Page i
Developing Theories of Intention Social Understanding and SelfControl Edited by Philip David Zelazo Janet Wil...
340 downloads
877 Views
3MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Page i
Developing Theories of Intention Social Understanding and SelfControl Edited by Philip David Zelazo Janet Wilde Astington David R. Olson University of Toronto
Page ii
Copyright © 1999 by Lawrence Erlbaum Associates, Inc. All rights reserved. No part of this book may be reproduced in any form, by photostat, microfilm, retrieval system, or any other means, without prior written permission of the publisher. Lawrence Erlbaum Associates, Inc., Publishers 10 Industrial Avenue Mahwah, NJ 07430 Cover design by Kathryn Houghtaling Lacey. Cover art: bill bissett, "th sky within us" (1998, mixed media). Private collection, Toronto, Canada. Reproduced with permission.
Library of Congress CataloginginPublication Data Developing theories of intention : social understanding and selfcontrol / edited by Philip David Zelazo, Janet Wilde Astington, and David R. Olson. p. cm. Includes bibliographical references and index. ISBN 080583141X (cloth : alk. paper)—ISBN 0805831428 (pbk. : alk. paper). 1. Selfcontrol in children. 2. Social perception in children. I. Zelazo, Philip David. II. Astington, Janet W. III. Olson, David R. BF723.S25D46 1998 153.8—dc21 9850446 CIP Books published by Lawrence Erlbaum Associates are printed on acidfree paper, and their bindings are chosen for strength and durability. Printed in the United States of America 10 9 8 7 6 5 4 3 2 1
Page iii
For William Kessen (d. 1999)— humanist, scholar, friend
Page v
CONTENTS Preface
ix
1 Introduction: Actions, Intentions, and Attributions
1
David R. Olson, Janet Wilde Astington, And Philip David Zelazo Part I Development of Intention and Intentional Understanding in Infancy and Early Childhood
15
2 Toddlers' Understanding of Intentions, Desires and Emotions: Explorations of the Dark Ages
17
Andrew N. Meltzoff, Alison Gopnik, And Betty M. Repacholi 3 Intentional Relations and Triadic Interactions
43
Chris Moore 4 Having Intentions, Understanding Intentions, and Understanding Communicative Intentions
63
Michael Tomasello 5 Intentions, Consciousness, and Pretend Play
77
Michael Lewis And Douglas Ramsay 6 Language, Levels of Consciousness, and The Development of Intentional Action Philip David Zelazo
95
Page vi
7 Development of Intention: The Relation of Executive Function to Theory of Mind
119
Douglas Frye 8 Executive Functions and Theory of Mind: Cognitive Complexity or Functional Dependence?
133
Josef Perner, Sandra Stummer, And Birgit Lang 9 The Theory of Ascriptions
153
David R. Olson And Deepthi Kamawar Part II Comparative Perspectives on Intentionality
167
10 Primate Representations and Expectations: Mental Tools for Navigating in a Social World
169
Marc D. Hauser 11 Social Understanding in Chimpanzees: New Evidence From a Longitudinal Approach
195
Daniel J. Povinelli Part III The Sociocultural Context of Intentionality
227
12 Making Sense of the Social World: Mindreading, Emotion, and Relationships
229
Judy Dunn 13 Influences on Maternal Attribution of Infant Intentionality
243
J. Steven Reznick 14 Intention and Emotion in Child Psychopathology: Building Cooperative Plans Jenny Jenkins And Rachel Greenbaum
269
Page vii
Part IV Intentionality and Language
293
15 The Language of Intention: Three Ways of Doing It
295
Janet Wilde Astington 16 Intentionality and Interpretation
317
Carol Fleisher Feldman 17 The Intentionality of Referring
329
Jerome Bruner
Author Index
341
Subject Index
351
Contributors
357
Page ix
PREFACE The current volume grew out of a small conference, Developing Intentions in a Social World, that was held at University College, University of Toronto, in April, 1997. The conference was designed to assess the "state of the art" of research on the development of intention visàvis social understanding and selfcontrol, but it also served to commemorate the university's long history as a center for inquiry into this topic. In 1891, J. Mark Baldwin established the first fully equipped psychology laboratory at University College (apparently the first in the British Commonwealth) and began a seminal series of studies (e.g., Baldwin, 1891, 1892a, 1892b, 1894) on the emergence of intentional imitation in infancy and its relation to the developing socius—the child's sense of self and other. These studies culminated in his landmark books, Mental Development in the Child and the Race (1895) and Social and Ethical Interpretations in Mental Development (1897), the influence of which has been pervasive but, until recently, largely filtered through the writings of Jean Piaget. A renewed interest in the issues addressed directly by Baldwin probably derives in part from several sources including: a growing appreciation of the sociocultural work of Vygotsky and Luria; recent work in developmental neuropsychology on the topic of prefrontal cortical function; and widespread fascination with the problem of consciousness. However, one obvious antecedent of this interest is research on children's beliefs about the mind—their "theory of mind." It will be noted that many of the contributors to the current volume were participants in the 1986 Developing Theories of Mind Conference at the University of Toronto. That conference, organized by Janet Astington, Lynd Forguson, Alison Gopnik, and David Olson, helped to establish theory of mind as a major focus on research. As a result of that research, we have learned a great deal about children's egocentrism and their ideas about thought and misleading appearances. Now, however, many theoryofmind researchers are returning to questions posed more than a century ago by Baldwin: questions about the control of action and how this control is related to children's developing selfconsciousness and their increasingly sophisticated appreciation of other people's perspectives. As will be clear from the
Page x
following chapters, intention is at the intersection of current research on imitation, early understanding of mental states, goaldirected behavior in nonhuman animals, executive function, language acquisition, and narrative understanding, to name just a few of the relevant foci. By bringing these lines of research together in a single volume, we hope to shed light on several questions: What are the psychological processes underlying intentional action? To what extent does the use of intentions (i.e., actionoriented plans) by individuals depend upon socialization? Is it necessary to know about intentions (in self and others) in order to act intentionally? Do individuals attribute intentionality to others by analogy to their own experience of acting intentionally? What is the relation between individual and shared (or mutual) intentions? No doubt, these are extremely difficult questions to answer, but they would seem to be difficult in direct proportion to their importance for an adequate understanding of mind. The talks at the conference were organized into sessions corresponding to the sections in this volume. However, several motifs soon became apparent, including: (1) the possibility of a causal relation between the development of intentional action in infancy and the subsequent understanding of intentional states in others; (2) the potential importance of parental attributions of intentionality to the development of intentional action in infancy and early childhood; (3) the close (and possibly complex) relation between executive function and theory of mind; and (4) the instrumental role that language appears to play in the development of social understanding and selfcontrol. These motifs are maintained in the chapters, which were contributed by nearly all of the 18 participants in the conference (regrettably, James Russell declined due to prior commitments). The conference was presented by the Cognitive Science Program at University College (University of Toronto), under the auspices of the Principal of University College, Lynd Forguson. Principal Forguson convened the conference planning committee, which included Keith Oatley, in addition to the editors of this volume. We would like to thank Lynd and Keith for their hard work and perspicacity, and Coleen McColeman, of the Principal's Office, University College, for her secretarial support to the conference organizers. Of course, the conference could not have been held without generous financial support from several sources including the Connaught Fund; University College; the Department of Psychology; the Office of the Vice President and Provost, University of Toronto; Field Services and Research, the Ontario Institute for Studies in Education/University of Toronto (OISE/UT); and the Laidlaw Research Centre, Institute of Child Study, OISE/UT. We are very grateful for that support. We are also grateful for the assistance of those students who graciously volunteered their time: Paul Andrews, Tate Avery, Carla Baetz, Sandra Bosacki, David Collister, Bruce
Page xi
Homer, Sophie Jacques, Deepthi Kamawar, Elizabeth Lee, Stuart Marcovitch, Bruce Morton, Terri Sloss, Tracy Solomon, and Jessica Sommerville. Finally, we are grateful to Judi Amsel, Executive Editor at Lawrence Erlbaum Associates, for her timely interest and helpful advice, and her unflagging efforts to expedite the publication of this volume. —PHILIP DAVID ZELAZO —JANET WILDE ASTINGTON —DAVID R. OLSON
Page 1
Chapter 1— Introduction: Actions, Intentions, and Attributions David R. Olson Janet Wilde Astington Philip David Zelazo University of Toronto Disciplines, like weedpatches, grow most prolifically at their edges, so perhaps it is not surprising that cognitive science, occupying a space between the biological sciences and the social sciences, should show its greatest signs of growth at its borders. At the first border, we find serious consideration of the question, ''How can a biological mechanism, the brain, give rise to intentional states?" At the second border, we see the equally important question, "How can seemingly private intentional states reflect public forms of talk and action?" Together, these questions capture Brentano's (1874/1973) insight that intentionality is the mark of the mental, and they help to frame developmental questions concerning the origins of intentionality. Development of Intentionality In Brentano's (1874/1973) sense, the term intentionality refers to that property of directedness at an internal object that distinguishes not just propositional attitudes (e.g., belief), but even the simplest mental states (e.g., pain; Brentano, 1874/1973, p. 89 ff.) from those states of the world that are merely physical (see Zelazo, 1996, for discussion). In a series of studies started just over a century ago at the University of Toronto, Baldwin (e.g., 1891, 1892a, 1892b) laid the foundation for his comprehensive account of the origin and development of intentionality (Baldwin, 1897). In this
Page 2
work, Baldwin addressed both the biological and the social aspects of intentionality, and considered intentionality not only in Brentano's (1874/1973) sense, but also in the everyday sense of "purposeful." The everyday sense of intentionality is closely related to Brentano's sense, both etymologically and conceptually (pace Meltzoff, Gopnik, & Repacholi, chap. 2, this volume), because intentional actions are directed at goals in very much the same way that mental states are directed at their objects (i.e., what they are of; see Crane, 1998, for a discussion of common misunderstandings of Brentano's notion of intentionality). Moreover, the two senses are definitionally dependent because, for many authors, intentional action is goaldirected behavior that is accompanied by (or caused by) a particular type of intentional state; namely, an intention (e.g., Adams, 1986). Similarly, Baldwin tied intentional action to the emergence of desire, deliberation, and effort: the conscious representation of a goal, the active consideration of alternative means and ends, and the feeling accompanying the selection and execution of a plan. Although early behaviors are mediated by consciousness, and are often complex and welladapted, they are not deliberate. For Baldwin, the maturation of a coordinating center at the end of the first year of life makes possible the conscious represention of goals and deliberation among means (cf. Olson, 1993, on "holding in mind"; Zelazo, chap. 6, this volume), and hence, intentional action. Baldwin (1897) clearly recognized that the development of intentional action—the deliberate use of goals and means—is inseparable from the development of the infant's understanding of intentionality (in its broader sense) in others. According to Baldwin, ego and alter are ideal ends of a continuum that he defined as the socius (which is a synthesis of Self and Other). Ego (one's sense of Self) and alter (one's sense of the Other) develop together through a circular process. First, one observes projective behavior, that is, behavior seen only in terms of its outward aspects. Then, by imitating this projective behavior, one comes to comprehend the subjective side of it, for example, the affect that accompanies it. Finally, one ejects this subjectivity back into the project. Thus, a father may prick himself with a pin, and his child will imitate the behavior, come to comprehend the painful consequence, and then believe that the father felt it, too. Third and firstperson perspectives on behavior are thereby synthesized. Through the dialectic, the child expands the scope of his or her identifications. Baldwin's (1897) slogan, "It is not I, but I am to become it" (p. 36) anticipated Freud's (1933/1940) famous dictum, "Where It was, there I shall be (Wo Es war; soll Ich werden)" (p. 86). By understanding oneself in relation to the Other one acquires means for acting intentionally and comes to conceptualize oneself as separate from a mindindependent world upon which one can act.
