haaga, D. A., Dyck, M. J., & Ernst, D. (1991). Empirical status of the cognitive theory of depression. Psychological Bulletin, 110, 215–236. Hartlage, S., Alloy, L. B., Vazquez, C., & Dykman, B. (1993). Automatic and effortful processing in depression. Psychological Bulletin, 113, 247–278. Hasher, L., & Zacks, R. T. (1979). Automatic and effortful processes in memory. Journal of Experimental Psychology: General, 108, 356–388. Hasher, L., & Zacks, R. T. (1988). Working memory, comprehension, and aging: A review and a new view. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 22, pp. 193–225). San Diego, CA: Academic Press. Healy, H., & Williams, J. M. G. (1999). Autobiographical memory. In T. Dalgleish & M. Power (Eds.), The handbook of cognition and emotion (pp. 229–242). Chichester: Wiley. Heller, W., & Nitschke, J. B. (1998). The puzzle of regional brain activity in depression and anxiety: The importance of subtypes and comorbidity. Cognition & Emotion, 12, 421–447. Hertel, P. T. (1994). Depressive deficits in word identification and recall. Cognition & Emotion, 8, 313–327. Hertel, P. T. (1997). On the contributions of deficient cognitive control to memory impairments in depression. Cognition & Emotion, 11, 569–584. Hertel, P. T. (1998). The relationship between rumination and impaired memory in dysphoric moods. Journal of Abnormal Psychology, 107, 166–172. Hertel, P. T. (2000). The cognitive-initiative account of depression-related impairments in memory. In D. Medin (Ed.), The psychology of learning and motivation (Vol. 39, pp. 47–71). New York: Academic Press.
212
Hertel, P. T., & Gerstle, M. (in press). Depression-related deficits in forgetting. Psychological Science. Hertel, P. T., & Hardin, T. S. (1990). Remembering with and without awareness in a depressed mood: Evidence of deficits in initiative. Journal of Experimental Psychology: General, 119, 45–59. Hertel, P. T., & Knoedler, A. J. (1996). Solving problems by analogy: The benefits and detriments of hints and depressed moods. Memory & Cognition, 24, 16–25. Hertel, P. T., Mathews, A., Peterson, S., & Kintner, K. (in press). Transfer of training emotionally biases interpretations. Applied Cognitive Psychology. Hertel, P. T., & Meiser, T. (2000). Capacity and procedural accounts of impaired memory in depression. In U. von Hecker, S. Dutke, & G. Sedek (Eds.), Generative mental processes and cognitive resources: Integrative research on adaptation and control (pp. 283–308). Dordrecht, The Netherlands: Kluwer Academic Publishers. Hertel, P. T., & Milan, S. (1994). Depressive deficits in recognition: Dissociation of recollection and familiarity. Journal of Abnormal Psychology, 103, 736–742. Hertel, P. T., & Parks, C. (2002). Emotional episodes facilitate word recall. Cognition & Emotion, 16, 685–694. Hertel, P. T., & Rude, S. S. (1991a). Depressive deficits in memory: Focusing attention improves subsequent recall. Journal of Experimental Psychology: General, 120, 301–309. Hertel, P. T., & Rude, S. S. (1991b). Recalling in a state of natural or induced depression. Cognitive Therapy and Research, 15, 103–127. Hull, C. L. (1943). Principles of behavior. New York: Appleton-Century-Crofts. Ingram, R. E. (1990). Self-focused attention in clinical disorders: Review and a conceptual model. Psychological Bulletin, 107, 156–176. Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30, 513–541. Jacoby, L. L., Jennings, J. M., & Hay, J. F. (1996). Dissociating automatic and consciously controlled processes: Implications for diagnosis and rehabilitation of memory deficits. In D. Herrmann, C. McEvoy, C. Herzog, P. T. Hertel, & M. K. Johnson (Eds.), Basic and applied memory research: Theory in context (pp. 161– 194). Mahwah, NJ: Erlbaum. Jacoby, L. L., & Kelley, C. M. (1987). Unconscious influences of memory for a prior event. Personality and Social Psychology Bulletin, 13, 314–336. Jenkins, W., & McDowall, J. (2001). Implicit memory and depression: An analysis of perceptual and conceptual processes. Cognition & Emotion, 15, 803–812. Joormann, J. (in press). Attentional bias in dysphoria: The role of inhibitory processes. Cognition & Emotion. Kwiatkowski, S. J., & Parkinson, S. R. (1994). Depression, elaboration, and mood congruence: Differences between natural and induced mood. Memory & Cognition, 22, 225–233. Lawson, C., MacLeod, C., & Hammond, G. (2002). Interpretation revealed in the blink of an eye: Depressive bias in the resolution of ambiguity. Journal of Abnormal Psychology, 111, 321–328. Lyubomirsky, S., Caldwell, N. D., & Nolen-Hoeksema, S. (1998). Effects of rumina-
213
tive and distracting responses to depressed mood on retrieval of autobiographical memories. Journal of Personality and Social Psychology, 75, 166–177. Matt, G. E., Vazquez, C., & Campbell, W. K. (1992). Mood-congruent recall of affectively toned stimuli: A meta-analytic review. Clinical Psychology Review, 12, 227–255. McCabe, S. B., & Gotlib, I. H. (1995). Selective attention and clinical depression: Performance on a deployment-of-attention task. Journal of Abnormal Psychology, 104, 241–245. McCabe, S. B., & Toman, P. E. (2000). Stimulus exposure duration in a deploymentof-attention task: Effects on dysphoric, recently dysphoric, and nondysphoric individuals. Cognition & Emotion, 14, 125–142. McNally, R. J., Metzger, L. J., Lasko, N. B., Clancy, S. A., & Pitman, R. K. (1998). Directed forgetting of trauma cues in adult survivors of childhood sexual abuse with and without posttraumatic stress disorder. Journal of Abnormal Psychology, 107, 596–601. Morris, C. D., Bransford, J. P., & Franks, J. J. (1977). Levels of processing versus transfer-appropriate processing. Journal of Verbal Learning and Verbal Behavior, 16, 519–533. Murray, L. A., Whitehouse, W. G., & Alloy, L. B. (1999). Mood congruence and depressive deficits in memory: A forced-recall analysis. Memory, 7, 175–196. Neill, W. T., & Mathis, K. M. (1998). Transfer-inappropriate processing: Negative priming and related phenomena. In D. Medin (Ed.), The psychology of learning and motivation (Vol. 38, pp. 1–44). New York: Academic Press. Nolen-Hoeksema, S. (1991). Responses to depression and their effects on the duration of depressive episodes. Journal of Abnormal Psychology, 100, 569–582. Nolen-Hoeksema, S. (2000). The role of rumination in depressive disorders and mixed anxiety/depressive symptoms. Journal of Abnormal Psychology, 109, 504–511. Nolen-Hoeksema, S., & Morrow, J. (1993). The effects of rumination and distraction on naturally occurring depressed moods. Cognition & Emotion, 7, 561–570. Parrott, W. G., & Hertel, P. T. (1999). Research methods in cognition and emotion. In T. Dalgleish & M. Power (Eds.), The handbook of cognition and emotion (pp. 61– 81). Chichester: Wiley. Parrott, W. G., & Spackman, M. P. (2000). Emotion and memory. In M. Lewis & J. Haviland-Jones (Eds.), Handbook of emotions (2nd ed., pp. 476–490). New York: Guilford. Peale, N. V. (1956). The power of positive thinking. Englewood Cliffs, NJ. Perrig, W. J., & Perrig, P. (1988). Mood and memory: Mood-congruity effects in absence of mood. Memory & Cognition, 16, 102–109. Polyani, M. (1958). Personal knowledge: Toward a post-critical philosophy. Chicago: University of Chicago Press. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information processing and cognition: The Loyola Symposium (pp. 55– 85). Hillsdale, NJ: Erlbaum. Powers, M. J., Dalgleish, T., Claudio, V., Tata, P., & Kentish, J. (2000). The directed forgetting task: Application to emotionally valent material. Journal of Affective Disorders, 57, 147–157.
214
Roediger, H. L., & McDermott, K. B. (1992). Depression and implicit memory: A commentary. Journal of Abnormal Psychology, 101, 587–591. Rokke, P. D., Arnell, K. M., Koch, M. D., & Andrews, J. T. (2002). Dual-task attention deficits in dysphoric mood. Journal of Abnormal Psychology, 111, 370–379, Rothkopf, J. S., & Blaney, P. H. (1991). Mood congruent memory: The role of affective focus and gender. Cognition & Emotion, 5, 53–64. Rude, S. S., Hertel, P. T., Jarrold, W., Covich, J., & Hedlund, S. (1999). Remembering to remember: A difficulty for depressed individuals? Cognition & Emotion, 13, 267–276. Rude, S. S., Wenzlaff, R. M., Gibbs, B., Vane, J., & Whitney, T. (2002). Negative processing biases predict subsequent depressive symptoms. Cognition & Emotion, 16, 423–440. Ruiz-Caballero, J. A., & González, P. (1994). Implicit and explicit memory bias in depressed and nondepressed subjects. Cognition & Emotion, 8, 555–569. Rusting, C. L., & DeHart, T. (2000). Retrieving positive memories to regulate negative mood: Consequences for mood-congruent memory. Journal of Personality and Social Psychology, 78, 737–752. Sonntag, P., Gokalsing, E., Olivier, C., Robert, P., Burglen, F., Kauffmann-Muller, F., et al. (2003). Impaired strategic regulation of contents of conscious awareness in schizophrenia. Consciousness and Cognition, 12, 190–200. Teasdale, J. D. (1983). Negative thinking in depression: Cause, effect, or reciprocal relationship? Advances in Behavior Research and Therapy, 5, 3–26. Teasdale, J. D., Segal, Z. V., Williams, J. M. G., Ridgeway, V. A., Soulsby, J. M., & Lau, M. A. (2000). Prevention of relapse/recurrence in major depression by mindfulness-based cognitive therapy. Journal of Consulting and Clinical Psychology, 68, 615–623. Tyler, S. W., Hertel, P. T., McCallum, M. C., & Ellis, H. C. (1979). Cognitive effort and memory. Journal of Experimental Psychology: Human Learning and Memory, 5, 607–617. Von Hecker, U., Conway, M., Meiser, T., & Holm, S. (2002). Inhibition deficits in dysphoric mood. Manuscript submitted for publication. Von Hippel, W., Hawkins, C., & Narayan, S. (1994). Personality and perceptual expertise: Individual differences in perceptual identification. Psychological Science, 5, 401–406. Watkins, E., & Teasdale, J. D. (2001). Rumination and overgeneral memory in depression: Effects of self-focus and analytic thinking. Journal of Abnormal Psychology, 110, 333–357. Watkins, P. C., Martin, C. K., & Stern, L. D. (2000). Unconscious memory bias in depression: Perceptual and conceptual processes. Journal of Abnormal Psychology, 109, 282–289. Watkins, P. C., Mathews, A., Williamson, D. A., & Fuller, R. D. (1992). Mood-congruent memory in depression: Emotional priming or elaboration? Journal of Abnormal Psychology, 101, 581–586. Watkins, P. C., Vache, K., Verney, S. P., Muller, S., & Mathews, A. (1996). Unconscious mood-congruent memory bias in depression. Journal of Abnormal Psychology, 105, 34–41.
215
Watts, F. N., MacLeod, A. K., & Morris, L. (1988). Association between phenomenal and objective measures of concentration problems in depressed patients. British Journal of Psychology, 79, 241–250. Weingartner, H., Cohen, R. M., Murphy, D. L., Martello, J., & Gerdt, C. (1981). Cognitive processes in depression. Archives of General Psychiatry, 38, 42–47. Weingartner, H., Miller, H., & Murphy, D. L. (1977). Mood-state-dependent retrieval of verbal associations. Journal of Abnormal Psychology, 86, 276–284. Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review, 101, 34–52. Wenzlaff, R. M., & Bates, D. E. (1998). Unmasking a cognitive vulnerability to depression: How lapses in mental control reveal depressive thinking. Journal of Personality and Social Psychology, 75, 1559–1571. Wenzlaff, R. M., Meier, J., & Salas D. M. (2002). Thought suppression and memory biases during and after depressive moods. Cognition & Emotion, 16, 403–422. Wenzlaff, R. M., Rude, S. S., Taylor, C. J., Stultz, C. H., & Sweatt, R. A. (2001). Beneath the veil of thought suppression: Attentional bias and depression risk. Cognition & Emotion, 15, 435–452. Wessel, I., Merckelbach, H., & Dekkers, T. (2002). Autobiographical memory specificity, intrusive memory, and general memory skills in Dutch-Indonesian survivors of the World War II era. Journal of Traumatic Stress, 15, 227–234. Williams, J. M. G., Teasdale, J. D., Segal, Z. V., & Soulsby, J. (2000). Mindfulnessbased cognitive therapy reduces overgeneral autobiographical memory in formerly depressed patients. Journal of Abnormal Psychology, 109, 150–155. Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1988). Cognitive psychology and emotional disorders. New York: Wiley.