Page 3
The essential role of the Other in the development of intentionality and intentional action has been a central concern of social theorists as well as psychologists. Radical social theorists such as Marx and Durkheim (Lukes, 1973) attempted to explain human behavior by appeal to social, not psychological, causes. The real fact, for such theorists, is "the fact of the dependence of the mental (and not only mental) activity of the individual on the system of culture established before him and completely independently of him, a system in which the [mental life] of every individual begins and runs its course" (Ilyenkov, 1977, p. 78). Vygotsky's (1978; see also Luria, 1961) contribution to this approach was to formulate the relation between mind and society in a developmental framework. According to Vygotsky, children's psychological structures come to resemble those of adults in their culture through the internalization of social practices, and behaviors first engaged in interpersonally come to be managed intrapersonally. This is no less true for simple procedures, such as how to play a game of marbles, than for understanding the intentional states of others. Beliefs, desires, and intentions are thus seen to both cause and reflect participation in a social world (see Bruner, chap. 17, this volume; Feldman, chap. 16, this volume). Although the contributions of social life to the development of intentionality likely are essential, it is clear that children must first possess a set of internal, biologically given, cognitive resources in order to benefit from social and cultural practices. These resources, which continue to develop throughout childhood, are much more scarce among our primate cousins, and possibly absent altogether in other animals (for discussion, see Hauser, chap. 10, this volume; Povinelli, chap. 11, this volume; Tomasello, chap. 4, this volume). What children bring by way of resources, and how this interacts with whatever it is that social interactions provide, remains a central, and still unresolved, issue. Indeed, the interaction between these biologically given resources and the social environment is prominent in recent work, including that of Frye (1981, 1991), Moore (1996; also see Moore, chap. 3, this volume), Meltzoff, Gopnik, and Repacholi (chap. 2, this volume), and Tomasello (chap. 4, this volume), and forms the central theme of the current volume: How are we, as scientists, to understand the interplay between biological and social influences on the development of the ability to have intentional states and attribute them to others? More specifically, however, the work reported in this volume can be seen as addressing a common set of three questions about intentionality in general and intentional action in particular. First, on what grounds can we call behavior intentional (i.e., accompanied by the intentional state of intention)? Second, what is the relation between having intentions and having a concept of intention? Third, what is the function or use of the ability to
Page 4
attribute intentional states (including intention) to oneself and others? These questions will now be considered, in turn. When Does Behavior Become Intentional? It is commonly suggested that children below a certain age, the age of accountability, are not intentional, moral agents. Similarly, other animals, including primates, are not moral creatures in that their actions cannot be judged by appeal to social norms. Praise and blame, as opposed to rewards and punishments, are relevant only to creatures who, as we say, know the difference between good and evil. But what are the criteria for deciding that behavior is intentional, and further, for deciding whether it merits reward and punishment, or praise and blame? Searle (1983) distinguished between intentioninaction and prior intention, the latter being involved in deliberate or premeditated actions (see Astington, chap. 15, this volume). Prior intention allows for planning an action and also for deliberately choosing not to act. It seems safe to suggest that actions based on a prior intention can clearly be construed as moral. However, what about intentioninaction? Certain unpremeditated actions (e.g., spontaneous gestures of comfort, or sudden aggressive acts) are often praised or blamed, although this is probably because we assume that they were motivated by a more general prior intention, to do good or to harm, for example. In considering intentionsinaction, there is an important distinction to be made between behavior that is directly "suggested" (Baldwin, 1891) by a situation—grasping for example—and what is properly referred to as "action" (and associated with intentionsinaction). Actions, in this sense, are distinguished by the dissociation between means and ends, as Baldwin (1892b) and Piaget (1936/1952) noted. They imply an end held in view while means are employed repeatedly or alternative means are sought (cf. Bruner, 1973). Such action may or may not be present in lower animals (see Hauser, chap. 10, this volume; Povinelli, chap. 11, this volume; Tomasello, chap. 4, this volume) but it is perhaps first seen in children at the end of the first year in their persistent attempts at imitation (Baldwin, 1892b). If this is the case, it raises the possibility that even intentioninaction is directly related to the recognition of the mental states of others. As Meltzoff shows, 18montholds imitate what others are intending to do, not what they actually do; that is, they appear to recognize the other's intention (see Meltzoff, Gopnik, & Repacholi, chap. 2, this volume). It might be thought that this creates a problem, in that the recognition of belief in others, as indicated by an understanding of false belief (Wimmer & Perner, 1983), does not appear until about 4 years of age. However, it may be that nonepistemic states, such as an intention, are more easily recognized, or it may be that the early recognition of intention does not
Page 5
involve metarepresentation (Astington, 1991). Certainly, other things we associate with the metarepresentation of intention, such as selfcontrol or acting in terms of a social or moral norm, continue to develop well past infancy. There are a number of important developmental achievements here, and the chapters in the first section of this volume attempt to sort them out (e.g., Lewis & Ramsay, chap. 5, this volume; Meltzoff, Gopnik, & Repacholi, chap. 2, this volume; Moore, chap. 3, this volume; Zelazo, chap. 6, this volume). Intentions and Concepts of Intention Historically, intention played a central role in psychological explanation. Conation or volition was one of the trinity of mental faculties—cognition, conation, and affect—that came down to us from classical times. Intention was changed from a faculty into a function in modern psychology by William James, John Dewey, and J. Mark Baldwin, among others. Baldwin (1897), for example, argued that a sense of personal agency arose in the context of solving problems (e.g., imitating others), and lay at the root of social consciousness. However, once one can do as one likes, one can also be held accountable for what one does and this accountability appears to play a role in the further development and elaboration of the concept of intention. Thus, the nature and development of the concept of intention is a central topic in moral development and the development of social cognition, and has been for some time. Nonetheless, it seems likely that the topic of intention has inherited some of its recent popularity from the now widespread interest in children's theory of mind. Like the present volume, Developing Theories of Mind (Astington, Olson, & Harris, 1988) grew out of a conference held at the University of Toronto and helped to crystallize what has become one of the liveliest areas of developmental research. In the introduction to that volume it was argued that in acquiring a theory of mind, "children begin to recognize mental states in themselves and others; they come to recognize beliefs as beliefs, desires as desires, and intentions as intentions" (Olson, Astington, & Harris, 1988, p. 1). Although much has since been learned about children's understanding of intentional states such as belief and, to a lesser extent, desire, we still know relatively little about children's understanding of intention. We know arguably even less about the development of intention itself, and the relation between intention and concepts of intention. In the rush of new work investigating children's theory of mind, it quickly became clear that important developments occur around 4 years of age in children's reasoning about beliefs. Children younger than age 4 tended to attribute to others the beliefs they themselves held, namely, those congruent with the current situation. Those older than age 4 ascribed
Page 6
beliefs on the basis of the causes of those beliefs, namely, what the other had seen, heard, and done. Consequently, these children could ascribe to others beliefs that they themselves knew to be false (hence the significance of the "false belief task" to the understanding of children's theory of mind; Wimmer & Perner, 1983). To some, this development seemed little more than yet another example of Piagetian egocentrism, rather than evidence of failure to understand beliefs, because children did ascribe to others the beliefs they held themselves. However, the picture was soon complicated by the finding that children were no better at understanding their own beliefs than those of others (Gopnik & Astington, 1988; Wimmer & Hartl, 1991). In contrast, children much younger than 4 years of age can understand that others may have desires different from their own, and can remember earlier desires, different from their present ones (Gopnik & Slaughter, 1991; Repacholi & Gopnik, 1997; but see Moore et al., 1995). Such young children also understand that desires motivate actions and they refer to desire states in their speech a year or more before they refer to beliefs (Wellman, 1991). Whereas the concepts of belief and desire quickly found their place in developmental and evolutionary theories of mind, the concept of intention has remained enigmatic. Intentions are more like desires than beliefs. That is, they are neither true nor false but are fulfilled (or not) by outcomes in the world and indeed, serve to motivate action to bring about those outcomes. Certainly, much of the time our desires and our intentions coincide. Such considerations may lead to the assumption that children understand intention when they understand desire. There are, however, important differences between desire and intention (Astington & Gopnik, 1991). Intentions are selfreferential, that is, they are not fulfilled by outcomes in the world unless the intention itself causes the action that brings about that outcome (Searle, 1983). Intentions are also more closely tied to beliefs (Moses, 1993); one can desire, but not intend, what one believes to be impossible. Frequently, theory of mind is used to explain behavior as a product of belief and desire, leaving unexplored just how beliefs and desires turn into intentions or how prior intentions and plans are formed or executed. Perhaps we may get a clearer picture of how intentions are related to other intentional states by thinking of intentional states as generated by a mental calculus. The Mental Calculus Consider the problem of determining when behavior is directly caused by external or internal stimuli and when behavior is, rather, a product of mental representation, including belief, desire, and intention. Can an animal's behavior, say that of a dog chasing a cat, be correctly characterized as intentional? The dog certainly has a goal, uses a variety of means, and even maintains the memory of the goal over time.
Page 7
However, intentional states, including beliefs, desires, and intentions, are arguably constituted by a calculus of agents, modes, and contents: AGENT
MODE
CONTENT
John—
believes—
the dog wants to go out
Mary—
intends—
to let the dog out
Sam—
wants—
the dog to be out
Searle (1983) remarked, "Only someone in the grip of a philosophical theory would deny that small babies can literally be said to want milk and that dogs want to be let out or believe that their master is at the door" (p. 5). However, if a creature can manifest only one of the modes, say, intending, then there are less secure grounds for saying that the behavior is intentional. Hence, contra Searle, the dog who, when shown a bone, can only try to get the bone, rather than think that it is a bone, may not be well described as intentional. Instead, the behavior would seem to be under the control of a schema for bone getting (suggested by the bone). It may be that only linguistic creatures with access to the sort of mental calculus just mentioned, creatures who can both treat expressions as asserting something as true ("This is a bone") and initiate an action can be correctly described as intentional in that they have a set of optional modes for relating themselves to the world. Agreeing on criteria (including linguistic criteria) for recognizing intention remains a critical task, and is explored in more detail in several of the chapters in this volume (Astington, chap. 15, this volume; Feldman, chap. 16, this volume). Structural and Functional Explanations of Intentional Understanding Although the origins of an understanding of intentional states in general and intentions in particular may be traced to structural changes in the organization of the mind and ultimately the brain, the origins of this understanding can also be examined in terms of the functions that these understandings sustain. Piaget's (1936/1952) account of the development of the representational function is a classic example of a structural change. A more recent structural change explanation of the development of an understanding of intentionality was articulated by Leslie (1987), who proposed a specific "decoupling" device that serves to sever the link between situation and content, this content thereby being eligible for metarepresentation. Others in this tradition have pointed to other plausible mechanisms: changes in the executive functions of prefrontal cortex that allow perspective taking (for discussion see Frye, chap. 7, this volume; Perner, chap. 8, this volume; Zelazo & Frye, 1998); changes in memory (e.g., Gordon & Olson,
Page 8
1998; Olson, 1993) that allow children to "hold in mind" a representation across changing situations; or the development of higherorder levels of consciousness that support selfreflection and psychological distancing (Lewis & Ramsay, chap. 5, this volume; Zelazo, chap. 6, this volume). But, it may be argued, structural change occurs to sustain particular functions and it is the function of participating in a social world that is responsible for the growth of an understanding of intentional states. The functions served by such an understanding are both social and cognitive. Socially, the attribution of intentional states allows for increasingly sophisticated social actions and interactions. Not only does it allow the child to recognize the intention behind the action of another, and therefore, to cooperate in the achievement of the action, but it also allows for the transmission of culture through learning and imitation, a point that is explored in the chapters in the third section of this volume (Dunn, chap. 12, this volume; Jenkins & Greenbaum, chap. 14, this volume; Reznick, chap. 13, this volume). The cognitive functions served by the development of intentionality are metacognitive ones, allowing a new kind of consciousness of mental life, not only for introspection of one's own thoughts but also of the sources of one's own beliefs. This is the reflective kind of consciousness that usually warrants the label thinking. Such metacognitive competence is critical to the formation of planned, premeditated actions, the kind of deliberate actions we associate with the "age of accountability." Indeed, the planfulness of one's own actions is identical in form to the participation in cooperative actions with others, a contingency that invites the inference that this aspect of cognitive development is more appropriately thought of as social development (see Jenkins & Greenbaum, chap. 14, this volume). A recent focus on individual differences in the development of intentional understanding has shown this understanding to be related to certain characteristics of children's social interactions (Dunn, chap. 12, this volume). For example, demographic approaches find relations between a child's understanding of false belief and the size of the child's family, particularly the number of siblings the child has (Jenkins & Astington, 1996; Lewis, Freeman, Kyriakidou, MaridakiKassotaki, & Berridge, 1996; Perner, Ruffman, & Leekam, 1994). A relation also exists between children's theory of mind and family background measures, such as parental occupation and education (Cutting & Dunn, in press), or parenting style (Vinden, 1997). Such findings lead to the argument that involvement in social relations is the precise causal mechanism through which social understanding and selfcontrol are acquired. Engagement in a social world, especially the world of complex social interactions mediated by language, is seen to be the immediate cause of the child's acquisition of an understanding of his or her own mental states, as well as those of other people. Dunn (chap. 12, this
Page 9
volume) cautions against making causal inferences from such correlational data; strong evidence, best provided by longitudinal and training studies, is required to support the causal argument. We do, however, know that particular patterns of social interaction and language use in the home, such as sibling cooperation, social pretend play, and family talk about feeling states, are associated with later differences in false belief and emotion understanding (Dunn, Brown, Slomkowski, Tesla, & Youngblade, 1991; Youngblade & Dunn, 1995). In addition, participation in training procedures involving, for example, conversations about surprise, has been shown to promote false belief understanding (Appleton & Reddy, 1996). Consequently, social factors do appear to have an impact on the child's acquisition of a theory of mind. Such social perspectives are sometimes seen to be in conflict with more cognitive ones insofar as these perspectives seem to see little need for specific cognitive mechanisms to acquire adult forms of competence (Raver & Leadbeater, 1993). Sometimes, these perspectives seem to assume that the adult forms of competence are the inevitable product of practice interpreting actions and speech in a social world. Nonetheless, most researchers agree that developing cognitive mechanisms are involved, and it remains to be determined how best to characterize these mechanisms. Any complete account would have to acknowledge both the structural and functional properties of intentionality: There are brain mechanisms that make the formation of intentional states possible; there are functional roles that these states play; these functional roles are, at least in part, and perhaps originally, social ones; and finally, these states can themselves become subject to reflection. Organization of the Volume The chapters in this volume, too rich to be summarized briefly, are organized into four sections, although a number of themes cut across the four sections. The first and longest section, focusing on the development of intention and intentional understanding in infancy and early childhood, is central to all of the issues we have raised in the introductory chapter. For example, Meltzoff, Gopnik, and Repacholi (chap. 2) argue that the origins of social understanding and selfcontrol can be seen in neonatal imitation, and they trace the development of social understanding from this early, intrinsically interpersonal context through what they refer to as the "dark ages" of the toddler period. Tomasello (chap. 4) argues that at the end of the first year, infants can infer others' intentions because they recognize the intentionality of their own actions, and because they recognize others as creatures like themselves. Zelazo (chap. 6) argues that increasingly higher levels of consciousness, mediated by language, allow more complex knowledge structures to govern children's intentional actions.
Page 10
Lewis and Ramsay (chap. 5) also discuss the development of increasingly higher order levels of intention that yield a selfreflective understanding. Frye (chap. 7) and Perner, Stummer, and Lang (chap. 8) consider how best to characterize the wellestablished relation between action control and theory of mind in preschoolers. Some authors' focus is not on the development of intention per se; instead, they are concerned with intentionality in the broader sense. For example, Moore (chap. 3) discusses infants' understanding of attention within an interactive context, arguing that infants understand intentional relations as a property of the interaction, not as a property of either the self or the other exclusively. Olson and Kamawar (chap. 9) discuss intentional attributions, focusing on belief attribution, and propose a strict criterion for the possession of intentional states, that is, the ability to ascribe such states to self and other. The chapters in the second section focus on comparative perspectives on intentionality; they too are concerned with issues of understanding and attributing intentional states, broadly considered. Hauser's (chap. 10) topic is the extent to which humans and other primates differ with respect to the causal relation between representation and expectation. He discusses primates' expectations concerning communicative signals, selfpropelled motions, and the greatest challenge, mental states, suggesting that primates know something about communication and animacy, if not belief. Povinelli (chap. 11) summarizes his extensive body of research on chimpanzees' understanding of seeing, leading to the perhaps startling conclusion that although chimpanzees follow another's gaze, there is no evidence for an intentional understanding of seeing, even though similar gazefollowing behaviors in children are correctly so interpreted. The chapters in the third section lead on from there to discuss the peculiarly human sociocultural context of intentionality. Dunn (chap. 12) makes a compelling argument for the significance of the emotional context in children's acquisition of intentional understanding. She argues that it is emotionally charged social experiences that play a vital role in the development of social understanding, right from the beginning. Reznick (chap. 13) shows how readily mothers attribute intentionality to infant behaviors displayed on videotape or described in a questionnaire, and Dunn (chap. 12) makes the point that in the real world such interactions are rarely emotionally neutral. Jenkins and Greenbaum (chap. 14) are also concerned with the link between cognition and emotion in their investigation of children's ability to engage in cooperative endeavors. They find that children whose emotional lives are dominated by feelings of anger and aggression are less able to negotiate cooperative goals, and less skilled in using intentional terms to discuss joint plans. The fourth and final section, on intentionality and language, picks up a number of issues from the earlier chapters: the themes that cut across the
Page 11
four sections. Astington's (chap. 15) discussion of the mediating role of language in the development of intentional understanding relates to Jenkins and Greenbaum's (chap. 14) findings, and reflects back to Zelazo's (chap. 6) focus on the role of language in the genesis of action. The acquisition of language depends on the infant's understanding of communicative intentions, as Tomasello (chap. 4) argues, and the development of language allows for the ascription of intentionality, as Olson and Kamawar (chap. 9) argue. Feldman (chap. 16) emphasizes that intentions are understood within an interpretive community, which links back to the chapters on the sociocultural context of intentional understanding, and also to Moore's (chap. 3) argument that intentional relations are first construed as part of the interactive context, not as independently belonging to self or other. Bruner's (chap. 14) focus on infant recognition and parental attribution of epistemic and intrumental intentions relates to Astington's (chap. 15) and Feldman's (chap. 16) concerns, and Tomasello's (chap. 4) and Reznick's (chap. 13) chapters also find echoes here. In conclusion, cognitive science, as heir to the discipline that William James (1890/1950, p. 1) defined as ''the Science of Mental Life" centers on intention and intentional understanding—the ability to make judgments of truth and goodness, and the ability to act intentionally in terms of those judgments. These abilities may well turn out to be found only in persons with sophisticated cognitive and linguistic powers—powers that are not only representational, but also metarepresentational and, more to the point, the product of participation in a social world. If so, then a full understanding of these abilities will require understanding of the cognitive and cultural determinants of intentionality and its development. We believe that the contributors to this volume take important steps toward this goal. References Adams, F. (1986). Intention and intentional action: The simple view. Mind and Language, 1, 281301. Appleton, M., & Reddy, V. (1996). Teaching 3yearolds to pass false belief tests: A conversational approach. Social Development, 5, 275291. Astington, J. W. (1991). Intention in the child's theory of mind. In D. Frye & C. Moore (Eds.), Children's theories of mind (pp. 157172). Hillsdale, NJ: Lawrence Erlbaum Associates. Astington, J. W., & Gopnik, A. (1991). Developing understanding of desire and intention. In A. Whiten (Ed.), Natural theories of mind: Evolution, development and simulation of everyday mindreading (pp. 3950). Oxford: Basil Blackwell. Astington, J. W., Olson, D. R., & Harris, P. L. (Eds.). (1988). Developing theories of mind New York: Cambridge University Press. Baldwin, J. M. (1891). Suggestion in infancy. Science, 18, 113117. Baldwin, J. M. (1892a). Infants' movements. Science, 19, 1516. Baldwin, J. M. (1892b). Origin of volition in childhood. Science, 20, 286288.