216
7
, , - , , ,
T
he conceptualization of schizophrenia has changed radically over the past decade. Cognitive deficits are now viewed as core symptoms and major disabilities of the disease (Green, 1996); virtually all cognitive functions are considered defective (Aleman, Hijman, de Haan, & Kahn, 1999; Heinrichs & Zakzanis, 1998). Among these deficits, memory impairments likely play a crucial role. Not all cognitive abilities are equally affected; several studies have shown that memory functions are disproportionately impaired (e.g., Bilder et al., 2000; Gold, Randolph, Carpenter, Goldberg, & Weinberger, 1992; Heinrichs & Zakzanis, 1998; Palmer et al., 1997; Saykin et al., 1991, 1994). Moreover, memory impairments are strongly correlated with disturbed functioning in everyday life (Green, 1996), and a large part of clinical work (e.g., collecting historical information; conducting psychodynamic, behavioral, or cognitive therapies; ensuring medication compliance) relies on the ability of the patient to learn and remember. In normal subjects, the evidence that memory is strongly influenced by emotions is overwhelming (e.g., Blaney, 1986; Bower, 1981; Conway & Pleydell-Pearce, 2000). Memory for emotional events is better than memory for neutral events, a phenomenon known as the emotionality effect (e.g., Dutta & Kanungo, 1975; Rubin & Friendly, 1986). Moreover, memory for positive events is often better than memory for negative events—a difference attributed to the Pollyanna tendency (Boucher & Osgood, 1969), a disposition for normal subjects to construct an affectively positive world view. Finally, normal subjects exhibit better memory for events whose emotional valence is congruent with their current mood (mood-congruent memory). How do memory and emotion interact in patients with schizophrenia? Do these patients exhibit a particular difficulty in remembering emotional events? Do they still show the Pollyanna tendency? Is there any impairment of the subjective experience of retrieving 217
emotional memories? Because emotional disturbances, which interfere dramatically with behavior control and with social interaction, are another major aspect of schizophrenia (Taylor & Liberzon, 1999), the answers to these questions are crucial for the development of cognitive models of schizophrenia that take emotions into account. They may help us identify the mechanisms underlying the formation and the persistence of poorly understood symptoms comprising a strong emotional component, such as delusional experience. In this chapter, we begin by reviewing the studies of memory impairment and emotion disturbances in schizophrenia. Then we discuss experimental investigations of how memory and emotion interact in schizophrenia. Because conscious awareness, which plays a crucial role both in memory and emotion, may represent the fundamental impairment in schizophrenia (Andreasen, 1999; Danion, Huron, & Robert, 2001; Frith, 1992), we emphasize not only objective accuracy of memory but also states of awareness associated with emotional memories.
Memory Impairment in
Schizophrenia
The Dissociation of Performance in
Explicit and Implicit Memory Tasks
The long-term memory impairment associated with schizophrenia is selective. Whether patients with schizophrenia display a deficit of performance depends on the task. Several studies have shown that patients display normal or quasinormal performance in implicit memory tasks—that is, when participants are not asked to retrieve material consciously. This pattern has been demonstrated in tasks exploring perceptual priming (Clare, McKenna, Mortimer, & Baddeley, 1993; Doniger, Silipo, Rabinowicz, Snodgrass, & Javitt, 2001; Gras-Vincendon et al., 1994), conceptual priming (Schwartz & Winstead, 1985), and procedural memory (Clare et al., 1993; Dominey & Georgieff, 1997; Michel, Danion, Grangé, & Sandner, 1998; Schwartz, Rosse, Veazey, & Deutsch, 1996). It has also been demonstrated in implicit learning tasks (Danion, Meulemans, KauffmannMuller, & Vermaat, 2001). In accordance with these findings, Kazès et al. (1999), who used the process dissociation procedure (Jacoby, 1991) to obtain uncontaminated estimates of memory processes, found that automatic influences of memory, but not consciously controlled uses of memory, were intact in schizophrenia. In contrast to normal performance in implicit memory tasks, impaired performance is typically observed in explicit or episodic memory tasks. This impairment has been consistently demonstrated in free-recall (Koh & Peterson, 1978), cuedrecall (Schwartz, Rosse, & Deutsch, 1993), and, to a lesser degree, recognition tasks (Aleman et al., 1999; Calev, 1984a, 1984b). Episodic memory impairment
218
occurs for verbal material, such as words, sentences (Koh & Peterson, 1978), and stories (Abbruzzese & Scarone, 1993); visual material, such as faces (Addington & Addington, 1998); and contextual information, such as space and time (Rizzo, Danion, Van der Linden, & Grangé, 1996; Rizzo, Danion, Van Der Linden, Grangé, & Rohmer, 1996). Autobiographical memories (memories for personal events and facts from one’s life) are also affected. Patients with schizophrenia display an abnormal pattern of memories when theirs are compared to those of healthy individuals: the most severe impairment occurs for early adulthood memories, childhood memories are quasi-normal, and recent past memories are deficient to a lesser extent (Feinstein, Goldberg, Nowlin, & Weinberger, 1998). Finally, despite difficulties with explicit memory, patients with schizophrenia remain able to learn (Goldberg, Weinberger, Pliskin, Berman, & Podd, 1989; Hawkins, 1999), although their learning curve is lower than normal (Aleman et al., 1999).
Recognition Memory
and Associated Subjective States
of Conscious Awareness
Most of the memory investigations carried out in normal controls and patients with schizophrenia primarily concern the objective accuracy of recall and recognition of stimuli. However, what often matters most is not the objective fact that we can say an event occurred but what it feels like and means to us to remember what took place. Recently, in recognition of the importance of subjective experience, many researchers have shifted their focus from questions about the absolute accuracy of memories and toward questions about the states of awareness accompanying memories. The notion of conscious awareness is pivotal to the distinction between explicit and implicit tasks. Therefore, we can infer from the pattern of learning and memory dysfunction observed in schizophrenia that memory is impaired when conscious awareness is required at encoding or at retrieval and is spared when conscious awareness is not required. In other words, conscious awareness might be the critical determinant of memory impairment both at encoding and at retrieval. However, this inference can be questioned because explicit and implicit memory tasks assess conscious awareness only indirectly, from an objective, or third-person, perspective. To obtain direct evidence of impaired conscious awareness, we should investigate the experiences of an individual from a subjective, or first-person, perspective. Tulving (1985) was the first to propose a first-person approach to conscious awareness associated with recognition memory. According to Tulving, conscious awareness associated with recognition memory is not a unitary phenomenon. It includes at least two distinct subjective states that may be investigated experimentally using the Remember/Know procedure (R/K; Gardiner, Java, & RichardsonKlavehn, 1996; Tulving, 1985). In a recognition task, the participants are asked
, ,
219
to report their subjective state of awareness at the time they recognize each individual item. They are instructed to make a Remember response if they consciously recollect something they experienced when they learned the item, that is, if they relive mentally the learning episode. This qualitatively rich mental experience includes perceptual, spatial, temporal, semantic, and emotional details that are attributed to a past event (Johnson, Hashtroudi, & Lindsay, 1993; Johnson & Raye, 1981). The participants are instructed to make a Know response if recognition is accompanied by feelings of familiarity yet no specific memories for the learning episode. Many experiments carried out with normal controls and the R/K procedure have shown functional dissociations between Remember and Know responses; these findings indicate that the two responses index two distinct states of conscious awareness (Gardiner et al., 1996; Gardiner & RichardsonKlavehn, 2000; Rajaram & Roediger, 1997; Yonelinas, Kroll, Dobbins, Lazzara, & Knight, 1998). Remember responses are influenced by variables that increase the salience of the events, or make them more distinctive: For instance, a deeper level of processing enhances the frequency of Remember, but not Know, responses (Gardiner, 1988). Know responses, on the other hand, depend on the ease with which an item can be processed either perceptually or conceptually: Repetition test priming increases Know responses, without influencing Remember responses (Rajaram, 1993). Moreover, neuropsychological data from brain-lesioned patients and brain imaging studies carried out in normal controls indicate that Remember and Know responses rely on different neural correlates (Duzel, Yonelinas, Mangun, Heinze, & Tulving, 1997; Henson, Rugg, Shallice, Josephs, & Dolan, 1999; Smith, 1993). Using the R/K procedure, Huron et al. (1995; also see Huron, Danion, Rizzo, Killofer, & Damiens, in press) and Danion, Rizzo, and Bruant (1999) studied patients with schizophrenia. They showed that recognition performance of patients was associated with low levels of Remember, but not Know, responses, as compared to performance of normal controls. Impairment of conscious recollection was attributed to a failure of strategic processes at encoding, at retrieval, or at both points. Episodic memories are not a literal reproduction of the past but instead depend on constructive and reconstructive processes that are sometimes prone to errors, distortions, and illusions (e.g., Schacter, Norman, & Koutstaal, 1998). Recently developed paradigms for investigating false memories (memories for events that never happened) have been used to address issues in constructive memory. Using both the R/K procedure and the procedure initially introduced by Deese (1959) and subsequently modified by Roediger and McDermott (1995) to investigate false memories, Huron and Danion (2002) showed that the frequency of false memories was lower in patients with schizophrenia than in normal controls. These low levels of false memories were associated with a reduced frequency of Remember, but not Know, responses. Danion, Gokalsing, Robert, Massin-Krauss, and Bacon (2001) also showed that schizophrenia impairs the relationship between subjective experience and behavior, suggesting that pa-
220
tients’ behavior is less determined by subjective experience, and less controlled by intentions, than that of normal controls. This is reminiscent of the traditional view that schizophrenia is characterized by a dissociation between thought and action (Bleuler, 1911). Because conscious recollection provides a basis for decision and action and a foundation for social relationships, these findings may have important clinical implications. Reduced conscious recollection leads to poor ability to apprehend subjective experiences in all their richness and may account for many behavioral disturbances exhibited by patients with schizophrenia (Danion, Huron, et al., 2001).
Emotional Disturbances
in Schizophrenia
According to a number of authors (e.g., Damasio, 1995; Ekman, 1993; Lazarus, 1991; Mandler, 1984; Zajonc, 1984), emotions are best conceptualized as a concatenation of cognitive evaluation with changes in body and brain state, arising in response to particular events or mental images. As the consequences of how people construe situations, emotions differ according to the complexity of the cognitive evaluation from which they stem. In some situations, this evaluation may be rudimentary, whereas in others the full complex emotion can develop only with substantial reasoning about the situation and its implications. Body state changes correspond to physiological modifications, many perceptible to an external observer; examples include changes in skin color, facial expression, or skin conductance. Brain changes correspond to the activation of specific cerebral regions and produce cognitive processing modifications; some of these modifications—such as the rate at which images are produced or the fluency with which affective stimuli are processed—are perceptible to the individual in whom they are enacted. In addition to these evaluative and physiological components, emotions may involve consciousness when the subject becomes aware of the link between emotion and the causal event. The implication of conscious awareness and its experiential dimension explains why, in humans, emotions are not only closely related to adaptive behavior at an automated level but also offer the flexibility of response based on memory for emotional events from the personal past. A primary function of emotions is to make thought, action, and social interaction more effective (Lazarus, 1991).
Inappropriate or Flat Affect Emotional disturbances such as inappropriate or flat affect are a cardinal symptom of schizophrenia. An inappropriate affect is one that is discordant with the content of the patient’s speech or ideation; sudden and unpredictable changes in affect may also occur. In flat affect, there is little or no affective expression in
, ,
221
general and a reduced range of facial expression in particular. The mechanism underlying this symptom has long been controversial. On the one hand, flat affect has been taken as evidence that the primary emotional disturbance in schizophrenia is an inability to experience pleasure, that is, anhedonia (e.g., Rado, 1962). According to this view, the absence of pleasure is a central feature of schizophrenia; hence, the diminished expression of positive emotions is secondary to anhedonia. On the other hand, classical authors such as Bleuler (1911) noticed that, although patients often report that they experience strong emotions, observers do not always confirm their reports. He believed that patients do experience emotions but that expression is blunted. Experimental evidence relevant to this debate has been obtained from laboratory studies. A disjunction between expressed and subjectively experienced emotions has been shown using emotion-eliciting films: patients with schizophrenia experience as many positive and negative emotions as normal controls, but they are less facially expressive of both positive and negative emotions (Kring, Kerr, Smith, & Neale, 1993; Kring & Neale, 1996). Myin-Germeys, Delespaul, and deVries (2000) recently assessed emotional experience of patients with schizophrenia and normal controls in daily life. They used a structured time-sampling technique to collect the data. Participants received a digital wristwatch and a set of assessment forms. Ten times a day on 6 consecutive days, the watch emitted a beep at unpredictable moments between 7:30 AM and 10:30 PM. After every beep, reports of thoughts, current activity, moods, and severity of symptoms were collected. The assessment forms contained 10 Likert-type items about positive and negative emotions. Patients were divided into blunted and nonblunted subgroups on the basis of Brief Psychiatric Rating Scale behavioral ratings of flat affect. Patients with schizophrenia experienced more intense and more variable negative emotions than normal controls did. In contrast, they experienced less intense and less variable positive emotions. This finding, at variance with evidence from laboratory studies that patients experience the same intensities of positive emotions as normal controls do, may be explained by the absence of social context in laboratory studies. Patients with schizophrenia may indeed be capable of experiencing the same positive emotions as normal controls but, in the naturalistic context of daily life, engage less often in situations likely to elicit positive experiences. These patterns of emotion were observed whether the patients were emotionally blunted or not. In another study that used the same structured time-sampling technique, Myin-Germeys, van Os, Schwartz, Stone, and Delespaul (2001) provided additional evidence that patients with psychotic illness reacted with more intense negative emotions to subjective appraisals of stress in daily life than did controls. In summary, these results bear out the view that patients with schizophrenia are more emotionally active and responsive than what is usually assumed on the basis of behavioral observations and facial modifications perceptible to an external observer. In some circumstances, patients may experience more intense negative emotions and,
222
possibly, less intense positive emotions than controls do. Therefore, the identification of the circumstances in which patients with schizophrenia do experience emotions like controls and those in which they do not is a prerequisite for the investigation of how memory and emotion interact.
Cognitive Evaluation
of Emotional Events
A large body of experimental evidence shows that patients with schizophrenia fail to recognize emotion in others (see the review by Bryson, Bell, & Lysaker, 1997; Walker, Marwit, & Emory, 1980). Whether this deficit reflects specific disturbances or is part of a more generalized impairment, such as, for instance, attentional, executive or evaluative functions, is a matter of debate. Kerr and Neal (1993) have shown that patients with schizophrenia do not exhibit greater impairment in affect recognition than in nonemotional perception tasks. However, evidence from two studies that varied the emotional valence of words in working memory tasks suggests that patients with schizophrenia may still be sensitive to the emotional salience of stimuli. Koh, Szoc, and Peterson (1977) used pleasant, unpleasant, and neutral words in a Sternberg task.1 Despite their overall poorer performance, patients, like normal controls, were faster for neutral words than for pleasant and unpleasant words. Moreover, in both groups, there was no difference between pleasant and unpleasant words. Kay (1982) used a modification of the Brown-Peterson task (Wickens, 1970) to investigate whether patients with schizophrenia were sensitive to the emotional salience of to-be-learned words. Participants were presented with three trials involving word triads (e.g., love, excellent, rest) drawn from a single affective valence (e.g., positive) and then switched to the opposite affective valence on the fourth trial (e.g., negative). After each trial, participants performed a distractor task over a short retention interval before recalling the word triads. Patients with schizophrenia performed progressively worse across trials but returned to the level of the first trial for the fourth trial, namely, when the affective valence shifted. This increase in performance induced by the shift of affective valence indicates a release of proactive interference, suggesting that patients with schizophrenia were still sensitive to the affective salience of the to-be-learned words. This finding is consistent with evidence from studies conducted in normal controls showing that words elicit an emotional evaluation automatically or with minimal strategic processes and conscious cognitive effort (review in Ochsner, 2000). In summary, these findings suggest that patients with schizophrenia exhibit defective emotional processing when a highly complex or very specific cognitive evaluation of the emotional stimuli is required, such as recognizing, discriminating, and interpreting emotions in others. In contrast, patients remain sensitive to the emotional valence of stimuli when cognitive evaluation is more automatic.