Page 12
Baldwin, J. M. (1897). Social and ethical interpretations in mental development: A study in social psychology. New York: Macmillan. Brentano, F. (1973). Psychology from an empirical standpoint. (O. Kraus, Ed.; A. C. Ranculrello, D. B. Terell, & L. L. McAlister, Trans.). London: Routledge & Kegan Paul. (Original work published in 1874) Bruner, J. (1973). Organization of early skilled action. Child Development, 44, 111. Crane, T. (1998). Intentionality as the mark of the mental. In A. O'Hear (Ed.), Current issues in philosophy of mind (pp. 229251). Cambridge, England: Cambridge University Press. Cutting, A., & Dunn, J. (in press). Theory of mind, emotion understanding, language and family background: Individual differences and interrelations. Child Development. Dunn, J., Brown, J., Slomkowski, C., Tesla, C., & Youngblade, L. (1991). Young children's understanding of other people's feelings and beliefs: Individual differences and their antecedents. Child Development, 62, 13521366. Freud, S. (1940). Neuefolge der Vorlesungen zur Einfhrung in die Psychoanalyse [New introductory lectures on psychoanalysis]. In A. Freud, E. Bibring, & E. Kris (Eds.), Gesammelte Werke: XV (Whole volume). London: Imago Publishing Co. (Original work published 1933) Frye, D. (1981). Developmental changes in strategies of social interaction. In M. E. Lamb & L. R. Sherrod (Eds.), Infant social cognition (pp. 315331). Hillsdale, NJ: Lawrence Erlbaum Associates. Frye, D. (1991). The origin of intention in infancy. In D. Frye & C. Moore (Eds.), Children's theories of mind (pp. 1538). Hillsdale, NJ: Lawrence Erlbaum Associates. Gopnik, A., & Astington, J. W. (1988). Children's understanding of representational change and its relation to the understanding of false belief and the appearance reality distinction. Child Development, 59, 2637. Gopnik, A., & Slaughter, V. (1991). Young children's understanding of changes in their mental states. Child Development, 62, 98110. Gordon, A. C. L., & Olson, D. (1998). The relation between acquisition of a theory of mind and information processing capacity. Journal of Experimental Child Psychology, 68, 7083. Ilyenkov, E. V. (1977). The concept of the ideal. In R. Daglish (Ed. and Trans.), Philosophy in the USSR (pp. 7199). Moscow: Progress Publishers. James, W. (1950). The principles of psychology. New York: Dover. (Original work published in 1890). Jenkins, J. M., & Astington, J. W. (1996). Cognitive factors and family structure associated with theory of mind development in young children. Developmental Psychology, 32, 7078. Leslie, A. M. (1987). Pretense and representation: The origins of "theory of mind." Psychological Review, 94, 412426. Lewis, C., Freeman, N. H., Kyriakidou, C., MaridakiKassotaki, K., & Berridge, D. M. (1996). Social influences on false belief access: Specific sibling influences or general apprenticeship. Child Development, 67, 29302947. Lukes, S. (1973). Emile Durkheim: His life and work. Markham, ON: Penguin Books. Luria, A. R. (1961). The role of speech in the regulation of normal and abnormal behaviour (J. Tizard, Ed.). New York: Pergamon Press. Moore, C., Jarrold, C., Russell, J., Lumb, A., Sapp, F., & MaCallum, F. (1995). Conflicting desire and the child's theory of mind. Cognitive Development, 10, 467482. Moses, L. J. (1993). Young children's understanding of belief constraints on intention. Cognitive Development, 8, 125. Olson, D. R. (1993). The development of representations: The origins of mental life. Canadian Psychology, 34, 114. Olson, D. R., Astington,J. W., & Harris, P. L. (1988). Introduction. In J. W. Astington, P. L. Harris, & D. R. Olson (Eds.), Developing theories of mind (pp. 1 15). New York: Cambridge University Press.
Page 13
Perner, J., Ruffman, T., & Leekam, S. R. (1994). Theory of mind is contagious: You catch it from your sibs. Child Development, 65, 12281238. Piaget, J. (1952). The origins of intelligence in children (M. Cook, Trans.). New York: Norton. (Original work published in 1936). Raver, C. C., & Leadbeater, B.J. (1993). The problem of the other in research on theory of mind and social development. Human Development, 36, 350362. Repacholi, B. M., & Gopnik, A. (1997). Early reasoning about desires: Evidence from 14 and 18montholds. Developmental Psychology, 33, 1221. Searle, J. (1983). Intentionality. Cambridge: Cambridge University Press. Vinden, P. (1997, April). The effects of parenting style on theory of mind understanding. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Washington, DC. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press. Wellman, H. M. (1991). From desires to beliefs: Acquisition of a theory of mind. In A. Whiten (Ed.), Natural theories of mind: Evolution, development and simulation of everyday mindreading (pp. 1938). Oxford: Basil Blackwell. Wimmer, H., & Hard, M. (1991). Against the Cartesian view on mind: Young children's difficulty with own false belief. British Journal of Developmental Psychology, 9, 125138. Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children's understanding of deception. Cognition, 13, 103128. Youngblade, L. M., & Dunn, J. (1995). Individual differences in young children's pretend play with mother and sibling: Links to relationships and understanding of other people's feelings and beliefs. Child Development, 66, 14721492. Zelazo, P. D. (1996). Towards a characterization of minimal consciousness. New Ideas in Psychology, 14, 6380. Zelazo, P. D. & Frye, D. (1998). II. Cognitive complexity and control: the development of executive function in childhood. Current Directions in Psychological Science, 7.
Page 15
PART I DEVELOPMENT OF INTENTION AND INTENTIONAL UNDERSTANDING IN INFANCY AND EARLY CHILDHOOD
Page 17
Chapter 2— Toddlers' Understanding of Intentions, Desires, and Emotions: Explorations of the Dark Ages Andrew N. Meltzoff University of Washington Alison Gopnik University of California, Berkeley Betty M. Repacholi Macquarie University Our sensory experience of other people tells us about their movements in space but does not tell us directly about their mental states. Although a few radical philosophers and psychologists may deny the existence of mental states, most regular "folk" feel sure that they themselves and others have them. It is adaptive to read another person's mental state because it allows us to explain the actions they have taken in the past and predict their actions in the future. The general aim of "theory ofmind" research is to illuminate the development of this everyday, folk psychological framework for understanding people (e.g., Astington & Gopnik, 1991b; Astington, Harris, & Olson, 1988; Flavell & Miller, 1998; Perner, 1991; Taylor, 1996; Wellman, 1990). This research has taught us that children's understanding of mental life is not all of one piece. There is no single moment at which children develop a theory of mind. Instead, children gradually converge on an adult understanding of mind. The focus on when children understand "false belief" has been misleading in this regard. Beliefs are only one of many mental states that children understand and use in their everyday interactions with people. Children may only develop a firm understanding of false beliefs at about 4 years old, but they have started on their path of developing a folk psychological understanding of people much earlier. Preschoolers understand a great deal about perceiving, wanting, and intending at an age when they still have only a shaky understanding of false beliefs (e.g., Astington & Gopnik, 1991a; Flavell, Flavell, Green, & Moses, 1990; Gopnik & Slaughter, 1991; Gopnik, Slaughter, & Meltzoff, 1994; Wellman, 1990, 1993).
Page 18
However, just as it is a distortion to think that children don't have an understanding of mind until they pass a falsebelief exam, it is likewise a distortion to hold that infants have the adult conception of the mind as soon as they show a special interest in people. An alternative is a genuinely developmental account of children's understanding of the mind. The view we favor is that infants are given a jump start in understanding people because of certain innate structures, but they gradually come to understand the whole range of psychological flora and fauna including pretenses, images, emotions, perceptions, desires, intentions, and beliefs. Newborns do not have anything like this full understanding of the mind, but they do have privileged ways of understanding other people and human acts. In earlier work, we argued that infant imitation may provide the first groundwork for later understanding of the mind (Gopnik & Meltzoff, 1994; Meltzoff & Gopnik, 1993). Imitation is a behavioral measure indicating that newborns, at some level of processing no matter how primitive, can map actions of other people onto actions of their own body. The findings of early imitation have now been replicated and extended in 25 different studies from 13 independent laboratories, both in this country and crossculturally (for a history and literature review, see Meltzoff & Moore, 1977, 1994, 1997). A comprehensive model of early imitation was offered by Meltzoff and Moore (1997) and dubbed the AIM (active intermodal mapping) model. The central notion is that imitation, even early imitation, is a matchingto target process. The goal or behavioral target is specified visually. Infants' selfproduced movements provide proprioceptive feedback that can be compared to the visually specified target. AIM proposes that when babies imitate, they are linking the visual appearance of other people to their own internal kinesthetic and proprioceptive feelings, connecting the visible bodily actions of others and their own internal states. This type of initial state would provide a jump start for infants' understanding of persons and commonsense psychology because it provides the first and most fundamental building block of the folk psychological framework: "Those entities are like me." Thus, when newborns look at the moving adults, they do not simply see "visual complexity," "highcontrast areas," or mere physical motions, but special acts that are like the acts they can and do perform. Newborns are not alone; they perceive that other entity is "like me."1 1
In using the English word "me," we do not suggest that the infant has the fullfledged adult sense of self. Indeed, we have argued that such a sense of self is a developmental product (Meltzoff & Moore, 1995). Our argument could be rephrased by purging the "me" word and instead saying, "That looks like this feels." Elsewhere we have attempted to describe the initial state in a precise technical manner, using a computational model and avoiding the glosses of everyday English (Meltzoff & Moore, 1997). Interested readers are referred to this work for detailed arguments about early selfother relations.
Page 19
Development in social cognition depends on twoway traffic between self and other, on what might be called "projection" from one's own case to the other and "appropriation" from the other to the self. But this depends on a prior assumption that self and other have anything whatever to do with one another. Newborn imitation provides a demonstration that at some primitive level this link has been made in the normal infant. Without this fundamental connectedness, there would be no reasoning bidirectionally from one's own case to another's because the two cases would not be known to be similar (Meltzoff & Moore, 1995). Regardless of our theory about the initial state, there is, admittedly, a substantial gap in the findings between early infancy and early childhood. We know something about the initial state of newborns from studying imitation and other early phenomena like interactional synchrony and face recognition. We know something about the state of 3yearolds who are on the verge of understanding belief. However, "the dark ages," from about 15 to 36 months, remain something of a mystery. Ask graduate students to test 2yearolds and they will often shudder and scurry out of the lab; the "terribletwos" lead to both subject and graduate student attrition. In the dark ages, the established techniques of infancy (e.g., preferencefornovelty procedures) do not work because the children are too old to sit and passively watch; conversely, tests demanding subtle verbal distinctions do not work ("When I first asked you, before we did X, what did you believe . . ."), because toddlers are too young for such verbal gymnastics. We're left guessing. The absence of empirical findings from this period has contributed to our difficulty in laying out a finegrained developmental theory. It is as if biologists had only seen frogs and tadpoles without the transitions in between. It would be hard to tell a developmental story, and no one would believe it if you did.2 A variety of techniques, however, have recently been developed to test children during the dark ages. One set of techniques uses toddlers' language abilities. Such studies suggest, for example, that 18monthold children understand that words refer to objects and can use an adult's attentional cues (e.g., gaze direction, gestures) to identify the referent of a novel label (e.g., Baldwin, 1993a, 1993b; Baldwin et al., 1996). At a similar age, children also take into account the intentions of the other person in their attempts to determine the referent of a novel word (e.g., Tomasello, 1995; Tomasello 2
We do not mean to imply that there has been no work in the 15 to 36monthold age range. There has been a great deal of work (e.g. Damon, 1998; Kagan, 1981), but not much from the "theoryofmind" viewpoint linking what infants know about persons and what 3 to 5yearolds know about the intentional mental states of persons. Researchers have discovered a good deal about social cognition in infancy and a good deal about social cognition in 3to 5yearolds, but not enough about what happens in between. We expect that future research will shed increasing light on "the dark ages."