, ,
223
Body State Changes Body state changes arising in response to emotional events have been widely investigated in schizophrenia with measurements of skin conductance. Forty to 50% of patients with schizophrenia exhibit a reduced phasic response of skin conductance to orienting stimuli (review in Bernstein, 1987). But patients may also exhibit a tonic hyperactivity of skin conductance (Dawson, Nuechterlein, Schell, Gitlin, & Ventura, 1994). Finally, in the study using emotion-eliciting films, Kring and Neale (1996) showed that, compared to normal controls, patients with schizophrenia evidenced greater reactivity of skin conductance to positive and negative films. Yet they also exhibited greater reactivity to neutral films than controls did. Thus, while they were reporting comparable levels of subjectively experienced emotions, the patients were more electrodermally aroused. The pattern of findings provided by these studies is complex: the level of physiological responses to emotional events is sometimes lower and sometimes higher in patients with schizophrenia than in normal controls. It suggests that schizophrenia is associated with a dysregulation of body responses, rather than a reduced reactivity to emotional events (Taylor & Liberzon, 1999).
Brain State Changes Advances in functional brain imaging have recently made it possible to investigate the functional brain correlates of mood changes in schizophrenia. Schneider et al. (1998) used functional magnetic resonance imaging to examine brain activity in patients with schizophrenia and normal controls during happy and sad mood induction. To induce happy and sad mood, researchers presented the participants with stimuli that consisted of happy and sad facial expressions. They were required to look at each face and use it to help them to feel happy or sad. In line with previous findings, results show brain activity in the amygdala of normal controls during negative affect. Unlike controls, patients with schizophrenia did not demonstrate amygdala activation during sadness, despite similar subjective ratings of negative affect. Using morphological procedures, researchers have reported structural abnormalities of the amygdala in schizophrenia (Bogerts, 1993). The absence of amygdala activation is consistent with the hypothesis that the inappropriate behavior displayed by patients with schizophrenia in an emotional context is related to a dysregulation of dopamine systems resulting from functional impairment of the amygdala (Louilot & Besson, 2000). Interestingly, three independent studies of patients with schizophrenia have recently shown that the response of dopamine terminals to systemic administration of amphetamine, a dopaminergic agonist, is enhanced (Abi-Dargham et al., 1998; Breier et al., 1997; Laruelle et al., 1996). However, to the best of our knowledge, there is no published study in which brain imaging techniques were used to investigate in patients with schizophrenia the functional brain cor-
224
relates of emotions arising specifically in responses to discrete emotional events. In particular, in view of the dopamine hypothesis about schizophrenia and of the role putatively played by dopamine in response to emotional stimuli, investigating the influence of emotional stimuli on dopamine responses in patients with schizophrenia should be revealing (Taylor & Liberzon, 1999). Moreover, because most patients are treated with neuroleptic drugs, that is, dopamine blockers, evaluating the influence of treatment on the different components of emotion should be worthwhile.
Implications for Research Both clinical and experimental findings show that schizophrenia is associated with an impairment of different components of emotions, including evaluative processes and body and brain correlates. They also show that patients with schizophrenia often exhibit a disjunctive relationship among expressive, subjective, and biological dimensions of emotions, indicating that the different components of emotion may not be defective altogether. In particular, the impairment of the expressive and biological dimensions does not necessarily imply a disturbance of the emotions subjectively experienced by the patient, which seem to be preserved in some situations. Moreover, patients are still sensitive to the emotional salience of events, provided that the cognitive evaluation of emotional valence is relatively automatic. Whether the subjective experience of emotion is intact depends on the type of information processing each emotional event requires. How emotion and memory interact in patients with schizophrenia should depend primarily on whether the sensitivity to the emotional salience of stimuli is intact. We can therefore conclude from these studies that emotional words, which elicit a cognitive evaluation of the emotional valence that can take place automatically, may represent appropriate stimuli for investigating the relationship between emotion and memory. Moreover, full-blown investigations of this relationship should include not only memory assessments but also measurements of body and brain changes induced by emotional memories.
Interaction of Emotion and
Memory in Schizophrenia
In comparison to the numerous studies that have addressed the issue of emotional disturbances, experimental evidence pertaining to the interaction of emotion and memory in schizophrenia is rare. With the exception of one study (Koh, Grinker, Marusarz, & Forman, 1981) that used photographs as to-be-memorized stimuli, all others used emotional words. None of them investigated the body and brain correlates of emotional memories. A series of studies (Kayton & Koh, 1975;
, ,
225
Koh et al., 1981; Koh, Kayton, & Peterson, 1976) compared the impact of the emotional valence of to-be-learned words on recall performance in patients with schizophrenia and normal controls. As a collection, these studies consistently showed that normal controls recalled significantly more pleasant words than unpleasant words, therefore exhibiting a Pollyanna tendency. In contrast, the studies did not provide consistent results in schizophrenia. Kayton and Koh used a free-recall task involving pleasant, unpleasant, and neutral words in 17 patients with schizophrenia and 12 normal controls. The results showed an absence of the Pollyanna tendency in schizophrenia: patients recalled pleasant words to the same extent as unpleasant words. Moreover, in comparison with the performance of normal controls, their recall performance for pleasant words, but not for unpleasant and neutral words, was impaired. In a second study, Koh, Kayton, and Peterson used a set of recall tasks to investigate the recall performance of patients with schizophrenia when they were asked to rate the pleasantness of words during the encoding phase. This orienting task was used to make sure that all participants encoded the affective valence of words to be recalled. The set of recall tasks included an incidental free-recall task in which participants assessed the pleasantness of the words but were not aware that they would have to recall them; an intentional free-recall task, in which participants were instructed to learn the words that they rated; and a cued recall task in which recall was cued by different emotional categories (e.g., pleasant or unpleasant). For all these recall tasks, the results showed that patients, like normal controls, tended to recall pleasant words more often than unpleasant words. However, this Pollyanna tendency was not statistically significant in incidental free recall by the group of patients with schizophrenia. Interestingly, across all these tasks, the recall performance of patients with schizophrenia was not defective: they recalled as many words as did the controls. This pattern of results led the authors to conclude that both the recall deficit and the absence of a Pollyanna tendency observed in their previous study of patients with schizophrenia was remediable when patients encoded the words according to their emotional features. Unfortunately, numerous methodological flaws marred the study by Koh et al. (1976). For instance, the words used in the incidental free-recall task were the same words participants had to recall in the subsequent cued-recall task. Similarly, the words that they had to recall in the intentional free-recall task had been used previously as distractors in a recognition task. These methodological concerns limit the conclusions that can be drawn from these findings. Moreover, another study (Koh et al., 1981) using the same orienting task produced inconsistent results: Although overall recall performance was comparable in patients with schizophrenia and normal controls, the Pollyanna tendency was absent in patients. In a second experiment, Koh et al. investigated the interaction between memory and emotion in schizophrenia using items characterized by a separation of target (photographs of faces) and emotional information (favorable or unfavorable personality-trait words). They confirmed the absence of a Pollyanna
226
tendency in schizophrenia: controls recognized significantly more favorably than unfavorably encoded faces, but this asymmetry was absent in patients. Finally, in a more recent study, Calev and Edelist (1993) investigated the effect of the emotional valence of words on the rate of forgetting. They compared immediate and delayed recall of neutral, positive, and negative words in patients with schizophrenia and normal controls. The results showed that patients, like normal controls, forgot neutral words more rapidly than emotional words but forgot negative words less rapidly than positive words, whereas normal controls exhibited equivalent forgetting rates for positive and negative words. Moreover, patients with schizophrenia recalled fewer words than normal controls in both immediate and delayed recall tasks. These findings do not provide a clear picture of the impact of emotion on memory in schizophrenia. Indeed, some studies (Kayton & Koh, 1975; Koh et al., 1981) that revealed a defective Pollyanna tendency in schizophrenia suggest that the memory performance of patients with schizophrenia might be less influenced by the emotional valence of stimuli than that of normal controls. In contrast, Koh et al. (1976) reported an intact Pollyanna tendency that suggests that the memory performance of patients remains influenced by the emotional valence of words.2 Similarly, the more rapid forgetting of positive than negative words in schizophrenia reported by Calev and Edelist (1993) suggests that the emotional valence of the words to-be-learned still has an influence on patients’ memory.
An Investigation of the Influence
of Emotional Words on Recall
Performance in Schizophrenia
In view of these inconsistent results, we investigated the emotionality effect in schizophrenia using words as emotional stimuli. Because patients with schizophrenia are sensitive to the emotional salience of words (Kay, 1982; Koh et al., 1977), we predicted that they should recall more emotional words than neutral words, even though they might recall fewer words overall in comparison to controls. However, to investigate whether a putative impairment of the emotionality effect could be remediable, we asked each subject to complete two sessions. In the first session, participants were asked to remember the words, no mention being made of their emotional component. The second session was identical to the first one, with the exception that it included an additional orienting task in which we explicitly asked participants to assess the emotional dimension of words (Koh et al., 1981; Koh et al., 1976). Method Twenty-four clinically stabilized patients (10 men, 14 women) fulfilling DSM-III-R criteria for schizophrenia took part in the study. Twenty-two patients were receiving a long-term neuroleptic treatment (M = 561 mg of chlorpromazine or equivalent, SD = 410 mg), whereas 2 patients were drug free.
, ,
227
Global psychiatric symptomatology (36.4, SD = 24.5) was assessed using the Brief Psychiatric Rating Scale (BPRS). Positive (24.5, SD = 27.6) and negative (38.4, SD = 31.1) symptoms were assessed using the Scale for the Assessment of Positive Symptoms (SAPS) and the Scale for the Assessment of Negative Symptoms (SANS), respectively (Andreasen, 1982). The normal control group comprised 24 participants (10 men, 14 women). The groups did not differ significantly in age (31.3 years, SD = 24.5, and 31.8 years, SD = 8.0, respectively) or education (11.0 years, SD = 4.8, and 11.4 years, SD = 3.0). Words were selected on the basis of their emotional valence (details of word selection procedure can be found in Danion, Kauffmann-Muller, Grangé, Zimmermann, & Greth, 1995; Grangé, Greth, & Danion, 1991). There were two lists of 30 words, each with 10 positive, 10 negative, and 10 neutral words, randomly selected from each of the three valence categories. Positive, negative, and neutral words were mixed in each list. The participants completed two sessions, separated by at least 24 hours. In the study phase of each session, the participants were exposed to the words of one list. The lists were counterbalanced across sessions within each group. The two sessions differed in study instructions. In the first session, we asked the participants to read the words aloud and remember them, no mention being made of the emotional component of words. We attempted to equalize recall performance for neutral words in both groups by giving more opportunity to patients than controls to learn the words and by shortening the delay between the study and test phases of the recall task. Normal controls were given 2 seconds per word and patients 5 seconds. In the second session, they were asked to read the words aloud, to remember them, and to perform an orienting task that required them to rate the words on a 100 mm visual analogue scale according to their subjective feelings of pleasantness and unpleasantness. Normal controls were given 5 seconds and patients 10 seconds per word. The test phases were conducted 10 minutes after the study phases in the control group and 3 minutes after in the patients group. In each of the two test phases, the participants were asked to write in 3 minutes as many words from the study list as possible. Finally, at the end of the second test phase, they were asked to rate the emotional valence of the words of the list presented in the first study phase. Results and Discussion For each subject, we computed the mean ratings of positive, negative, and neutral words. Patients and normal controls discriminated among positive, negative, and neutral words to the same extent. In normal controls, the mean ratings for positive words (70.6, SD = 10.8) were higher than those for neutral words (52.8, SD = 7.2), which were higher than those for negative words (28.3, SD = 14.3). A similar profile was observed in patients (67.5, SD = 14.4; 56.9, SD = 10.1; 32.2, SD = 17.7, respectively). An analysis of variance (ANOVA) carried out on these ratings yielded a significant word type effect,
228
Mean Percent Recalled
F(2, 88) = 91.2, p < .0001, but neither the group effect nor the interaction between group and word type was significant. The mean proportions of words recalled were subjected to an ANOVA, with word type (positive vs. negative vs. neutral) and encoding condition as withinsubject factors and group as a between-subject factor. Figure 7.1 displays the results pooled across the two sessions; recall performance was not affected by the orienting task, as indicated by a lack of a significant effect of encoding conditions or interactions (all Fs < 1). Patients recalled fewer words than controls, F(1, 46) = 9.80, p < .005. This result indicated that the attempt to equalize recall performance in the two groups failed, probably as a consequence of the severity of the memory impairment that characterized the patients. There was a significant word-type effect, F(2, 92) = 5.98, p < .005; the mean proportion of recalled words was significantly higher for positive words compared to neutral words, t(47) = 3.35, p < .002, and for negative words compared to neutral words, t(47) = 2.81, p = .007. The difference between positive and negative words was not significant, nor was the interaction between group and word type. Therefore, as we predicted, while patients’ recall performance was poor, both groups recalled emotional words better than neutral words. This pattern of performance was observed under conditions in which patients and normal controls discriminated between positive, negative, and neutral words to the same degree. It was independent of whether emotional encoding was induced by an orienting task forcing the participants to process the emotional valence of words. Although the hypothesis that participants deliberately noticed valence in the first task, regardless of instructions, cannot be excluded, this finding is consistent with the assumption that the encoding of emotional valence
Controls Patients
25 20 15 10 5 0
Positive
Neutral
Negative
Materials Figure 7.1. Free-recall performance as a function of affective valence of words in patients with schizophrenia and normal participants.