Page 20
& Barton, 1994; Tomasello, Strosberg, & Akhtar, 1996). Finally, analyses of naturalistic studies of early conversation have illuminated children's understanding of the mind (e.g., Bartsch & Wellman, 1995). A second newly developed technique, called the "behavioral reenactment procedure" (Meltzoff, 1995), also explores what children know about the mind, but does not rely on language. The behavioral reenactment procedure capitalizes on children's natural tendency to reenact or imitate the behaviors they see, but uses it in a more abstract way. A host of research indicates that children, even young infants, do not imitate by rote; they do not immediately imitate the events they see, but rather their interpretation of them (Meltzoff & Moore, 1995, 1997, 1998). This means that if we arrange a situation correctly, we can use their natural tendency to reenact adult behavior as a "read out" of how they understand the world. Such an approach has been extensively used in the psycholinguistic literature to assess children's linguistic structure. This work shows that children who are asked to imitate a sentence of the adult language tend to respond with a similar sentence, often synonymous with the tobeimitated one, but which conforms to the child's own linguistic rules. The behavioral reenactment procedure uses the imitation of goaldirected acts to examine the psychological structures children use in interpreting human behavior. A third set of techniques capitalizes on children's very early tendency to read meaning into human emotional expressions. This underlies early "social referencing" studies but also has been developed in a more sophisticated way by Repacholi (Repacholi, 1998; Repacholi & Gopnik, 1997). There is evidence that basic emotions such as happiness, sadness, and disgust are associated with particular facial expressions from early infancy and universally across cultures (e.g., Darwin, 1872; Ekman, 1980). Emotions are closely and intricately connected to intentions and desires in our adult understanding of mind. In the everyday adult framework, we assume that getting what we want or acting as we intend to will lead to happy emotions, whereas failure will lead to negative emotions. We also assume that we act in a way that will bring about positive but not negative emotions. Some of the new techniques we discuss in this chapter exploit the early nonverbal ability to read emotional expressions as a way of investigating children's understanding of the mind. Taken together, these recent procedural advances in addressing questions to very young children are starting to reveal some of what children understand in the dark ages. There are two important foci of children's developing understanding of the mind in this period. One is their understanding of perception and attention (e.g., Baldwin, 1993a, 1993b; Gopnik, Slaughter, & Meltzoff, 1994; Gopnik & Wellman, 1994; O'Neill, 1996; Slaughter & Gopnik, 1996; Tomasello, 1995). The other is an understanding
Page 21
of what Searle (1983) called "worldtomind states" such as desire and intention. We focus on the latter in the present chapter. Understanding Directedness In fullfledged adult psychology, an important feature of desires and intentions is that they are "directed at" objects and events. In fact, in adult psychology this is one thing that distinguishes desires and intentions from mere moods, feelings, or emotions and makes them similar to perceptions and beliefs. At the same time, desires, intentions, and emotions, unlike perceptions and beliefs, share what we might call valence. Desires carry with them an implication of certain positive or negative attitudes towards objects and events. Getting what we want is good, and being frustrated is bad. Doing what we intend to do is good, failing to do so is bad. A Conceptual Clarification: Intention and Intentionality What we are calling the "directedness" of these states sometimes is referred to in psychology as the "intentional" character of desire and intention, but this derives from a misunderstanding (or loose adaptation) of a technical philosophical term. Intention in the technical philosophical sense refers to the propositional character of a mental state, not solely the fact that it is directed at objects in the world. This is an important distinction inasmuch as some mental states may be directly or causally related to real objects or events in the world, without being intentional. The classical philosophical example is "seeing" (as opposed to "believing" or ''seeing that"). This mental state is related to real objects and events in the world, not to mental representations of events, and this has important consequences. I can substitute different descriptions of the same event and preserve the truth of the sentence when "see" is used in the nonintentional sense. For example, it is true to say that someone sees the author of Waverley when they see Scott, but it is not true to say that someone, who believes this person is Scott, also believes that he is the author of Waverley. The same holds for worldtomind states such as wanting and intending. These states may simply be directed at objects or events, which means that descriptions can be substituted preserving truth, or they may be genuinely intentional (in the technical, philosophical sense), which means that they cannot. The terminological distinction is important for developmental psychologists because it lets us discriminate between two different ways young children might understand the "aboutness" or "directedness" of mental states. It also should prevent us from assuming that if a child has the minimal idea of the directedness of mental states, they must also have an under
Page 22
standing of the fact that mental states have propositional content, involve representations, and therefore are "intentional." In fact, there is likely a developmental change from children first understanding the nonintentional aspects of mind to later understanding its intentional character. To make things even worse, these philosophical concerns about what it means to be "intentional" have literally nothing to do with the everyday use of "intention," as in the intention to act—it is just a homonym, although it is related to the "intension" of sentences (no wonder philosophers prefer to talk about x's and y's). To assume a deep link between the two intentions, as psychologists sometimes do, is like assuming that the Federal bank must be an historical outcome of the economic importance of rivers. Of course, intentions, in the sense of intending to act, are mental states and can be "intentional" (in the sense of having propositional content), although they need not necessarily be, just as desires, perception, and other mental states may or may not be ''intentional." But then the Federal bank may (for all we know) have something to do with the economic importance of rivers. The similarity of the two words is obviously not an indication of any deeper link between the two concepts: Beliefs are "intentional" mental states par excellence, and there's no homonym at play in this case. Finally, it may be true (we think it is) that the earliest "intentional" concepts of children concern intention, desire, and perception, but if so, this will be an empirical discovery of developmentalists, not a logical truth embodied in the homonyms. Exploring Toddlers' Understanding of the Directedness of Emotional Attitudes Infant imitation and other phenomena of early infancy show that infants can link their own feelings and those of others. However, these early behaviors do not involve objects. The feelings that children understand at first are just that, purely internal feelings. When do infants understand both the valenced and directed character of adults' attitudes toward objects? The literature on social referencing suggests to some that this may be understood as early as 9 months of age. In these studies, mothers reacted to objects and events with particular emotions, and babies seemed to adopt these emotional attitudes. However, a closer look at the experiments in that paradigm suggests that this conclusion may be unwarranted. The fact that infants adopted the mother's attitude does not demonstrate that they understood that this attitude was directed at a particular object. First, typically only one object is presented, so it remains unclear whether infants truly understood that the emotional message was directed toward this and not other objects. Second, the fact that infants adopted the mother's attitude does not necessarily mean that they understood that her
Page 23
attitude was directed at any object whatever. Various forms of emotional contagion and/or simple associative processes (e.g., temporal contiguity, stimulus salience) could be at work in these studies, as argued by advocates of a "lean interpretation" of the standard social referencing work (Baldwin & Moses, 1994). To demonstrate that children really understand directedness, you need to show that they understand that a person can have one attitude toward one object and a different attitude toward another object. At least two objects should be presented, and these should be equally attractive and salient. Some researchers have recently adopted a dualobject display (e.g., Baldwin & Moses, 1994; Mumme, Won, & Fernald, 1994), but this by itself has not been sufficient to eliminate the possibility of all simpler nonreferential processes (see Repacholi, 1998, for a discussion). Repacholi (1998) developed a technique to test infants' understanding of directedness of emotional signals that rules out nonreferential mechanisms. In these studies, 14monthold infants saw the same experimenter produce two different emotional expressions, an expression of disgust and an expression of joy, toward two different objects. The objects were in closed boxes so that the children did not see the objects at the time that they saw the emotions. Instead they saw the experimenter peek into each box and make a different emotional response. The children were then given the two boxes. The results showed that children touched and handled both boxes equally quickly and frequently, indicating that there was no simple emotional contagion at work. The important result was that they opened the box with the "happy" object inside significantly more frequently than they opened the box with the "disgust" object. We conclude that the children had inferred that the adult's attitude was specifically directed at the object inside the containers, even though they had not seen the emotions and the objects at the same time. On the basis of these results, we can also firmly address the temporal contiguity and salience issues that have bedeviled socialreferencing research. How do we know that infants were not simply noting the temporal contiguity between the emotional signal and whichever stimulus they were looking at, at the time the signal was issued? This is ruled out because the only visible objects were the two boxes, not the objects they contained. Moreover, infants should have connected an emotion to whichever box was the focus of their own attention when the expression was displayed, and the results showed that this was not the case (because they handled both boxes equally). Similarly, the paradigm rules out the possibility that simple salience was at work. Both boxes were extremely salient and the experimenter's actions (e.g., picking the boxes up, opening their lids) made them all the more so. Yet infants did not link these salient stimuli
Page 24
to the emotions: They were not loathe to touch and examine either box, even the box containing the disgust object. They just did not want to grab the object that was hidden inside that box. Because the object was hidden, it could hardly have been visually salient when the emotion was originally displayed. We conclude that the interpretation of the social referencing demonstrated in 9 to 13monthold infants is still equivocal. Perhaps a "lean" interpretation of those effects are still in order. However, the work discussed here shows that by 14 months infants genuinely understand that emotional attitudes with particular valences may be directed at particular objects. They understand that the objects to which an adult's emotions are directed may be ones that are neither perceptually salient, nor even perceptually present, contiguously with the emotional expression. This is an important step toward the adult understanding of desire. Early Understanding of Simple Intentions Another important aspect of the adult understanding of the mind is a distinction between the actual actions someone performs and their intention in performing those actions. Wittgenstein (1953) asked, "What is left over if I subtract the fact that my arm goes up from the fact that I raise my arm?" (p. 161). Answer: Intention. This pithy example shows that intention is not wholly reducible to bodily movement. Intentions are mental states and bodily movements are physical events in the world. The two have an intimate relation because intentions underlie and cause bodily movements. If we know a person's intentions, we (often) can predict her actions, and conversely if we see her actions, we can often reason backwards to what her intentions must have been. Moreover, within the adult framework, only certain types of movements are ascribed to intention. Chairs and boulders move, but their rocking and rolling are not seen as intentional. Most prototypically, human acts are the types of movement patterns that are seen as caused by intentions. Just as the youngest infants do not show evidence of understanding the directedness of mental states, they also show little evidence of understanding this distinction between underlying intentions and visible movements, although they do link their own intentional movements to the intentional movements of others (as in body imitation). When do children begin to differentiate bodily movements from the underlying psychological states that cause them, and when do they begin to understand that only certain types of movements and not others are intentional? To address these questions, it is not enough to show that young infants act intentionally themselves. We want to know when they begin to under
Page 25
stand the intentions of others, and most important, when they begin to differentiate surface actions from underlying intentions. There has been some excellent research on this question using verbal tests on young children just beyond "the dark ages," between 3 and 4 years of age (e.g., Astington & Gopnik, 1991b; Moses, 1993; Shultz, 1980; Shultz, Wells, & Sarda, 1980). Our goal was to use the behavioral reenactment technique to pose the question to preverbal children. In one study, we tested whether 18monthold children could read below the literal surface behavior demonstrated in an adult act (Meltzoff, 1995). The study involved showing infants an unsuccessful act. For example, the adult accidentally under or overshot his target, or he tried to perform a behavior but his hand slipped several times. Thus the goalstate was not achieved (Fig. 2.1). To an adult, it was easy to read the actor's intentions, even though he was not able to fulfill them. The experimental question was whether children interpreted this behavior in purely physical terms or whether they too read through the literal body movements to the underlying goal or intention of the act. The measure of how they interpreted the event was what they chose to reenact, in particular whether they chose to produce the intended act despite the fact that it was never present to the senses. In a sense, the "correct answer" was to not copy the literal movement, but the intended act that remained unfulfilled and therefore invisible. Using this behavioral reenactment paradigm, Meltzoff (1995) tested four groups of 18monthold infants. The demonstrationtarget group saw the adult successfully fulfill his intentions and perform a series of target acts on five different objects. The demonstrationintention group saw accidental failures for five different events. With each object, the adult strove
Fig. 2.1. The displays used in the study of toddlers' understanding of simple intentions. The top panel shows the human's acts. The adult tried to pull apart the dumbbell three times. Each time, his hand slipped off the end of the cube (first to one side, then the other, and then the first side again). The children did not copy this surface behavior. When given their turn with the dumbbell, they wrapped their hands around the cubes and firmly pulled it apart. The bottom panel shows the inanimate device. Results showed that children did not try to pull the dumbbell apart after seeing these motions. They interpreted the human acts differently than the similar motions of the inanimate device. (From Meltzoff, 1995.)
Page 26
to reach the goal but did not successfully carry out his intention. The adult's motor actions were realisticlooking attempts to reach the goal, but he did not verbalize or show facial expressions of frustration at his failures. Two control groups were used. The control1 group simply omitted any adult demonstrations. The control2 group saw the adult perform control actions on the objects for the same length of time as in the two demonstration groups, but the adult showed neither the target acts nor the intention to achieve them. The control acts were carefully designed so as to control for the possibility that spatial proximity of the adult's hands to the target, or proximity of two objects with each other, might "suggest" the target behavior (see Meltzoff, 1995, for details). The results were very clear cut. Infants in the control groups did not tend to produce the target acts spontaneously or by chance. However, infants in the two demonstration groups produced more than 75% of the target acts. They reproduced the targets after observing the adult do so, which is straightforward imitation. The important, new finding is that they also produced the targets in the intention condition. Indeed infants were equally likely to produce the target after seeing the adult "trying" but failing as they were when the target was actually achieved. They seemed to interpret the adult's effortful behavior as going beyond the literal surface behavior itself, and as being about something else, about the unseen but inferred goal of the act (see also Tomasello & Barton, 1994, for work using a slightly different approach and yielding compatible results). Several followup studies have now been completed that confirm and expand the original findings. If infants are picking up the underlying goal or intention of the human act from seeing the failed attempt they should be able to achieve the act using a variety of means. This was tested in a recent study (Meltzoff, 1996b). As before, the adult showed the failed attempt with his hands sliding off the ends. Then he handed the infant a gigantic dumbbell that was too big for the infant's hands. The infants did not even make an attempt to grasp the ends of the dumbbell. They did not appear to be trying to mimic the surface behavior. Instead, they used novel ways to struggle to get the gigantic toy apart. They put one end of the dumbbell on the table and used both hands to pull the other end upwards; or they put their hands inside the toy and pushed outwards, and so on. In short, they use different means than had been demonstrated by the experimenter, but used them toward the same end. Of course, the interesting thing is that they had never seen the end. They inferred the end and then used previously unseen means to get there. This eliminates the possibility that infants in the original study had merely tried to imitate the surface behavior of the adult (hands slipping off the cubes) and had pulled the toy apart by mistake. It is consistent with the hypothesis that
Page 27
infants had inferred the goal of the act, differentiating it from the literal surface behavior that was observed. Another experiment pressed this point further. In this study, infants were shown the standard "failed attempt" display, but they were handed a trick toy. The toy had been glued shut before the study began. When infants picked it up and attempted to pull it apart, their hands slipped off the ends of the cubes. This of course matched the surface behavior of the adult. The question was whether this match to the adults' behavior satisfied them. The results showed it did not. Infants were not satisfied when they matched the surface behavior of the adult; this did not terminate their behavior. They repeatedly grabbed the toy and yanked on it in different ways, and appealed to their mothers and the adult. Ninety percent of the infants looked up at an adult immediately after the infant failed to pull the trick toy apart. They did so with an average latency of less than 2 seconds and accompanied by vocalizations while they stared into the adults' faces (Meltzoff, 1996b). Why were they appealing for help? They had matched the adult's surface behavior, but evidently were striving toward another goal, not the behavior itself. (Of course it did not make a lot of sense to seek the adult's help because he had already failed. If a large adult failed, the infant's failure was perhaps inevitable. This subtlety escaped them, however.) We have begun to explore the aspects of the adult's behavioral envelope that carry the information that an action was a failedattempt and not a success. When we see an adult relaunch the act and vary the means, we interpret the adult as "effortfully trying" to accomplish something beyond what he is doing. We tested whether infants were sensitive to this tryandtryagain aspect of intentional action (Meltzoff, 1996a). In this study, four independent groups of 18montholds saw the adult perform either the failedattempt or the successful target act either one or three times. (Recall that infants in the original Meltzoff, 1995, study saw three failed attempts.) Infants who saw one failed attempt performed poorly. Their behavior dropped to chance levels, significantly lower than infants who saw three failed attempts. On the other hand, the infants who saw the adult perform the successful target behavior only once did as well as if they saw it three times. This establishes that it is not an across theboard sensory limitation of some kind. These results suggest that seeing a person relaunch his behavior several times is an important cue to the purposiveness of the act for infants, just as it is for adults (Heider, 1958). When an adult relaunches his behavior several times, using different but related actions, infants infer that there is a common cause unifying this surface behavior. In short, 18montholds use the whole pattern of behavior to indicate whether the adult is aiming to do what they are doing or something else.
Page 28
What Kind of Entities Are Interpreted as Purposive? The results we have discussed so far suggest that normal infants can pick up the simple intentions of human actors. One interesting question is whether this intentional reading of behavior is specific to people or at least most readily ascribed to them. To begin to examine this, Meltzoff (1995) tested how 18montholds responded to a mechanical device that mimicked the same movements as the actor in the failedattempt condition. An inanimate device was constructed that had poles for arms and mechanical pincers for hands. It did not look human but it could move very similarly to the human (Fig. 2.1, bottom panel). For the test, the pincers "grasped" the dumbbell at the two ends just as the human hands did. One mechanical arm was then moved outwards, just as in the human case, and its pincer slipped off the end of the dumbbell just as the human hand did. The movement patterns of machine and man were closely matched from a purely spatiotemporal description of movements in space. The results showed that infants did not attribute a goal or intention to the movements of the inanimate device. Although they were not frightened by the device and looked at it as long as at the human display, they simply did not see the sequence of actions as implying a goal. Infants were no more (or less) likely to pull apart the toy after seeing the failed attempt of the inanimate device than they did in baseline levels when they saw nothing. Another study pursued this point. In this study, we had the inanimate device succeed. The inanimate device held the dumbbell from the two ends and successfully pulled it apart. After witnessing this display, infants were given the dumbbell. The results showed that they too pulled it apart (Meltzoff, 1996b). It appears that infants can pick up certain information from the inanimate device (they pull it apart after seeing the device do so), but they cannot pick up other information (concerning failed attempts). We believe 18montholds interpret the person's actions within a psychological framework that differentiates between the surface behavior of people and a deeper level involving goals and intentions. When they see a person's hands slip off the ends of the dumbbell, they infer what the adult was "trying" to do (which is different from what he did do). When they see the inanimate device slip off the end of the dumbbell, they see it as mechanical slippage and sliding with no implications for purposiveness.3 3
It is quite likely that displays can be constructed that fool infants, analogously to those that fool adults (is my computer intentional?). We do not know the necessary and sufficient conditions for attribution of intention, but under certain circumstances infants may see purposiveness in the actions of pretend humans (e.g., stuffed dolls or animals) or in dynamic displays that may seem to be ambiguous as to animacy (2D spots that leap and move spontaneously, as in Gergely, Nádasdy, Csibra, & Bíró, 1995). This does not contradict our thesis, but underscores the need for research on boundary conditions. The 3D, clearly inanimate object used by Meltzoff (1995) gives a lower boundary (infants fail) and real people give an upper boundary (infants succeed). There is a lot of room in between.
Page 29
Human Acts Versus Mechanical Motions On the basis of these findings, it is useful to introduce a distinction that is discussed later in the chapter. We wish to distinguish between construing the behaviors of others in purely physical versus psychological terms. To help keep this distinction clear, we call the former movements or motions and the latter human acts. The behavior of another person can be described using either physics or psychology. We can say, "The president's hand contacted the cup; the cup fell over," or "The president was trying to pick up the cup." Strict behaviorists (and some lawyers) stick to the former description precisely because they eschew appealing to invisible psychological states. Evidently, infants are not behaviorists (not to mention lawyers). They do not construe the behavior of others simply as, "Hold the dumbbell and then remove one hand quickly," but rather construe it as an effort at pulling. Moreover, the new work shows that infants have a differentiation in the kinds of attributions they make to people versus things. By 18 months of age children have already adopted a fundamental aspect of a folk psychology: Persons are understood within a framework involving goals and intentions. Human acts are seen as dripping with purposiveness and are mapped onto one's own like acts. Using a Person's Emotional Reactions to Understand Their Intentions In the adult framework, it makes sense that the same external event may cause one person to become happy and another sad. This is because emotions do not map directly onto outcomes, but are mediated by the person's desires. Using a variety of verbal tests, it has been shown that young children understand quite a bit about the linkage between desires, actions, and emotional reactions. For example, they know that fulfilled desires lead to happiness and a cessation of searching, whereas unfulfilled ones lead to sadness and a continuance of search for the desired object (Hadwin & Perner, 1991; Harris, 1989; Stein & Levine, 1989; Wellman & Banerjee, 1991; Yuill, 1984). There is clear evidence for this sort of understanding in 3 to 4yearolds and some evidence that children as young as 34 months (Wellman & Woolley, 1990) can correctly predict whether a protagonist in a story will be happy or sad, depending on the match between her desires and the outcome. The behavioral reenactment procedure provides a nonverbal way of beginning to explore children's understanding of the links between desire/intention action emotional reaction. In one study, children ranging from 18 to 36 months old were shown an adult performing an action and the adult's emotional reaction to her action was systematically
Page 30
manipulated (Meltzoff, 1996b). After the action was completed, the adult reacted in one of two ways. For half the children, the adult reacted with happy/satisfied facial expressions and exclaimed, "Yeah! There!" and for the other half she reacted with unhappy/dissatisfied facial expressions and exclaimed, "Uhoh! Oh dear!" The important point is that the adult's physical actions were identical in both cases. What differed was the adult's reaction to the event she caused. The question was whether the children's inference about the adult's desire/intention changed as a function of her emotional reactions. Using the behavioral reenactment procedure, three groups of infants aged 18, 24, and 36 months old were randomly assigned to the see the happy/satisfied and unhappy/dissatisfied reaction. The events in themselves were carefully designed on the basis of pilot studies to be ambiguous in themselves. For example, the adult put a toy unstably on top of a 12inchhigh shelf and the toy toppled off making a banging sound as it hit the table. The usual response after the toy fell was to look at his face to clarify the ambiguous event. Immediately after the object fell, the adult gave one of the reactions, happy/satisfied or sad/dissatisfied. The objects were then given to the child. This proved to be a very motivating task, and children leapt at the opportunity to play with the objects. The results revealed an interesting developmental change. The behavior of the 18 and 24montholds did not systematically vary as a function of the adult's emotional reactions. The 36montholds showed a highly significant and orderly response. They carefully put the toy stably on the shelf (which is not what they saw) in the case that the adult seemed dissatisfied by the outcome. Conversely, they exaggeratedly knocked the toy off the shelf if the adult had shown the happy/satisfied reaction. We conclude that in an ambiguous situation children use the adult's emotional reactions to clarify the meaning of the adult's behavior. On this interpretation, children by 36 months but not 18 to 24montholds can reason backwards from an emotional reaction to what the adult was striving to do. It is interesting that this age estimate fits well with that obtained by verbal methods (Wellman & Woolley, 1990). A modest additional piece of information provided by the behavioral reenactment procedure is that children are not simply presented with a multiplechoice verbal response ("will he be happy or sad"). The children are surrounded with the clutter of realworld activity and have to create for themselves the desired end state. The adult never put the toy stably on the shelf and the children who tuck it firmly up there, far from the edge, are imagining and creating the result of what the adult "had in mind" but never achieved. Another interesting point is that the children were forced to reason backwards from emotional reaction to the unseen desire or intention, not forward, as in many of the verbal story scenarios, from the desire and events to the