, ,
229
of words does not require any controlled, strategic processes but involves more automatic processes (e.g., Hasher & Zacks, 1979). It is also consistent with evidence that automatic memory processes are intact in schizophrenia (GrasVincendon et al., 1994; Kazès et al., 1999). To investigate whether affective flattening (limited outward expression of positive and negative emotions in face, voice, and gestures) blunted the emotionality effect on memory performance, we carried out a complementary ANOVA of memory performance on two subgroups of 12 patients. We categorized these patients according to the score (below or above the median of the group) of the “affective flattening” item of the SADS, a clinical scale measuring the intensity of negative symptoms of schizophrenia. There was neither a significant group effect nor an interaction between group and word type. Therefore, our experiment suggests that the enhancement of memory performance for emotional words is not attenuated by affective flattening; it provides evidence of a disjunction between the expressed dimension of emotion, which is reduced, and emotions experienced at encoding and re-experienced at retrieval, which are preserved. Discrepancies between our results and those from previous studies suggesting that the emotionality effect on memory is disturbed in schizophrenia may be linked to their methodological differences: differences in the emotional stimuli used and differences in samples of patients. Some former studies did not use explicit diagnostic criteria of schizophrenia.
Interaction of Emotion and
Conscious Awareness Associated
With Recognition Memory in
Normal Controls
Most of the previous investigations of the interaction of emotion and memory carried out in normal controls and in patients with schizophrenia primarily concerned the objective accuracy of memory for emotional stimuli. However, as we said, what often matters most is not the objective fact that an event occurred but what it feels like and how it is re-experienced. This is particularly important for emotional memories experienced by patients suffering from psychiatric diseases, such as schizophrenia, that are accompanied by major disturbances of emotion. However, most of the studies carried out in normal controls with the R/K procedure used neutral material (review in Gardiner & Java, 1993). On the other hand, studies that have addressed the issue of the subjective experience associated with emotional memories in normal controls shared a common approach: The participants were simply asked to rate the vividness of their memories at retrieval. These studies found that memories for emotional events are more vivid than memories for neutral events (Ochsner, 2000). Recently, Ochsner (2000) conducted three experiments using the R/K procedure with student participants to investigate the subjective states of aware-
230
ness accompanying recognition of emotional events. He reasoned that emotional events possess unique attributes that make them more distinctive than neutral events. They elicit evaluative and physiological responses not generated by neutral events, and they induce semantic processing and emotional subjective states not evoked by neutral events. Because the distinctiveness with which stimuli are encoded enhances the frequency of conscious recollection (Dewhurst & Conway, 1994; Huron et al., in press; Rajaram, 1993), Ochsner postulated that emotional events should be consciously recollected more often than neutral events. Because familiarity is thought to rely on how easily a stimulus is processed, Ochsner also predicted that Know responses (which reflect familiarity) would be more frequent for emotional, and especially negative, stimuli than for neutral stimuli. Several lines of evidence suggest that without prior study, emotional stimuli are processed more fluently than neutral stimuli. Speeded perceptual or conceptual access for emotionally arousing stimuli has been shown in various tasks (review in Ochsner, 2000). Increased fluency seems to be more reliably found for negative stimuli than for positive stimuli. However, by considering that nonstudied affective stimuli seen at test should be processed more fluently as well, Ochsner reasoned that increases in familiarity for negative stimuli should be expected only if encoding significantly boosts fluency above an already elevated baseline fluency. Photographs of scenes and objects selected from the International Affective Picture System (IAPS; Lang, Greenwald, Bradley, & Hamm, 1993) were used as emotional stimuli. In experiment 1, the participants were presented with these photos and asked to rate each one according to its valence, arousal, and visual complexity. Experiment 2 was a replication of experiment 1, but the participants were not asked explicitly to attend and to rate the stimuli along these dimensions. In experiment 3, the participants were asked to encode stimuli by judging their subjective brightness. Ochsner conducted two series of analyses. Considering that the subjective states of awareness as assessed by Remember and Know responses are mutually exclusive, he first computed the proportions of Remember and Know responses for old and new items as a function of word type. Computed values are measures of discrete states of awareness. However, whether these values reflect the underlying processes is a matter of debate. Initially, it was suggested that the states of awareness captured by the R/K procedure map onto the underlying processes of conscious recollection and familiarity (Gardiner & Java, 1993). According to this view, an item either could be recollected, or familiar, but not both. However, several authors have argued against such a direct mapping (e.g., Jacoby, 1991; Yonelinas et al., 1998). They have suggested that the underlying processes of conscious recollection and familiarity are independent, rather than mutually exclusive. According to their view, some items can be exclusively recollected, some can be exclusively familiar, and some can be both recollected and familiar. In other words, whether a given stimulus is consciously recollected does not constrain whether it is familiar. Therefore, in a second series of analyses, Ochsner
, ,
231
used the independence model recently proposed by Yonelinas et al. to estimate the respective contributions of conscious recollection and familiarity to recognition memory. Conscious recollection is an estimate of how often words were correctly recollected relative to the number of opportunities participants had to do so. Familiarity is derived from the participants’ tendencies to give a correct Know response to old words and an incorrect Know response to new words. The results of the three experiments confirmed most of Ochsner’s predictions. Remember responses were more frequent for emotional than for neutral stimuli and for negative than for positive stimuli; estimates of conscious recollection followed the same pattern. However, contrary to Ochsner’s prediction, differences in familiarity were slight and nonsignificant. These findings are consistent with the notion that emotional events are better recalled or recognized than neutral events, but they go further by demonstrating that greater accuracy is due primarily to better conscious recollection of emotional stimuli. The results also support the hypothesis that emotion leads to increases in the distinctiveness with which stimuli are encoded and re-experienced in memory later on. Moreover, the results indicated that the emotionality effect obtained even when participants did not explicitly evaluate the emotional valence of stimuli (Experiment 2) or evaluated the subjective brightness of stimuli (Experiment 3). They thereby reconfirmed the hypothesis that the emotionality effect does not depend on being instructed explicitly to attend to emotional aspects of words. Finally, the pattern of differences observed in this study with positive and negative photographs is opposite to the Pollyanna tendency usually observed with positive and negative words. In keeping with Conway and Dewhurst (1995), Ochsner suggested that a more general factor underlying the emotionality effect on conscious recollection might be the personal significance of a stimulus and not its absolute emotional valence. This interpretation is also supported by an investigation of autobiographical memories in normal controls that Conway, Collins, Gathercole, and Anderson (1996) conducted. This experiment showed that the personal significance attributed to autobiographical events interacts with emotion to form a major determinant of conscious recollection for autobiographical memories.
An Investigation of the Influence of
Emotional Words on Recognition
Memory and Associated Conscious
Awareness in Schizophrenia
Ochsner’s results suggest that the R/K procedure is a more sensitive tool than traditional recall and recognition tasks for investigating the relationship between emotion and memory. The fact that emotional stimuli differentially influenced conscious recollection and familiarity may explain why some previous studies failed to find a clear relationship between emotion and memory in normal con-
232
trols. This may also account for some of the inconsistent results obtained by investigations of the interaction of memory and emotion in schizophrenia. However, the studies of conscious awareness in patients with schizophrenia were conducted using neutral stimuli, leaving unexplored the states of awareness that accompany memory for emotional events. Danion, Kazès, Huron, and Karchouni (in press) addressed this issue recently. They used the R/K procedure and the recognition model of Yonelinas et al. (1998) to investigate the influence of emotional words both on the subjective states of awareness accompanying recognition memory and the processes of conscious recollection and familiarity that might underlie Remember and Know responses. During the study phase of a recognition task, 24 patients with schizophrenia and 24 normal controls were presented with positive, negative, and neutral words. They were asked to remember the words and to perform an orienting task that required them to rate the words according to their subjective feelings of pleasantness and unpleasantness. In the test phase, the participants were presented with a list of studied and nonstudied words, and they were asked to recognize words from the study list. In addition, they were told to make a R/K judgment. Because patients with schizophrenia remain sensitive to the emotional salience of words, the researchers predicted that remember responses and conscious recollection should be more frequent for emotional than for neutral events. Results confirmed this prediction. During the study phase of the task, patients evaluated the emotional valence of words just as normal controls did. At test, patients produced fewer Remember responses and exhibited lower levels of conscious recollection than normal controls did. They gave more Know responses than controls but familiarity, as estimated using the model of Yonelinas et al., was slightly reduced. Their Remember responses were more frequent for emotional than for neutral words and for positive than for negative words, with parallel variations in conscious recollection. This pattern of responses was similar to the one observed in normal controls. The levels of Know responses and familiarity for emotional and neutral words were similar. Therefore, provided that the cognitive evaluation of emotional valence is intact, patients, like controls, consciously recollected emotional words better than neutral words.
Concluding Comments This review of the investigations of conscious awareness in schizophrenia shows that, whereas several determinants of conscious recollection such as word frequency (Huron et al., 1995), picture superiority (Huron & Danion, 2002), and directed-forgetting effects (Sonntag et al., 2003) are impaired in patients with schizophrenia, the emotionality effect is preserved when words are used. However, this review leaves numerous questions unanswered. For instance, are the body correlates and the brain correlates of this emotionality effect similar in
, ,
233
patients with schizophrenia and in normal controls? Studies of the emotionality effect of words in schizophrenia using psychophysiological and brain imaging techniques would be worth carrying out to answer this question. Are patients with schizophrenia still able to re-experience at retrieval the specific attributes that make an event emotional at encoding, such as particular emotional details or aspects of the event, components of the cognitive evaluation of the emotional valence, and physiological changes? One way to investigate this issue might be to use the R/K procedure to evaluate the subjective states of awareness that characterize the emotional component of the event. Another, complementary way to investigate this issue might be to compare the qualitative characteristics of the subjective experiences associated with emotional memories in patients with schizophrenia and in normal controls. The Memory Characteristics Questionnaire (MCQ), designed by Johnson, Foley, Suengas, and Raye (1988) to assess attributes of memories, could be used to achieve this goal. The MCQ asks the participants to rate the specific qualitative characteristics of their subjective experiences, such as visual details, mental images, and emotional reactions. These approaches to experiential aspects of emotions should provide information about not only the influence of emotional events on the frequency of conscious recollection in patients with schizophrenia but also the ability of patients to relive what makes this event distinctive, its emotional dimension. It is plausible that the relationship between emotion, memory, and conscious awareness depends on the type of information processing required by the emotional items at encoding and at retrieval. For instance, Koh et al. (1981) failed to observe a Pollyanna tendency, as noted earlier; the experiment used items characterized by a separation of target (photographs of faces) and emotional information (favorable or unfavorable personality-trait words). This failure may be related to a defective binding of target and emotional information. If so, whether this defective binding is specific to emotional information is not clear. Instead, it could simply be a special case of a generalized defect in binding separate dimensions of items. Such a deficit has been described in schizophrenia for items that comprise several spatially, temporally, or cognitively separated components (Danion et al., 1999; Rizzo, Danion, Van der Linden, & Grangé, 1996; Rizzo, Danion, Van Der Linden, Grangé, & Rohmer, 1996). These results raise the issue of the influence of emotion on conscious recollection in schizophrenia when emotional stimuli other than words are used—such as stimuli that induce stronger emotional responses; stimuli that are associated with primary, innate emotions (instead of secondary, acquired emotions; e.g., Damasio, 1995); or stimuli that require a more controlled, strategic evaluation of emotional valence. Finally, patients with schizophrenia often suffer from depressive symptoms, and Brebion, Gorman, Malaspina, Sharif, and Amador (2001) have suggested that the deficit of verbal memory observed in schizophrenia may be related to these symptoms. It would therefore be worthwhile to investigate whether schizophrenic patients suffering from
234
depressive symptoms exhibit a mood congruency effect, that is, better memory performance for negative than for positive stimuli [see chapter 6—Eds.]. All the investigations of emotion and memory in schizophrenia reviewed in this chapter shared a primary concern with episodic memory, as assessed in the laboratory using experimental procedures far removed from real-life situations. A crucial issue, albeit virtually unexplored in schizophrenia, is the relationship between emotion and autobiographical memories, that is, memories for personal events and facts from one’s life. There is converging evidence that emotional experience at the time of an event can influence memory and associated conscious recollection for that event (Conway et al., 1996). Because autobiographical memories are emotional in essence and act as organizers of autobiographical knowledge, a crucial function of the interaction of emotion and autobiographical memory is the development of the self and of personal identity (Conway & Pleydell-Pearce, 2000). Investigating the interaction of emotion and autobiographical memories may thus provide insights into the current abnormalities of personal identity, as well as the formation of the abnormal personal identity that characterizes schizophrenia. Abnormalities in personal identity emerge at the onset of schizophrenia, which occurs in adolescence and early adulthood, that is, when autobiographical knowledge is being acquired and organized to form personal identity. What are the clinical implications of the investigations of emotion and memory in schizophrenia? Evidence that the emotionality effect on memory and subjective experience may be preserved in patients with schizophrenia has positive clinical implications. Emotions are closely related to adaptive behavior at an automated level. But because emotions enhance conscious recollection, they also offer the flexibility of response based on memory for emotional events from the subject’s personal past. Evidence of a preserved emotionality effect therefore suggests that patients may still benefit from such flexibility, at least in some situations. Other clinical implications may be negative. As emotion is the consequence of how people construe situations, some neutral events or experiences may become emotional by virtue of false beliefs or delusions. Because these events will be more easily recollectable later and, hence, more richly and vividly experienced in memory, they could contribute to the persistence and the enrichment of delusional experiences. According to this view, symptoms of schizophrenia such as delusions may arise as a consequence of a combination of impaired and spared aspects of emotions and cognition.
Notes 1. In a typical Sternberg task, subjects are presented with a short list of items (memory set) to remember. After a short delay (retention interval), they are presented with a probe item and instructed to decide as quickly as possible whether the probe belongs to the previously presented memory set. Performance is measured by the time to respond and the number of errors.
, ,
235
2. This conclusion implies that the nonaffective characteristics (e.g., frequency, imagery) of words to-be-learned can be considered equivalent.