Page 31
predicted emotion. The results indicate that at least by 36 months old, children know that the adult may desire/intend to do something different from what they do and that the person's emotional reactions after the event are a clue to the underlying desire/intention of the person. It makes some developmental sense that children can first read the goal of the adult's act (by 18 months) and then later can detect regularities in how they themselves and others emotionally react to the successes and failures of goaldirected actions. Presumably, this is the database that allows them (at 36 months) to give meaning to messy, naturally occurring events such as an actor who reacts either positively or negatively to toppling toys. Understanding Differences between One's Own Desires and Those of Others The work discussed so far has focused on one important aspect of our adult understanding of intention, the ability to "read through" surface actions to determine the intentions of the person who performed them. One important aspect of this ability is that the child seems to go beyond the immediate action itself. Children also seem to take into account other aspects of the situation, for example the actor's attendant emotional reactions, to determine intentions. This research assumes that the child identifies their own intentions with the intentions of adults. In fact, one reason both the behavioral reenactment and social referencing paradigms are so effective is that the children so readily "take on" the intentions, desires, and attitudes of adults, even when those intentions are not their own initially. However, an important aspect of our adult theory of mind is the fact that we can differentiate between our own intentions and desires and the intentions and desires of others. We understand that our desires and intentions may differ from and even be in conflict with, the desires of those around us. In some ways these differences among desires parallel the differences in belief that are tested in falsebelief paradigms. This understanding of differences in desires emerges considerably earlier than the understanding of differences in belief (e.g., Flavell et al., 1990; Gopnik & Slaughter, 1991). As in the case of intention, there is a considerable body of work suggesting that children understand this aspect of desire by the time they are about 3 years old (Astington & Gopnik, 1991a; Bartsch and Wellman, 1995), but the origins of this understanding have been lost in "the dark ages." Repacholi and Gopnik (1997) devised a nonverbal method to explore young children's understanding of differences in desires. The method, like that in the previous study of intention and the earlier study on "di
Page 32
rectedness,'' capitalized on young children's ability to detect emotions. Here, however, we asked a rather different question about emotions. In this work, 14 and 18 monthold infants were presented with a plate of raw broccoli and a plate of goldfish crackers. Infants consistently prefer the crackers. The experimenter indicated her preference for one object or another by tasting it and producing a particular emotional expression (disgust or pleasure). Infants were randomly assigned such that half of them saw the adult apparently like the goldfish crackers (the child's own preference) and half saw her apparently like the broccoli (the child's nonpreferred food). In the crucial test condition, the experimenter then reached her hand out to the infant midway between the foods and asked the infant to give her an (unspecified) food. The results showed that 18monthold infants consistently gave the adult the object for which she had expressed a preference, even when the preference differed from their own desire. They gave her broccoli when she had previously expressed a desire for the broccoli, and crackers when she expressed a desire for crackers. This is a developmental achievement inasmuch as 14montholds did not do this. Instead, they always gave the experimenter crackers, their own preference, regardless of the experimenter's expressed desires. This work suggests that even very young children, 18montholds, may have a nonegocentric understanding of the differences between their own mental states and those of others in some cases. This early understanding of desire, like the understanding of intention, goes beyond the simple cues of action or emotion themselves. It confirms that 18monthold children are not limited to the immediate evidence of the experimenter's perceptually present action, or their own present feelings, in determining the adult's desires. Instead, they take into account an earlier, and superficially quite different, piece of evidence about the experimenter's underlying mental state, namely her emotional expression. Like the earlier understanding of social referencing, it also shows that young children understand the directedness of mental states. They understand that disgust and pleasure were directed at different objects. But this understanding also goes beyond understanding the directedness and valence of mental states and the fact that they underlie, but are not identified with, actions. It shows that 18montholds, although not 14montholds, understand differences between their own desires and those of others. By 18 months, a complexity of folk psychology has dawned on children. They have come to understand that people not only have mental states, just as they do, but these mental states may sometimes not be the same as their own. Other people are like me but do not necessarily have my likes. The children no longer live in a mental Garden of Eden without conflict in which everyone is conceived of as sharing the same desires.
Page 33
Developing Theories of Intention for Ourselves and in Our Children We first summarize some of the developmental changes that have been described in children's understanding of mind between 0 and 3 years of age. Next we sketch three mechanisms of change that may induce these developments. Conceptual Change Between Birth and 3 Years of Age Newborns are not as sophisticated as 18montholds, no less 3yearolds. Newborns can imitate actions themselves. They can link the actions they see to internal feelings. However, they do not initially appreciate that those feelings may be directed toward objects. Imitation of objectdirected acts arises in the second half year of life (Meltzoff, 1988; Meltzoff & Moore, 1998). Similarly, young infants can imitate what the adult actually does, but they can't imitate what the adult intends to do but fails to achieve. One of us has tried in vain to get neonates to imitate intended actions (e.g., an adult straining to produce a tongue protrusion or mouth opening). They simply do not seem to read through the actions. Hence a developmental change from the youngest infants imitating what we do to older ones imitating what we meant to do. Younger infants also initially do not appreciate fully the differences between their own states and those of others. In fact, we have seen that they seem to begin by assuming that the two will be similar. By 18 months, they will have learned about all these characteristics of the mind. So there is a rich initial state but also profound developmental change. The fact that newborns have any way at all of interpreting others as equivalent to the self provides a foundation for the development of the notion of persons that will eventually include desire and intention. As we have seen, the 18monthold's abilities are quite different from the newborn's abilities. At the same time, they are also quite different from young 3yearold abilities. Eighteenmontholds differentiate between intentional and unintentional actions, between their own desires and those of others, and understand that desires and intentions are directed at objects. However, there is no evidence that they conceptualize desires or intentions as mental states that exist in the mind prior to and independent from any action at all, although there is evidence that older children do so. Nor, in fact, is there any evidence that they differentiate between desires, intentions, and emotional attitudes toward objects, although we have been using those adult terms differentially in this chapter. Reconstructing the child's world view in terms of adult language is always difficult. One idea we find helpful is Searle's (1983) notion of
Page 34
"intentioninaction." Searle suggests that even adults often understand action as informed and shaped by desires and intentions, even if they do not think a separate mental state preceded that action. When I sit down in the morning to drink my tea and go over my plans for the coming day, I may be said to have formulated the intention and the desire to drive to the office at 9:45 a.m. When in the course of that drive, I swerve to avoid the construction pothole, I do so intentionally and have the desire to do so, but I could not be said to have had those desires and intentions before I started out that morning or even right before I swerved. When, in the course of the drive, my mind is so full of the intention and desire to make a good argument at the 10:00 a.m. meeting that I actually drive into the next construction pothole, the damage to my axle is neither intentional nor desired. We suggest that the 18monthold's conception of desire/intention is akin to my conception of the intention involved in swerving to avoid the pothole. It is not identified with or reducible to any bodily movement in particular, but it is assumed to accompany actions. Mechanisms of Development We are suggesting then that there are important developmental changes in the child's conception of the mind, in particular in their understanding of desire and intention, between birth and 3 years (for a more complete account see Gopnik & Meltzoff, 1997). What mechanisms are responsible for those changes? We suggest that three mechanisms may be particularly important: imitation and interpretation of human acts, spontaneous experimentation, and integration of the evidence. Imitation and Interpretation of Human Acts as "Like Me" We believe that infants start off with a wellstocked toolkit for developing a folk psychology. We can be precise about how the initial state enables later development. In our view, the "human act" may be the most elementary parsing of the world for social cognition. Human acts are especially relevant to infants because they look like the infant feels himself to be and because they are things infants can intend. When a human act is shown to a newborn baby, it may provide a primordial "Aha" experience: "Something interpretable! That (seen) event is like this (felt) event." It is not simply the attractive eyes and lips of the adults that are special for infants but the way the body moves and its relation to the self. The fact that infants can recreate the act allows them to imbue it with special meaning. Thus, we propose that the initial parsing infants impose on the world is not any one of the "usual suspects" found in textbooks and commonly discussed at the biennial meetings of the Society for Research on Child Development. It is not the GelmanSpelke distinction between "animate
Page 35
versus inanimate" (because armadillos are only of passing interest to infants). It is not the PremackLeslieMandler distinction between "selfpropelled versus moved byaseen force" (because swinging clock pendulums and falling leaves are not viewed by infants as special). It is not even the philosopher's distinction between "people (as adults know them) versus things." We believe that the primordial distinction may be something closer to "human acts versus other events'' (see Meltzoff & Moore, 1995, 1997, for more detailed arguments along this line). Infants' construing certain movements in the environment in terms of human acts that can be imitated has cascading developmental effects: (a) The world of material objects is then divisible into those that perform human acts (people) and those that do not (things); and (b) having made the division in the external world, new meanings are possible. I can imitate others, and those entities out there can generatively imitate me. Persons are special entities, the only entities in the world with whom I can share behavioral states. Furthermore, the recognition that others share your states lays the foundation for making further progress toward ascribing psychological properties to these people. This may come about in part because the infant detects regularities in their own behaviors and feelings. When they are acting in a "tryandtryagain" manner they are striving to reach a goal that is not achieved. When they feel happy, they regularly produce a smiling face. There are regularities between the way they act and the way they feel. This would have no interpersonal significance if infants could not recognize that others are acting just like them. But as we have discovered from studies of imitation, infants can readily act like others and recognize when others are acting like them. This recognition of sharing behavioral states is crucial because it allows a foothold for infants attributing like mental states to others. We envision a threestep developmental sequence: (a) When I perform that bodily act I have such and such a phenomenal experience, (b) I recognize that others perform the same type of bodily acts as me, (c) the other is sharing my behavioral state; ergo, perhaps the other is having the same phenomenal experience. (For further analysis of this developmental sequence, see Meltzoff, 1990; Meltzoff & Moore, 1995, 1997.) On this view, imitation and the crossmodal representation of human acts provide a kick start for getting folk psychological thinking off the ground. Without it, people would not be seen as psychological entities, 'just like me." The "likemeness" of others is the essential foundation for all later social cognition—from the attribution of mental states, to empathy, to moral judgments. Experimentation We have suggested elsewhere that young children use psychological devices that bear an interesting similarity to the cognitive devices that are involved in theory change in science (Gopnik, 1996, 1998;