References Abbruzzese, M., & Scarone, S. (1993). Memory and attention dysfunctions in story recall in schizophrenia: Evidence of a possible frontal malfunctioning. Biological Psychology, 35, 51–58. Abi-Dargham, A., Gil, R., Krystal, J., Baldwin, R. M., Seibyl, J. P., Bowers, M., et al. (1998). Increased striatal dopamine transmission in schizophrenia: Confirmation in a second cohort. American Journal of Psychiatry, 155, 761–767. Addington, J., & Addington, D. (1998). Facial affect recognition and information processing in schizophrenia and bipolar disorder. Schizophrenia Research, 32, 171–181. Aleman, A., Hijman, R., de Haan, E. H., & Kahn, R. S. (1999). Memory impairment in schizophrenia: A meta-analysis. American Journal of Psychiatry, 156, 1358–1366. Andreasen, N. C. (1982). Negative symptoms in schizophrenia. Definition and reliability. Archives of General Psychiatry, 39, 784–788. Andreasen, N. C. (1999). A unitary model of schizophrenia. Archives of General Psychiatry, 56, 781–787. Bernstein, A. S. (1987). Orienting response research in schizophrenia: Where we have come and where we might go. Schizophrenia Bulletin, 13, 623–641. Bilder, R. M., Goldman, R. S., Robinson, D., Reiter, G., Bell, L., Bates, J. A., et al. (2000). Neuropsychology of first-episode schizophrenia: Initial characterization and clinical correlates. American Journal of Psychiatry, 157, 549–559. Blaney, P. H. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229– 246. Bleuler, E. (1911). Dementia Praecox ou groupe des schizophrénies. Paris: E.P.E.L. Bogerts, B. (1993). Recent advances in the neuropathology of schizophrenia. Schizophrenia Bulletin, 19, 431–445. Boucher, J., & Osgood, C. E. (1969). The Pollyanna hypothesis. Journal of Verbal Learning and Verbal Behavior, 8, 1–8. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129–148. Brebion, G., Gorman, J. M., Malaspina, D., Sharif, Z., & Amador, X. (2001). Clinical and cognitive factors associated with verbal memory task performance in patients with schizophrenia. American Journal of Psychiatry, 158, 758–764. Breier, A., Su, T. P., Saunders, R., Carson, R. E., Kolachana, B. S., de Bartolomeis, A., et al. (1997). Schizophrenia is associated with elevated amphetamine-induced synaptic dopamine concentrations: Evidence from a novel positron emission tomography method. Proceedings of the National Academy of Sciences of the USA, 94, 2569–2574. Bryson, G., Bell, M., & Lysaker, P. (1997). Affect recognition in schizophrenia: A function of global impairment or a specific cognitive deficit. Psychiatry Research, 71, 105–113. Calev, A. (1984a). Recall and recognition in chronic nondemented schizophrenics: Use of matched tasks. Journal of Abnormal Psychology, 93, 172–177.
236
Calev, A. (1984b). Recall and recognition in mildly disturbed schizophrenics: The use of matched tasks. Psychological Medicine, 14, 425–429. Calev, A., & Edelist, S. (1993). Affect and memory in schizophrenia: Negative emotion words are forgotten less rapidly than other words by long-hospitalized schizophrenics. Psychopathology, 26(5–6), 229–235. Clare, L., McKenna, P. J., Mortimer, A. M., & Baddeley, A. D. (1993). Memory in schizophrenia: What is impaired and what is preserved? Neuropsychologia, 31(11), 1225–1241. Conway, M. A., Collins, A. F., Gathercole, S. E., & Anderson, S. J. (1996). Recollections of true and false autobiographical memories. Journal of Experimental Psychology: General, 125, 69–95. Conway, M. A., & Dewhurst, S. A. (1995). Remembering, familiarity, and source monitoring. Quarterly Journal of Experimental Psychology, 48, 125–140. Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobiographical memories in the self-memory system. Psychological Review, 107, 261–288. Damasio, A. R. (1995). Toward a neurobiology of emotion and feeling: Operational concepts and hypotheses. Neuroscientist, 1, 19–25. Danion, J., Gokalsing, E., Robert, P., Massin-Krauss, M., & Bacon, E. (2001). Defective relationship between subjective experience and behavior in schizophrenia. American Journal of Psychiatry, 158(12), 2064–2066. Danion, J. M., Huron, C., & Robert, P. (2001). Schizophrenia and autonoetic awareness. European Neuropsychopharmacology, S133–S134. Danion, J. M., Kauffmann-Muller, F., Grangé, D., Zimmermann, M. A., & Greth, P. (1995). Affective valence of words, explicit and implicit memory in clinical depression. Journal of Affective Disorder, 34, 227–234. Danion, J. M., Kazès, M., Huron, C., & Karchouni, N. (in press). Do patients with schizophrenia consciously recollect emotional events better than neutral events? American Journal of Psychiatry. Danion, J. M., Meulemans, T., Kauffmann-Muller, F., & Vermaat, H. (2001). Intact implicit learning in schizophrenia. American Journal of Psychiatry, 158, 944–948. Danion, J. M., Rizzo, L., & Bruant, A. (1999). Functional mechanisms underlying impaired recognition memory and conscious awareness in patients with schizophrenia. Archives of General Psychiatry, 56, 639–644. Dawson, M. E., Nuechterlein, K. H., Schell, A. M., Gitlin, M., & Ventura, J. (1994). Autonomic abnormalities in schizophrenia. State or trait indicators? Archives of General Psychiatry, 51(10), 813–824. Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal of Experimental Psychology, 58, 17–22. Dewhurst, S. A., & Conway, M. A. (1994). Pictures, images, and recollective experience. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20, 1088–1098. Dominey, P. F., & Georgieff, N. (1997). Schizophrenics learn surface but not abstract structure in a serial reaction time task. Neuroreport, 8(13), 2877–2882. Doniger, G. M., Silipo, G., Rabinowicz, E. F., Snodgrass, J. G., & Javitt, D. C. (2001). Impaired sensory processing as a basis for object-recognition deficits in schizophrenia. American Journal of Psychiatry, 158, 1818–1826.
, ,
237
Dutta, S., & Kanungo, R. N. (1975). Affect and memory: A reformulation. New York: Pergamon Press. Duzel, E., Yonelinas, A. P., Mangun, G. R., Heinze, H. J., & Tulving, E. (1997). Eventrelated brain potential correlates of two states of conscious awareness in memory. Proceedings of the National Academy of Sciences of the USA, 94(11), 5973–5978. Ekman, P. (1993). Facial expression and emotion. American Psychologist, 48, 384– 392. Feinstein, A., Goldberg, T. E., Nowlin, B., & Weinberger, D. R. (1998). Types and characteristics of remote memory impairment in schizophrenia. Schizophrenia Research, 30, 155–163. Frith, C. D. (1992). The cognitive neuropsychology of schizophrenia. East Sussex, England: Erlbaum. Gardiner, J. M. (1988). Functional aspects of recollective experience. Memory & Cognition, 16, 309–313. Gardiner, J. M., & Java, R. I. (1993). Recognising and remembering. In A. Collins, M. A. Conway, S. E. Gathercole, & P. Morris (Eds.), Theories of memory (pp. 163– 188). Hillsdale, NJ: Erlbaum. Gardiner, J. M., Java, R. I., & Richardson-Klavehn, A. (1996). How level of processing really influences awareness in recognition memory. Canadian Journal of Experimental Psychology, 50, 114–122. Gardiner, J. M., & Richardson-Klavehn, A. (2000). Remembering and Knowing. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 229–244). New York: Oxford University Press. Gold, J. M., Randolph, C., Carpenter, C. J., Goldberg, T. E., & Weinberger, D. R. (1992). Forms of memory failure in schizophrenia. Journal of Abnormal Psychology, 101, 487–494. Goldberg, T. E., Weinberger, D. R., Pliskin, N. H., Berman, K. F., & Podd, M. H. (1989). Recall memory deficit in schizophrenia. A possible manifestation of prefrontal dysfunction. Schizophrenia Research, 2, 251–257. Grangé, D., Greth, P., & Danion, J. M. (1991). Application des méthodes de classification à la construction d’un test de mémoire. La Revue de Modulad, 8, 31–43. Gras-Vincendon, A., Danion, J. M., Grangé, D., Bilik, M., Willard-Schroeder, D., Sichel, J. P., et al. (1994). Explicit memory, repetition priming and cognitive skill learning in schizophrenia. Schizophrenia Research, 13, 117–126. Green, M. F. (1996). What are the functional consequences of neurocognitive deficits in schizophrenia? American Journal of Psychiatry, 153, 321–330. Hasher, L., & Zacks, R. T. (1979). Automatic and effortful processes in memory. Journal of Experimental Psychology: General, 108, 356–388. Hawkins, K. A. (1999). Memory deficits in patients with schizophrenia: Preliminary data from the Wechsler Memory Scale-Third Edition support earlier findings. Journal of Psychiatry and Neuroscience, 24, 341–347. Heinrichs, R. W., & Zakzanis, K. K. (1998). Neurocognitive deficit in schizophrenia: A quantitative review of the evidence. Neuropsychology, 12, 426–445. Henson, R. N., Rugg, M. D., Shallice, T., Josephs, O., & Dolan, R. J. (1999). Recollection and familiarity in recognition memory: An event-related functional magnetic resonance imaging study. Journal of Neurosciences, 19(10), 3962–3972.
238
Huron, C., & Danion, J. M. (2002). Impairment of constructive memory in schizophrenia. International Clinical Psychopharmacology, 17, 127–133. Huron, C., Danion, J. M., Giacomoni, F., Grangé, D., Robert, P., & Rizzo, L. (1995). Impairment of recognition memory with, but not without, conscious recollection in schizophrenia. American Journal of Psychiatry, 152(12), 1737– 1742. Huron, C., Danion, J. M., Rizzo, L., Killofer, V., & Damiens, A. (in press). Subjective qualities of memories with the picture superiority effect in schizophrenia. Journal of Abormal Psychology. Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 110, 306–340. Johnson, M. K., Foley, M. A., Suengas, A. G., & Raye, C. L. (1988). Phenomenal characteristics of memories for perceived and imagined autobiographical events. Journal of Experimental Psychology: General, 117, 371–376. Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 3–28. Johnson, M. K., & Raye, C. L. (1981). Reality monitoring. Psychological Review, 88, 67–85. Kay, S. R. (1982). Conceptual disorder in schizophrenia as a function of encoding orientation. Journal of Nervous and Mental Diseases, 170, 154–163. Kayton, L., & Koh, S. D. (1975). Hypohedonia in schizophrenia. Journal of Nervous and Mental Diseases, 161, 412–420. Kazès, M., Berthet, L., Danion, J. M., Amado, I., Willard, D., Robert, P., et al. (1999). Impairment of consciously controlled use of memory in schizophrenia. Neuropsychology, 13, 54–61. Kerr, S. L., & Neale, J. M. (1993). Emotion perception in schizophrenia: Specific deficit or further evidence of generalized poor performance? Journal of Abnormal Psychology, 102, 312–318. Koh, S. D., Grinker, R. R., Marusarz, T. Z., & Forman, P. L. (1981). Affective memory and schizophrenia anhedonia. Schizophrenia Bulletin, 7, 292–307. Koh, S. D., Kayton, L., & Peterson, R. A. (1976). Affective encoding and consequent remembering in schizophrenic young adults. Journal of Abnormal Psychology, 85, 156–166. Koh, S. D., & Peterson, R. A. (1978). Encoding orientation and the remembering of schizophrenic young adults. Journal of Abnormal Psychology, 87, 303–313. Koh, S. D., Szoc, R., & Peterson, R. A. (1977). Short-term memory scanning in schizophrenic young adults. Journal of Abnormal Psychology, 86, 451–460. Kring, A. M., Kerr, S. L, Smith, D. A. & Neale, J. M. (1993). Flat affect in schizophrenia does not reflect diminished subjective experience of emotion. Journal of Abnormal Psychology, 102, 507–517. Kring, A. M., & Neale, J. M. (1996). Do schizophrenic patients show a disjunctive relationship among expressive, experiential, and psychophysiological components of emotion? Journal of Abnormal Psychology, 105, 249–257. Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O. (1993). Looking at pictures: Affective, visceral, and behavioral reactions. Psychophysiology, 30, 261–273.
, ,
239
Laruelle, M., Abi-Dargham, A., Van Dyck, C. H., Gil, R., D’Souza, C. D., & Erdos, J. (1996). Single photon emisssion computerized tomography imaging of amphetamine-induced dopamine release in drug-free schizophrenic subjects. Proceedings of the National Academy of Sciences of the USA, 93, 9235–9240. Lazarus, R. S. (1991). Cognition and motivation in emotion. American Psychologist, 46, 352–367. Louilot, A., & Besson, C. (2000). Specificity of amygdalostriatal interactions in the involvement of mesencephalic dopaminergic neurons in affective perception. Neuroscience, 96, 73–82. Mandler, G. (1984). Mind and body: Psychology of emotion and stress. New York: Norton. Michel, L., Danion, J. M., Grangé, D., & Sandner, G. (1998). Cognitive skill learning and schizophrenia: Implications for cognitive remediation. Neuropsychology, 12, 590–599. Myin-Germeys, I., Delespaul, P. A., & deVries, M. W. (2000). Schizophrenia patients are more emotionally active than is assumed based on their behavior. Schizophrenia Bulletin, 26, 847–854. Myin-Germeys, I., van Os, J., Schwartz, J. E., Stone, A. A., & Delespaul, P. A. (2001). Emotional reactivity to daily life stress in psychosis. Archives of General Psychiatry, 58(12), 1137–1144. Ochsner, K. N. (2000). Are affective events richly recollected or simply familiar? The experience and process of recognizing feelings past. Journal of Experimental Psychology: General, 129, 242–261. Palmer, B. W., Heaton, R. K., Paulsen, J. S., Kuck, J., Braff, D., Harris, M. J., et al. (1997). Is it possible to be schizophrenic yet neuropsychologically normal? Neuropsychology, 11, 437–446. Rado, S. (1962). Psychoanalysis of behavior. Vol. 2. New York: Grune & Stratton. Rajaram, S. (1993). Remembering and knowing: Two means of access to the personal past. Memory & Cognition, 21, 89–102. Rajaram, S., & Roediger, H. L. III. (1997). Remembering and knowing as states of consciousness during retrieval. In J. D. Cohen & J. W. Schooler (Eds.), Scientific approaches to consciousness: Carnegie Mellon symposia on cognition (pp. 213– 240). Mahwah, NJ: Erlbaum. Rizzo, L., Danion, J. M., van der Linden, M., & Grangé, D. (1996). Patients with schizophrenia remember that an event has occurred, but not when. British Journal of Psychiatry, 168, 427–431. Rizzo, L., Danion, J.-M., Van Der Linden, M., Grangé, D., & Rohmer, J. G. (1996). Impairment of memory for spatial context in schizophrenia. Neuropsychology, 10, 376–384. Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 803–814. Rubin, D. C., & Friendly, M. (1986). Predicting which words get recalled: Measures of free recall, availability, goodness, emotionality, and pronunciability for 925 nouns. Memory & Cognition, 14, 79–94. Saykin, A. J., Gur, R. C., Gur, R. E., Mozley, P. D., Mozley, L. H., Resnick, S. M., et al.
240
(1991). Neuropsychological function in schizophrenia. Selective impairment in memory and learning. Archives of General Psychiatry, 48, 618–624. Saykin, A. J., Shtasel, D. L., Gur, R. E., Kester, D. B., Mozley, L. H., Stafiniak, P., et al. (1994). Neuropsychological deficits in neuroleptic naive patients with firstepisode schizophrenia. Archives of General Psychiatry, 51, 124–131. Schacter, D. L., Norman K. A., & Koutstaal, W. (1998). The cognitive neurosciences of constructive memory. Annual Review of Psychology, 49, 289–318. Schneider, F., Weiss, U., Kessler, C., Salloum, J. B., Posse, S., Grodd, W., et al. (1998). Differential amygdala activation in schizophrenia during sadness. Schizophrenia Research, 34, 133–142. Schwartz, B. D., & Winstead, D. K. (1985). Icon formation in chronic schizophrenics. Biological Psychiatry, 20, 1015–1018. Schwartz, B. L., Rosse, R. B., & Deutsch, S. I. (1993). Limits of the processing view in accounting for dissociations among memory measures in a clinical population. Memory & Cognition, 21, 63–72. Schwartz, B. L., Rosse, R. B., Veazey, C., & Deutsch, S. I. (1996). Impaired motor skill learning in schizophrenia: Implications for corticostriatal dysfunction. Biological Psychiatry, 39, 241–248. Smith, M. E. (1993). Neurophysiological manifestations of recollective experience during recognition memory judgments. Journal of Cognitive Neuroscience, 5, 1– 13. Sonntag, P., Gokalsing, E., Olivier, C., Robert, P., Burglen, F., Kauffmann-Muller, F., et al. (2003). Impaired strategic regulation of contents of conscious awareness in schizophrenia. Consciousness and Cognition, 12, 190–200. Taylor, S. F., & Liberzon, I. (1999). Paying attention to emotion in schizophrenia. British Journal of Psychiatry, 174, 6–8. Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26, 1–12. Walker, E., Marwit, S. J., & Emory, E. (1980). A cross-sectional study of emotion recognition in schizophrenics. Journal of Abnormal Psychology, 89, 428–436. Wickens, D. D. (1970). Encoding categories of words: An empirical approach to meaning. Psychological Review, 77, 1–15. Yonelinas, A. P., Kroll, N. E. A., Dobbins, I., Lazzara, M., & Knight, R. T. (1998). Recollection and familiarity deficits in amnesia: Convergence of rememberknow, process dissociation, and receiver operating characteristic data. Neuropsychology, 12, 323–339. Zajonc, R. B. (1984). On the primacy of affect. American Journal of Psychology, 39, 117–123.
, ,
241
’
T
he events of our lives are imbued with emotion. In the everyday ups and downs of daily life, the cyclical joys and sorrows, or the heart-stopping ecstasy, grief, or fear that comes just a few times in a lifetime, emotion is woven deep into the fabric of our personal past. Moreover, we don’t simply experience emotional events; we share them with others, both as they are occurring and in retrospect. How we experience and reminisce with others about the meaningful and emotional events of our lives plays a critical role in how we come to understand and evaluate our past and ourselves. For young children who are just beginning to construct an autobiographical memory, the way in which emotion and memory are intertwined provides a foundation for understanding of self and other (Fivush & Buckner, 2000; Fivush, Berlin, et al., 2003). Most important, early autobiographical memory is co-constructed in the context of parent-child reminiscing; parents influence both the structure and the content of young children’s emerging autobiography (Fivush, Haden, & Reese, 1996; Haden, Didow, Ornstein, & Eckerman, 2001; Nelson, 1993; Nelson & Fivush, 2002). In this chapter, we discuss two aspects of children’s developing memories of emotional events. First, we examine parent-child reminiscing about everyday emotional experiences. We argue that through participating in parent-guided reminiscing, children develop an emotional self-concept that simultaneously influences the way in which the past is remembered and forms the basis for understanding self in the present. We then turn to an examination of children’s memories of highly emotional and stressful experiences and how these memories may differ from memories of highly positive experiences. Given our perspective on the role of parent guided reminiscing in children’s developing autobiographical memories, we also discuss the role of parent-child reminiscing about these kinds of stressful experiences and how parents might help or hinder young
242
242
children in understanding and coping with aversive events. Throughout, we take a functional perspective on autobiographical memory (Bluck & Alea, 2002; Fivush, 1988). Our focus is on how autobiographical memories are used to link the past with current understanding of self through the creation of an autobiographical narrative, help define self in relation to others through time, and contribute to understanding, regulating, and coping with aversive experiences. Thus, we emphasize meaning making rather than accuracy; how do individuals make sense of their past experiences and how does this process emerge developmentally within joint reminiscing? To place this research in perspective, we first briefly review the development of children’s autobiographical memory and emotional understanding more broadly.
Autobiographical Memory
Development
Although memory functions from birth, autobiographical memory—defined as consciously available and socially shareable memories of specific events experienced by the self at a specified time and place—emerges slowly across the preschool years (Nelson & Fivush, 2002; Pillemer & White, 1989). Children begin to converse about specific past experiences at about 16 to 18 months of age (Eisenberg, 1985; Harley & Reese, 1999). However, at this early point, adults provide much of the content and structure of what happened; children participate by confirming, denying, or repeating what the adult says. Very quickly, children become more competent participants in autobiographical reminiscing, and by 3 years of age, children are able to give accurate detailed accounts of their personal past (Fivush, Gray, & Fromhoff, 1987), although they still rely on adults to provide much of the coherence, or narrative structure (Hamond & Fivush, 1990; Ornstein, 1995). By the end of the preschool years, children can provide relatively clear and organized narratives of their past (Fivush, Haden, & Adam, 1995), and these skills continue to develop throughout childhood (Hudson & Shapiro, 1991; McCabe & Peterson, 1991). Extensive research confirms that children are learning the forms and functions for talking about the past through participating in adult-guided reminiscing. In accord with Vygotsky’s (1978) social-cultural theory of development, adults provide the structure, or “scaffold,” for children’s early reminiscing; over time, children begin to internalize the forms used by adults and begin to provide the structure for themselves. More specifically, children learn the canonical narrative forms and the culturally appropriate content to include when recalling the past (see Fivush, 1994, and Fivush & Haden, 2003, for a review). Importantly, there are clear and enduring individual differences in both parents’ and children’s autobiographical reminiscing (Fivush & Fromhoff, 1988; Hudson, 1990; McCabe & Peterson, 1991). Some parents display a highly elaborative
’
243
reminiscing style, talking about the past in great detail and encouraging their children’s participation. In contrast, some parents display a less elaborative reminiscing style, taking about the past in sparse detail, often simply asking the same questions over and over. Not only is parental reminiscing style consistent over time but highly elaborative parents facilitate their children’s developing autobiographical memory skills. Children of more elaborative parents eventually come to tell more richly embellished narratives of their own past experience than children of less elaborative parents (Harley & Reese, 1999; McCabe & Peterson, 1992; Reese, Haden, & Fivush, 1993). In addition to individual differences within cultures, there are also substantial differences among cultures in parent-child reminiscing. Within Asian cultures, in which self is conceptualized as interdependent and integrated into a social and moral community (Oyserman & Markus, 1993), there is less focus overall on the past; parents in these cultures are less elaborative than Caucasian parents (Leichtman, Wang, & Pillemer, 2003). Provocatively, there are also differences within middle-class American culture between boys and girls. Parents are more elaborative and more evaluative overall when reminiscing with daughters than with sons (Reese, Haden, & Fivush, 1996). By middle childhood, the effects of these differences in early parent-child reminiscing become apparent. Caucasian children tell longer and more elaborated narratives of their personal past than do Asian children (Han, Leichtman, & Wang, 1998), and middle-class Caucasian girls tell longer and more elaborated personal narratives than boys (Buckner & Fivush, 1998). And as adults, Caucasians and women provide longer and more detailed narratives of their personal past than do Asians and adult men (see Pillemer, 1998, for a review). The emerging effects of culture and gender differences indicate that parental reminiscing style is an important avenue for the socialization of self and other, as well as for the construction of a culturally “appropriate” autobiography (Fivush & Haden, 2003).
Emotional Development Children begin talking about their emotions as well as their past early in development. By about 13 months of age, children begin referencing their emotional states, and by 2 years of age, children are integrating information about their own and others’ emotions into their everyday conversations (Bretherton, Fritz, ZahnWaxler, & Ridgeway, 1986). Further, and again similar to the development of autobiographical memory, the way in which parents incorporate emotion into conversations plays a critical role in children’s developing understanding of emotion. Mothers who talk more about emotion early in development have children who talk more about emotion later in development (Denham, Zoller, & Couchoud, 1994; Dunn, Bretherton, & Munn, 1987). More broadly, families that discuss emotional experiences in more open and integrative ways have children
244
who develop better prosocial skills, have more positive peer relations, and show better psychological adjustment (see Halberstadt, Denham, & Dunsmore, 2001, for a review). And just as there are gender differences in parent-child reminiscing, there are also gender differences in emotion socialization. Mothers talk more about emotions overall with girls than with boys, and girls begin to talk more about emotion than boys do as early as the preschool years (Dunn et al., 1987; Zahn-Waxler, Cole, & Barrett, 1991). To summarize, early socialization patterns mirror adult patterns of emotional expression and understanding. Adult women express more emotion than men and report experiencing emotions more frequently and intensely than do men (see Fischer, 2000, for a review). Specific to autobiography, adult women include more emotion when narrating their past than do adult men (Bauer, Stennes, & Haight, in press; Davis, 1990).
Parent-Child Reminiscing About
Emotional Events
The developments in autobiographical memory and emotional development suggest that children may be socialized in both domains in parallel ways and along similar timelines. Not surprisingly, when reminiscing about the past, parents and children include information about the emotional aspects of their shared experiences. Reminiscing about past emotional experiences differs in important ways from talking about emotions experienced in the present (Dunn, Brown, & Beardsall, 1991; Fivush, 1993). First, parents and children are not in the immediate heat of the emotional moment when reminiscing about past emotions and therefore may be better able to reflect on and interpret emotional experience. Second, in reminiscing, parents can choose to talk about particular emotions rather than others; for example, they can emphasize times when their children were sad but downplay times when their children were angry. Finally, parents can choose to talk about various aspects of emotional experience; they can focus on the emotion itself and the way it is expressed; they can focus on the causes of emotional experience, or on how negative emotion is resolved; and they can talk about both the child’s and others’ emotional experiences and explain similarities or differences. Thus, in the context of reminiscing about emotional experiences, parents can provide a framework for children’s developing understanding of how and why they and others experience emotion. Through parent-guided interpretation and evaluation of emotional experiences, children begin to develop an “emotional self-concept” (Fivush & Buckner, 2003; Fivush, Berlin, et al., 2003), composed of three functions: self-defining (the kind of emotional person I am), self-in-relation (how I express and share my emotions with others), and self-regulation (how I cope with and resolve negative emotion).
’
245
In line with gender differences in many other aspects of emotional processing (Fischer, 2000), several studies have documented gender differences in the emotional content of mother-child reminiscing (Adams, Kuebli, Boyle, & Fivush, 1995; Fivush, 1989, 1991; Kuebli, Butler, & Fivush, 1995; Kuebli & Fivush, 1992). In these studies, mothers and their preschool children are visited in their homes and asked to talk about several past experiences they have shared. No further instructions are given. Tape-recorded conversation are transcribed and coded for various dimensions. Across studies, mothers talk more about emotion with their preschool daughters than sons, especially about sadness. Mothers also use a wider variety of emotion words with girls than with boys, for example, talking about being “sad” and “upset” and “crabby” rather than simply “sad.” As children grow older, mothers talk increasingly of other people’s emotions as well as the child’s, especially with girls. Mothers also place emotions in a more social and relational context with girls than with boys. In reminiscing with daughters, mothers talk about how emotions emerge from, and are modulated by, interactions with other people, whereas with sons, they are more likely to discuss emotions as internal and autonomous experiences. These patterns suggest that mothers are providing a more embellished and differentiated understanding of emotion with daughter than with sons. Although there are no differences in how girls and boys talk about past emotions early in the preschool years, by the end of the preschool years, girls are taking significantly more about their past emotional experience than are boys (see Fivush & Buckner, 2000, for a review). These findings, along with previous findings of differences in adults’ autobiographical narratives, raise the question of possible differences between mothers and fathers reminiscing with daughters and sons. To explore this issue, we asked mothers and fathers to independently discuss four specific past experiences with their 4–year-old children when the child experienced happiness, sadness, anger, and fear (Fivush, Brotman, Buckner, & Goodman, 2000). Overall, mothers talked more about emotion (M = 5.23 emotion words per event discussed) than fathers (M = 3.99 emotion words), but both mothers and fathers talked about sadness, especially the causes of sadness, more with daughters (M = 4.05 mentions) than with sons (1.95 mentions). In this study, girls talked more about being scared than did boys, but there were no differences in conversations about feeling happy, sad, or angry. Finally, as in previous research, both mothers and fathers placed emotions in a more social and relational context with daughters than with sons. Thus, overall, this body of research indicates that emotional experiences are discussed differently depending on both the gender of the parent and the child. Parents, especially mothers, talk about emotional experiences more, use a wider variety of emotional language, and place past emotions in a more social and relational context with their preschool daughters than with their sons. In turn, by the end of the preschool years, girls are talking more about the emotional
246
aspects of their past experiences and using a wider variety of emotion terms than boys. These patterns suggest that children are being socialized into “genderappropriate” ways of understanding their emotional experience. Parents facilitate the development of a more complex and more differentiated emotional sense of self in the past with daughters than with sons (Fivush, 1998b). Whereas previous research on the development of autobiographical memory has demonstrated distinct parental reminiscing styles that vary in elaborativeness, research on reminiscing about emotional events focuses on differences in the emotional content of parent-child reminiscing. To integrate these research findings, we examined whether level of maternal elaboration is related to the type of emotional event discussed (Fivush, Berlin, et al., 2003). We were particularly interested in how mothers reminisce about emotionally negative events, because negative emotions are often difficult to cope with, and children must learn how to express and resolve negative affect in culturally appropriate ways. Are mothers more elaborative when reminiscing about fear rather than sadness or anger? Further, does content vary depending on the type of emotion discussed? Do mothers differ in their focus on the emotion itself, the causes of emotional experience, or resolution of negative affect depending on the type of emotional event under discussion or the gender of the child? Mothers reminisced with their 4-year-old children about three specific events, a time the child was scared, angry, and sad. With daughters, mothers were more elaborative overall, talking in more embellished detail about emotional experiences than with sons. Mothers were also more evaluative with daughters than with sons, providing more feedback and more confirmation of their daughters’ contributions to the conversations than their sons’. Similarly, girls were more elaborative than boys. But there were also differences depending on the type of emotion discussed. Across all three emotions, mothers discussed the causes of the emotional experiences to a greater extent than the emotion itself or its resolution, suggesting that mothers are helping their children to understand how emotional experiences arise. However, when discussing fearful events, mothers elaborated more, and they talked more about the facts concerning the event itself than when discussing sadness or anger. Mothers also focused on resolving fear more so than anger (but not sadness). Children also seemed to be more concerned with fearful experiences than with sadness or anger. They talked more about fear overall, and they provided more evaluations and resolutions when discussing fear than when discussing sadness or anger. In contrast, in conversations about anger, mothers did not elaborate, they did not discuss the facts surrounding the event itself, and they did not resolve anger as much as they did in conversations about fear. Rather, when discussing anger, mothers focused on emotional attributions. Children also did not talk as much about anger, they did not evaluate information about anger, and they did not discuss emotional resolutions of anger to the same extent as
’
247
resolutions of fear or sadness. Finally, sadness falls between these two extremes. Conversations about sadness were not as elaborative overall as conversations about fear, nor did mothers focus on the event itself, as they did for fear, or on emotional attributions, as they did for anger. Rather, mothers focused on evaluating and resolving sad feelings. Children also focused on resolving sadness. These conversational patterns are illustrated in table 8.1, which presents excerpts from conversations between one mother-daughter dyad and one mother-son dyad, each reminiscing about three emotional events. Mothers’ focus on the causes of their children’s emotions is apparent, as are the gender differences. Across all three emotional events, the mother and daughter talk more about the emotional event, and the mother uses more nuanced emotion language and talks in more detail about the daughter’s experience and expression of emotion than in the mother-son dyad. These patterns confirm the previous suggestion that girls are socialized into a more elaborated and differentiated understanding of their past emotions. Perhaps most important, research on parent-child reminiscing underscores that even when recalling the everyday events of our lives, emotion is an integral aspect of what we remember. Beginning very early in development, parents and children are co-constructing personal narratives replete with emotional experience; through these narratives, parents are helping their children to interpret and evaluate their experience. Mothers seem to reminisce more about emotions than do fathers, and both mothers and fathers reminisce about emotions more with daughters than with sons. Through participating in richly emotional reminiscing, girls may be constructing a more emotionally laden and emotionally nuanced sense of self in the past. Further, by placing emotions in a more interpersonal context with daughters than with sons, parents may be teaching girls to integrate emotions into relationships with others to a greater extent than are boys. Thus, girls may be developing a more embellished, more differentiated, and more relational emotional self-concept than are boys. However, there are also important differences depending on the type of emotion discussed. These patterns suggest that parents are influencing the way in which their children are learning to understand, express, and perhaps even experience specific emotions. An important limitation of this research is the focus on white middle-class American families. Emotional expression is clearly culturally mediated (Lutz & White, 1986); therefore, it is not surprising that there are cultural differences in parent-child reminiscing about emotional events (see Leichtman et al., 2003, for a review). For example, Wang (2002) asked American and Chinese mothers to reminisce about times when their preschool child was happy, scared, angry, and sad. American mothers used what Wang labeled an “emotion-explaining” style, a pattern similar to what we have just described, whereas Chinese mothers used an “emotion-criticizing” style. An emotion-explaining style helps create shared emotional experiences that define one in relation to others; an emotion-
248
. Excerpts From Mother-Child Conversations About Everyday Emotional Events (“. . .” indicates some missing dialogue) Mother-daughter dyad
Mother-son dyad
Conversations about fear M: And what happens when there’s a big thunder and lightening storm? C: I’d want to be with my mom and dad [whispering]. M: Yes, you want to be with your mom and dad. And what happens if you’re sleeping, and sleeping and there’s a big bunch of thunder in the middle of the night? What happens sometimes? C: Scared M: Scared, do you tremble like that? Do you shake? Huh? Then what do you do? C: I get up and go to my mom and dad [whispering]. M: You get up and go to your mom and dad. And what do we do? C: Say don’t worry. M: Say don’t worry. Do we hold you? Yes, does holding you help you when you’re scared? Yes.
M: . . . why did you come in bed with mommy and daddy? Cause how did you feel, how did that storm make you feel? C: Umm, I feel really scared. M: A little bit scared? C: Really scared. M: You were really scared. What scares you about the thunderstorm honey? C: . . . brokes the TV. M: Yeah, that scared you because the TV went off.
Conversations about anger M: Well, I seem to remember that yesterday you and Patrick were swinging on the swings and he was doing something that really annoyed you. . . . Do you remember? C: He wouldn’t give me a swing. M: Oh! That’s right . . . and did you wait patiently for a long time, hmm? C: Yes. M: And is that when you started to get really angry? Uh-huh, how do you feel inside when you get angry? C: . . . Mad. M: Mad! And did your mouth turn down at the corners? And do you feel like going grrrr? Conversations about sadness ... M: Well, one thing that made you really sad was when your best friend Shena moved away, right? Yeah, and did we watch all her things go on the moving truck? Uh-huh, and do you remember why she had to move away? C: . . . Because Shena’s Dad had to work.
M: Now how do you feel when Mommy makes you take you vitamins in the morning and you don’t want to take them? C: I feel mad. M: You do! And what do you do? C: I hide. M: . . . Uh-huh, then what happens to Mommy? How does Mommy feel when you didn’t take your vitamins and your calcium? C: Mad. M: Then what happens to [you]? C: Get in trouble. M: Yeah.
M: How did you feel when Mandy wouldn’t let you play the other day, when she was playing with Jess and she told you boys to go away? How did that make you feel? C: Umm, sad. M: And why does Mandy do that? (continued)
’
249
. (continued) Mother-daughter dyad
Mother-son dyad
Conversations about sadness M: Shena’s Daddy was going to start working at a new job . . . And do you still miss Shena when you think about it? Yes? C: Yes. M: It makes you sad, doesn’t it? But is she still your friend even far away? Yes! What can you do even though she’s far away? C: Give her a happy letter with a [drawing] on it. M: Give her a happy letter, right, and we have a drawing, don’t we?
C: Because she’s mean. M: [laughing] Is she mean? Is she mean all the time? When she doesn’t let you play, you’re not very happy, are you? C: Umm.
criticizing style uses emotion to instill proper behavior and values. Wang argues that these different styles emerge from larger cultural constructions of self as independent or interdependent, as discussed earlier. Independent selves own and express their emotional experience, whereas interdependent selves use emotions to regulate appropriate social behavior (see also Mullin & Yi, 1995). Obviously, these are not orthogonal functions, and mothers and children from both cultures likely show elements of both styles, but Wang’s research highlights that children may be developing different understandings of self and emotion depending on the larger social and cultural context within which reminiscing is embedded. Thus far, we have discussed how parent-child reminiscing modulates children’s developing understanding of their everyday emotional experiences. But what of more stressful and traumatic events? How do extreme levels of stress affect children’s memories, and how might parent-guided reminiscing play a role in helping children to understand and cope with aversive experiences?
Memories of Stressful
and Traumatic Events
Intertwined with everyday emotional experiences are the unfortunate accidents, disasters, and other stressful experiences that occur over the course of a lifetime. The concept of stress is familiar to both laypersons and professionals. In fact, it is not uncommon to hear the words stress or trauma in everyday conversations. However, as Mason (1975) notes, “The single most remarkable historical fact concerning the term ‘stress’ is its persistent, widespread usage in biology and medicine in spite of almost chaotic disagreement over
250
its definition” (p. 9). Despite the lack of a universal definition, the topic of stress and memory has been of interest to researchers from many areas of study. Because much of the developmental research stems from forensic issues involving children as witnesses or victims of violence, the focus has been on how stress affects the amount, accuracy, and suggestibility of children’s recall of past experiences (see Ceci & Bruck, 1993, and chapter 10, this volume). Although not the focus in this chapter, the consensus on accuracy is clear (see Westcott, Davies, & Bull, 2002, for an overview). Even quite young children can respond to open-ended free-recall questions (e.g., “What was the man wearing?”) with accurate details of their past experiences. Moreover, repeatedly interviewing children about the same event does not compromise accuracy in the absence of suggestive and misleading questions (Fivush, Peterson, & Schwartzmueller, 2002). The problem is that preschoolers, though accurate, provide little information in response to these open-ended questions, and errors increase dramatically as questions become more close-ended. Accuracy is compromised in young children’s memory reports in two ways. First, responses to specific close-ended questions (e.g., “What was the man wearing on his head?”), especially yes/no questions (e.g., “Was the man wearing a hat on his head?”), are vulnerable to error during the preschool years (Fivush et al., 2002). Second, preschoolers are substantially more susceptible to misleading and suggestive questions (e.g., “What color was the man’s jacket?” when no jacket was worn) than older children and adults (Ceci & Bruck, 1993). And suggestive questions asked over repeated interviews can induce many errors into young children’s memory reports (Leichtman & Ceci, 1995). What is still in question is the extent to which these kinds of misleading questions change the actual memory representation or whether children are responding to the implicit social demands of the interviewer’s questions. Related to this, although children change their responses to specific questions, it is not clear that these errors are incorporated into children’s subsequent free recall (Cassal & Bjorkland, 1995; Fivush et al., 2002). Moreover, the extent to which stress interacts with accuracy of recall is still controversial. First and foremost, this research domain is plagued by the problem of operationalizing stress. Several methods have been devised to assess children’s stress levels, including self-report, global ratings of stress level by parents or doctors, objective stress ratings based on specific behavioral indices, and physiological measures such as heart rate, skin conductance, and hormonal levels. When directly compared, the different measures of stress do not correlate highly with one another (Eisen, Goodman, Ghetti, & Qin, 1999; Ornstein, 1995; Parker, Bahrick, Merrit, Lundy, & Fivush, 1998); thus, it is not surprising that they do not correlate systematically with memory measures. In addition, different measures of memory have been used across studies, some studies assessing free recall and other studies assessing cued recall or recognition. Also, different aspects of the
’
251
event are targeted for recall, such as central or peripheral information. Moreover, studies have examined events ranging from mildly stressful, such as getting an inoculation, to highly traumatic, such as witnessing the murder of a parent. Finally, stress may influence memory differently at different developmental points, further complicating comparisons across studies. Several clinical case studies describe children’s memories of truly horrendous experiences, such as being kidnapped and buried alive or witnessing the murder of a parent (Malmquist, 1986; Terr, 1979, 1988). Terr (1979) reports that school-age children who were kidnapped from a school bus and essentially buried alive overnight were able to recall the horrific event in vivid detail up to 4 years later. Similarly, in describing long-term memories of children traumatized during the preschool years, Terr (1988) reports that children 3 years of age or older at time of traumatization continue to report vivid and elaborated memories of events such as abuse, kidnapping, and other violent acts. However, children under the age of 3 at the time of the event do not develop the ability to recall the event verbally as they grow older, although aspects of the traumatic experiences may be present in their behavior. These findings suggest that real-life traumatic events are remembered in great detail over long periods of time, at least in children 3 or older at the time of the event. Although clinical studies allow rich description of memories of real-life trauma, they do not allow for systematic comparisons across children or type of event. To examine this question, developmental researchers have taken advantage of necessary but painful medical procedures that many children must endure. Studies use one of two approaches. One approach compares one group of children’s memories of stressful events, such as inoculations, to another group of children’s memories of more mundane events such as a routine doctor examination. Children in the stressful situation generally show enhanced memory for the event in comparison to the participants experiencing a more neutral event, suggesting that stressful events are better recalled than emotionally neutral events (see Pezdek & Taylor, 2001, for a review). The other methodological approach has been to select a single stress-producing event and examine individual differences in stress level. For instance, Merritt, Ornstein, and Spicker (1994) examined children’s recall of a stressful medical procedure, a voiding cystourethrogram. They found that children who experienced more stress during this procedure, as measured by behavioral ratings, recalled less accurate information than children who displayed less stress during the procedure. Similarly, Vandermass, Hess, and Baker-Ward (1993) found that high anxiety or stress had a debilitative effect on older children’s recall of a dental operation but not on younger children’s, suggesting developmental differences in the stress-memory relation. More specifically, preschool children may experience and recall stressful events differently than older children. In a systematic, longitudinal research program examining children’s memories of a stressful event, Peterson and others have examined children’s memories
252
of an injury resulting in emergency room treatment (Peterson, 1999; Peterson & Bell, 1996; Peterson & Whalen, 2001). Injuries were relatively minor, requiring only outpatient procedures, and included mostly lacerations requiring stitches, severe burns, and broken bones. Children recalled their injury and treatment within 2 weeks of the event and again at 6 months, 1 year, 2 years, and 5 years after the event. Although there are several nuances to the results, the basic finding is that children 3 years of age or older at the time of the experience recalled this event extremely well over the 5–year period. In fact, Peterson and Whalen report virtually no forgetting of the injury itself and little forgetting of the hospital treatment. However, there were no relations between stress, as assessed by maternal and child report on a Likert scale, and either amount or accuracy of recall. These results confirm that children recall stressful events extremely well over long time periods, although they also contribute to the mixed findings in the literature over possible relations between stress and memory. Importantly, children under the age of 3 at the time of experience showed a somewhat different pattern (Peterson & Rideout, 1998). One- and 2-year-old children who were not able to verbally recall the event when it occurred did not show accurate recall as they grew older. Although many of these children did respond to the interviewer’s questions about what happened at later interviews, about 50% of what they reported was wrong. Thus, similar to the clinical descriptions provided by Terr (1988), more systematically controlled research confirms that children under the age of 3 may have particular difficulty verbally recalling events even as they grow older and develop the required language skills. Most important, Peterson and Rideout present evidence suggesting that it is not age per se that is critical, but rather children’s ability to verbally recall the event when it occurred. Those 2-year-olds who were able to give a coherent, although obviously still limited, verbal report of what happened when first interviewed remained able to recall the event accurately as they grew older. Thus, it seems that language or narrative skill at the time of experience is a critical factor in predicting whether children will remain able to recall events accurately as they grow older (see Fivush, 1998a, and Fivush, Pipe, Murachver, & Reese, 1997, for review and further discussion of this issue). One of the limitations of this body of research is the focus on memories of medical procedures. Whereas many of these procedures are quite painful and stressful for young children, it is not clear that they approach the level of stress experienced during traumatic events. Further, children often take their emotional cues from the adults around them as to the meaning and severity of the event; in the case of medical procedures, parents may be stressed, but they are not in fear of either their own or their children’s lives (at least for the events that have been studied thus far). One of the prevalent theories about the relation between memory and stress comes from the Yerkes-Dodson law that predicts an inverted-U function (see Christianson, 1992, for an overview). Basically, it is postulated that as stress increases to moderate levels, memory will be enhanced
’
253
due to increased attention and processing, but as stress increases to extreme levels, memory will be hindered because the system will be overwhelmed. Although this theory is widely cited, little evidence shows that this function holds. One problem in assessing this function is that control for level of stress while holding the event to be recalled constant is difficult. And, of course, creating extremely high levels of stress in the laboratory is not ethical. Only one study has systematically examined relations between children’s levels of stress and memory for a naturally occurring life-threatening traumatic event. Bahrick, Parker, Fivush, and Levitt (1998) examined how stress affected preschool children’s memories for Hurricane Andrew, a devastating storm that hit the Florida coast in 1992. Stress was defined by the amount of property damage to the family’s home. Children were placed into either a low, moderate, or high damage group. Families who prepared for the hurricane, but sustained little property damage, were placed in the low damage group. If they experienced substantial damage to the perimeter of their home, such as the yard, windows, or roof, but the house itself remained intact, they were placed into the moderate damage group. Families who experienced a severe storm that penetrated the home, often resulting in partial collapse of the roof and multiple broken windows or doors, were placed into the high damage group. Three- and 4-year-old children were interviewed about their experiences within 2 to 6 months of the storm. Interviews began with an open-ended question about the hurricane, followed by nondirective prompts until the child did not recall any additional information. Then the experimenter asked a series of more specific questions about preparing for the storm, the storm itself, and the aftermath of the storm. Children’s recall was related to the amount of stress experienced, such that the children exposed to a moderate amount of damage recalled more than children who experienced high or low damage did. This pattern would seem to support the proposed inverted-U function between memory and stress. However, 6 years later, when the children were 9 and 10 years old, we interviewed them again about their hurricane experiences (Fivush, Sales, Goldberg, Bahrick, & Parker, in press). Figure 8.1 displays the mean amount of information children recalled at each interview by stress group. All children were able to recall the event in vivid detail 6 years later, and there were no differences in overall amount recalled among children in the low, moderate, or high damage groups. Thus, there is no longer an inverted-U function. In fact, children now reported more than twice as much information about the experiences as they had at the initial interview. We do not want to argue that no forgetting occurs; we argue that children are now able to verbally report much more of what they remember than they could when they were younger. In this situation, children were able to recall the event when it occurred, and, as their language skills increased, their verbal report of the event increased as well (see Peterson & Whalen, 2001, for similar findings and arguments). This confirms our earlier discussion of Peterson’s research (Peterson & Rideout,
254
Mean propositions recalled
140 120 100 80
Low Moderate High
60 40 20 0 Immediate Recall
6 year follow-up Recall
Figure 8.1. Mean number of propositions recalled by children in each stress group over time.
1998). In that research, children who were unable to recall the event at the time of occurrence were not able to subsequently recall the event with any accuracy, but like that research, we also found that children who could recall the core aspects of the event when it occurred remained able to recall the event in vivid detail over extended time, indicating that memory of stressful events is quite long lived. Importantly, amount of recall was not related to either retention interval or amount of rehearsal at either interview time. However, there were still effects of stress on memory 6 years later. Children in the high damage group needed more prompts and cues to recall as much information as children in the moderate and low damage groups, as shown in figure 8.2. This changes the interpretation of the original findings. Obviously, if children in the high stress group are now recalling as much information as children in the other two groups, arguing that stress interferes with memory per se becomes difficult. Rather, stress may interfere with either the ability or the willingness to retrieve information for recall. If we interpret the lower recall in the high damage group at the first interview as a form of avoidance, in that highly stressed children simply did not want to talk about their experiences, then 6 years later children in the high damage group may express avoidance as being less willing to respond freely to the experimenter because they found it stressful to recall their experiences openly. However, when directly questioned and prompted, these children demonstrated that they do indeed recall as much as the children who experienced lesser damage. Thus, stress may interfere with retrieval in different ways at different developmental ages or as time since the stressful experience increases.
’
255
Mean Amount Recalled
120 100 80 60
Free Recall Cued Recall
40 20 0 Low Stress
Moderate Stress
High Stress
Figure 8.2. Mean number of propositions recalled by children in each stress group in free and cued recall.
Although stress affected recall at both interviews, children in all three damage groups recalled detailed information about their hurricane experience. Though we were unable to compare amount of recall about this event directly to other events recalled by these same children, comparing across studies of children’s memory, we found that children recalled substantially more about the hurricane event than their age-peers recall about more mundane or even mildly stressful events (see Fivush, 1998a, for a review). Further, although only a subset of the children’s recall was judged for accuracy by mothers, ratings of accuracy were extremely high, above 90%. Thus, it appears that highly stressful events are well recalled overall, although as stress increases, children may be less willing to think about and talk about their experiences. To illustrate the richness and detail of children’s recall, table 8.2 presents excerpts from a child in the high damage group and a child in the moderate damage group at each interview. These are children’s responses to the first openended question, “What do you remember about Hurricane Andrew?” and the nondirective prompts that followed this question. As shown, both children recalled the event in great detail at each interview. Even 6 years later, children were able to narrate a detailed, coherent account of the storm and its consequences. These excerpts also illustrate how emotionally compelling the memories are, especially for children who experienced extensive damage to their homes. Children seem to remain highly affected by their experience even years later. As we already mentioned, most of the research on stress and memory has focused on accuracy and overall amount of information children recall. However, relations between memory and stress are important not only for forensic reasons but also
256
table 8.2 Excerpts of Children’s Recall of Hurricane Andrew Immediate interview
Follow-up interview
Child who experienced high damage E: Can you think really hard about the hurricane and tell me everything you remember about it? C: [Child nods yes] E: Yeah? What do you remember first? C: Um, I remember we had friends over. E: You had friends over, yeah, what else? C: Um, I was, um, stuffed in a corner. E: You were stuffed in a corner, yeah, what else can you remember? C: I was in my bedroom. E: In your bedroom, yeah. C: And um . . . and um, during the hurricane, there was a window in my bedroom, and um, well we, I was stuffed in the corner and then we, then we, um, we had the Farmigan’s over and they, um, stayed at our house for a little bit longer after the hurricane. E: Yeah, what else can remember about that? C: Um, that they were sitting on the furniture and I wasn’t. E: And you weren’t. O.K. Can you remember anything else about the hurricane? C: [Child shakes head no]
E: Can you tell me everything you can remember about Hurricane Andrew? C: Okay. Um, I remember the hurricane was the day after my fourth birthday party. And we were scared. My grand—, grandparents came over. And our neighbors Sue and Terry. And it was me, my mom and my dad. And we were in my room and my grandparents were sleeping in my mom and dad’s room. And I remember I had on, I sat, I wanted to go close to the window because we had a mattress over our window, because there’s only one window in my room. And I sat and then this big lightening bolt came I was scared a lot and I remember half way through the night, part, a corner of our roof blew off. And our grandparents came and told us that it was off and we put two or three trashcans under it to catch it. And Terry, he had out little weather radio and he said he was sorry because there were no cartoons on, there was only weather about the hurricane. And I remember, I had my bed is in the corner of the room and about five feet before my closet and they made me, my mom and my dad made me stay there the whole night. Cause they were afraid if the window blew off, the mattress would hit me [she goes on like this for several more sentences]. . . . And I was really scared because I had never been through anything like that, and um, I don’t remember a lot.
Child who experienced moderate damage E: So, can you think really hard and tell me what you remember about Hurricane Andrew? C: Ah. . . . You know, in my neighbor’s house. E: Ah. C: All the screens ‘cause she has a column of screens . . . when . . . when the screen was knocked over, they had to get new screen, and there was all trees. My next door neighbor had to go all [makes noise] like that.
E: Can you think really hard about the hurricane and tell me everything that you can remember about it. We want to hear anything and everything that you can think of. C: Well, I remember we had a coconut palm in our backyard, and my dad had to cut all the coconuts off to make sure none of the windows broke or anything. E: Uh-huh. C: And, well, and that coconut palm broke in the Hurricane Andrew so now that’s (continued)
’
257
table 8.2 (continued) Immediate interview
Follow-up interview
Child who experienced moderate damage E: Anything else you remember about that? C: Only this . . . only my little screen fell off and nothing else. And all the trees fell off, a lot a lot, but not all. E: Right. C: Any you know what, and even when it was almost the hurricane we were cutting coconuts, and then, then we got back inside, and then, and my dad, and then he is all sweaty and he has ants all over. That’s it. E: Ok. Anything else that you remember about the hurricane C: Um, I think nothing else.
not there anymore. But we didn’t, we were pretty lucky and we didn’t have much happen. All it was was, like, fences and stuff. E: Uh-huh. C: And our pool was very dirty. E: Really? C: Yeah. We, uh, stayed in our, what used to be our play room, which is now our computer room, and we had little shutters that we put over the glass so no glass would break there so we were very safe there. And we had a, we had a patio by the pool. E: Uh-huh. C: And we have some sort of electrical shutters to close everything up in the patio, so we can put lots of our stuff in there. And I don’t remember much of the hurricane because I don’t know how, but I fell asleep in the middle of it. E: Really? C: Uh-huh. So that’s about it.
Note. C stands for Child, E stands for Experimenter
because how we remember and incorporate the stressful occurrences of our lives has implications for future coping and physical and emotional well-being (Pennebaker, 1997). What we remember and how we come to understand the events of our lives may be more important for our everyday functioning than whether our memories are perfect replicas of the actual event (Fivush et al., in press). As the excerpts illustrate, children do not simply recall what happened during the hurricane; they are actively trying to understand and process what occurred, as indicated by the many references to their thoughts and emotions. Thus, we were interested in examining the content of children’s recall of Hurricane Andrew in relation to stress in more detail (Sales, Fivush, Parker, & Bahrick, 2002). In particular, based on research on expressive writing about negative events with adults (see Pennebaker, 1997, for a review), we focused on the inclusion of cognition and emotion words that have been related to higher levels of stress and coping with aversive events. Further, given the mixed findings in previous research, we were interested in examining relations among multiple measures of stress and memory. We included two different types of stress measures at the initial interview: Likert scale ratings by mother and child assessing global assessment of experienced stress and more concrete, behaviorally based measures of stress, including the Frederick
258
Posttraumatic Stress Disorder Reaction Index (Frederick, 1985), which asks mothers and children to rate 20 specific behaviors associated with stress (e.g., “Do thoughts about the hurricane make you feel afraid or upset?” and “Have you had bad dreams since the hurricane?”); a child well-being score, which assessed 35 specific behaviors associated with stress (e.g., “Wants to sleep with adults” and “Laughs easily”); and the damage groupings, as previously defined. Only the PTSD Reaction Index was completed at the follow-up interview. As shown seen in table 8.3, the damage assessment, the child well-being survey and the PTSD Reaction Index, which all rate specific physical events or specific child behaviors, were highly correlated with one another, but not related to the more global and subjective Likert scale stress measures. Thus, the most reliable measures of stress seem to be those that decrease the amount of subjective interpretation of the rater and depend instead on more observable events and behaviors. Furthermore, only the objective ratings of stress were related to the content of children’s recall. As shown in table 8.4, which displays correlations among the stress and memory measures, highly stressed children initially included less positive emotion, fewer cognitive processing words, and less information overall than less stressed children. The inclusion of less language indicative of cognitive and emotional processing of the event, as well as shorter narratives, suggests that the highly stressed children had difficulty processing the event immediately following the storm. However, 6 years later they included more negative emotion words and more cognitive words in their recall than less stressed children, indicating that these highly stressed children are still trying to process and understand the event they experienced when they were only 3 and 4 years old to a greater extent than children who were less stressed at the time. In contrast, children who exhibited more emotional and cognitive processing indicative of better coping during their initial interview had better psychological outcomes 6 years later, as indicated by lower PTSD scores. These patterns accord with our previous interpretation that children who were highly stressed by Hurricane Andrew actively tried to avoid thinking about and processing the event. However, this table 8.3 Correlations Between Stress Variables Dmg MSR CSR CWB PTSD1 PTSD2
.43*
.03 -.36* .43** .13
MSR
CSR
CWB
.09
-.25 .19 .16
.16
-.07 .21
-.79**
-.13
PTSD1
.24
Note: Dmg = damage rating; MSR = mother’s rating of child’s stress; CSR = child’s rating of own stress; CWB = child wellbeing score; PTSD1 = Time 1 total PTSD score; PTSD2 = Time 2 total PTSD score. **p