Page 36
Gopnik & Meltzoff, 1997). One such device is active experimentation. We have proposed that normal infants have an early, and perhaps innate, drive to actively experiment with the world in a way that will increase their understanding of it and that this experimentation plays an important role in development. There is evidence for a period of experimentation accompanying the new understanding of desire and intention that emerges in "the dark ages." We can demonstrate this discovery in the laboratory, but it is also dramatically apparent in ordinary life. Parents know it as the "terrible twos." (The dark ages, in development as in history, are dark in both senses, in mystery and in retchedness.) What makes the terrible twos so terrible is not that the babies do things you do not want them to do, but that they do things because you do not want them to. Twoyearolds are deliberately perverse, what the British call bloodyminded. The 2yearold does not even look at the forbidden computer keyboard as you type your grant proposal. Instead his hand goes out as he looks, steadily, gravely, and with great deliberation, at you. Why do they torture us by seeking to play with the very things in the world we desire that they do not touch—the computer, the lamp cord, the lipstick, the power tools? This perverse behavior may turn out to be quite rational. Consider that 2yearolds are only just in the course of discovering that people may have different desires. The broccoli experiment shows that children first start to realize that there are differences between their own desires and those of others when they are about 18 months old. The terrible twos seem to involve a systematic exploration of that idea, almost a kind of experimental research program (see Gopnik & Meltzoff, 1997; Repacholi &.Gopnik, 1997, for further arguments). Toddlers are systematically testing the dimensions on which their desires and the desires of others may be in conflict. The grave look is directed at you because you and your reactions, rather than the forbidden power tools, are the really interesting things. If the child is a budding psychologist, parents are the laboratory rats. Moreover, the experimentation is striking because it actually conflicts with the child's apparent interest in domestic peace. The Fall has come: The young child now understands that their own desires and those of others are not only not the same but that they often conflict. They are forced from the mental Garden of Eden. Integration of Evidence A further common factor in both conceptual changes in childhood and theory change in science is the importance of relevant evidence. Children have extensive evidence about the nature of human action, intention, and desire. There are two sets of experimental findings suggesting that evidence about desires and intentions may induce
Page 37
developmental change and propel toddlers toward a fuller understanding of mind. First, there is a consistent finding in the literature that younger siblings do better on theoryofmind tasks than only or older children (Jenkins & Astington, 1996; Perner, Ruffman, & Leekam, 1994). The most likely explanation for this effect is that siblings provide children with rich evidence about the mind and particularly about differences in minds. Remember that much of what children learn involves the differences between their own minds and the minds of others. They largely take the similarities for granted; in fact, the assumption that we are like other people seems to be part of a basic foundation for understanding mind found in infancy. Parents, and perhaps especially some mothers, tend to minimize the distance between their own mental states and those of the babies. They look for commonality and understanding instead of difference, and their lessons are largely lessons about congruence. Siblings may provide a necessary counterweight. They are much more likely to emphasize differences between what they want and the baby wants, or to witheringly contrast their highly superior 4yearold knowledgeand the baby's pitiful 2yearold ignorance. Other data showing a positive correlation between early parentchild talk about feelings and later performance on theoryofmind tasks (Dunn, Brown, & Beardsall, 1991; Dunn, Brown, Slomkowski, Tesla, & Youngblade, 1991) might also be interpreted from this viewpoint, assuming that the frequencyofconversations measures capture increased talk about conflicts in desires, not solely increased congruence talk, a reasonable assumption given the functions of everyday conversations (e.g., Bruner, 1990). The second set of data comes from work in which we explicitly tried to induce changes in children's understanding of the mind by giving them evidence (Slaughter & Gopnik, 1996). The focus in this study was on developments in falsebelief, appearancereality, and source understanding between 3 and 4 years of age. The theory guiding the research was similar to that of this chapter—that an understanding of mental states such as beliefs emerge from a prior understanding of the mind that includes concepts such as desire, intention, perception, and so on. One prediction from this viewpoint is that giving children experience with understanding these earlier states should induce them to develop more quickly and to acquire an understanding of belieflike mental states even though the latter were not part of the training. Such an acceleration study was conducted and the results confirmed that providing children with evidence relevant to desire and perception significantly increased their understanding of belieflike mental states, including passing the falsebelief exam. This acceleration study strongly suggests that experience with reasoning about desires and perceptions are developmental precursors to understanding of belief.
Page 38
Conclusions The challenge now is to articulate a theory about early development that takes seriously the richness of the initial state of infants' understanding of persons, as revealed in modern infancy research, and still embraces developmental change (Gopnik & Meltzoff, 1994, 1997; Meltzoff & Moore, 1997, 1998). We have argued for what we call a "startingstate nativism" that doesn't reduce to Fodor's (1987) "modularity or final state nativism" in which the outcomes are fixed to begin with and simply mature with age. On the contrary, we propose that development proceeds by a constant process of revision, like the process of theory change in science. Infants never face the empiricist dilemma of inducing the existence of the mind from the raw data of behavior. On the other hand, they also are not trapped by the constraints of a single, biologically fixed construal of other people. The analogy to science enables us to suggest that the sophisticated mental life of the 4yearold could emerge without being preprogrammed in the mind of the newborn. The folk psychological framework of Western adults is neither innate nor maturationally determined; it is fashioned by the child largely to account for his experiences with other persons. Children, like the adults who study them, start off with certain powerful assumptions, they experiment, and the theory they construct is deeply influenced by the evidence they receive. Our understanding of children and their understanding of us is not fixed by nature but cobbled together as we interact with each other. References Astington, J. W., & Gopnik, A. (1991a). Developing understanding of desire and intention. In A. Whiten (Ed.), Natural theories of mind: Evolution, development and simulation of everyday mindreading (pp. 3950). Oxford: Basil Blackwell. Astington, J. W., & Gopnik, A. (1991b). Theoretical explanations of children's understanding of the mind. British Journal of Developmental Psychology, 9, 731. Astington, J. W., Harris, P. L., & Olson, D. R. (1988). Developing theories of mind. New York: Cambridge University Press. Baldwin, D. A. (1993a). Early referential understanding: Infants' ability to recognize referential acts for what they are. Developmental Psychology, 29, 832843. Baldwin, D. A. (1993b). Infants' ability to consult the speaker for clues to word reference. Journal of Child Language, 20, 395418. Baldwin, D. A., Markman, E. M., Bill, B., Desjardins, R. N., Irwin, J. M., & Tidball, G. (1996). Infants' reliance on a social criterion for establishing wordobject relations. Child Development, 67, 31353153. Baldwin, D. A., & Moses, L. J. (1994). Early understanding of referential intent and attentional focus: Evidence from language and emotion. In C. Lewis & P. Mitchell (Eds.), Children \ early understanding of mind: Origins and development (pp. 133156). Hillsdale, NJ: Lawrence Erlbaum Associates.
Page 39
Bartsch, K., & Wellman, H. M. (1995). Children talk about the mind New York: Oxford University Press. Bruner, J. S. (1990). Acts of meaning. Cambridge, MA: Harvard University Press. Damon, W. (1998). Handbook of child psychology (5th ed.). New York: Wiley. Darwin, C. (1872). The expression of the emotions in man and animals. Chicago: University of Chicago Press. Dunn, J., Brown, J., & Beardsall, L. (1991). Family talk about feeling states and children's later understanding of others' emotions. Developmental Psychology, 27, 448455. Dunn, J., Brown, J., Slomkowski, C., Tesla, C., & Youngblade, L. (1991). Young children's understanding of other people's feelings and beliefs: Individual differences and their antecedents. Child Development, 62, 13521366. Ekman, P. (1980). The face of man: Expressions of universal emotions in a New Guinea village. New York: Garland STMP Press. Flavell, J. H., Flavell, E. R., Green, F. L., & Moses, L.J. (1990).Young children's understanding of fact beliefs versus value beliefs. Child Development, 61, 915 928. Flavell, J. H., & Miller, P. H. (1998). Social cognition. In W. Damon (Series Ed.) & D. Kuhn & R. Siegler (Eds.), Handbook of child psychology: Vol. 2. Cognition, perception, and language (pp. 851898). New York: Wiley. Fodor, J. A. (1987). Psychosemantics: The problem of meaning in the philosophy of mind. Cambridge, MA: MIT Press. Gergely, G., Nádasdy, Z., Csibra, G., & Biró, S. (1995). Taking the intentional stance at 12 months of age. Cognition, 56, 165193. Gopnik, A. (1996). The scientist as child. Philosophy of Science, 63, 485514. Gopnik, A. (1998). Explanation as orgasm. Minds & Machines, 8, 101118. Gopnik, A., & Meltzoff, A. N. (1994). Minds, bodies, and persons: Young children's understanding of the self and others as reflected in imitation and theory of mind research. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Eds.), Selfawareness in animals and human: Developmental perspectives (pp. 166186). New York: Cambridge University Press. Gopnik, A., & Meltzoff, A. N. (1997). Words, thoughts, and theories. Cambridge, MA: MIT Press. Gopnik, A., & Slaughter, V. (1991). Young children's understanding of changes in their mental states. Child Development, 62, 98110. Gopnik, A., Slaughter, V., & Meltzoff, A. N. (1994). Changing your views: How understanding visual perception can lead to a new theory of the mind. In C. Lewis & P. Mitchell (Eds.), Children's early understanding of mind: Origins and development (pp. 157181). Hillsdale, NJ: Lawrence Erlbaum Associates. Gopnik, A., & Wellman, H. M. (1994). The theory theory. In L. A. Hirschfeld & S. A. Gelman (Eds.), Mapping the mind: Domain specificity in cognition and culture (pp. 257293). New York: Cambridge University Press. Hadwin, J., & Perner, J. (1991). Pleased and surprised: Children's cognitive theory of emotion. British Journal of Developmental Psychology, 9, 215234. Harris, P. L. (1989). Children and emotion: The development of psychological understanding. Oxford: Basil Blackwell. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Jenkins, J. M., & Astington, J. W. (1996). Cognitive factors and family structure associated with theory of mind development in young children. Developmental Psychology, 32, 7078. Kagan, J. (1981). The second year. Cambridge, MA: Harvard University Press. Meltzoff, A. N. (1988). Infant imitation and memory: Ninemontholds in immediate and deferred tests. Child Development, 59, 217225. Meltzoff, A. N. (1990). Foundations for developing a concept of self: The role of imitation in relating self to other and the value of social mirroring, social modeling, and self practice in infancy. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood (pp. 139164). Chicago: University of Chicago Press.
Page 40
Meltzoff, A. N. (1995). Understanding the intentions of others: Reenactment of intended acts by 18monthold children. Developmental Psychology, 31, 838850. Meltzoff, A. N. (1996a, April). The role of imitation in developing a theory of mind. In P. D. Zelazo (Chair), Intentionality: An interdisciplinary approach. Symposium conducted at the meeting of the International Conference on Infant Studies, Providence, RI. Meltzoff, A. N. (1996b, August). Understanding intentions in infancy. In A. Leslie (Chair), Children's theory of mind. Symposium conducted at the XXVI International Congress of Psychology, Montreal, Canada. Meltzoff, A. N., & Gopnik, A. (1993). The role of imitation in understanding persons and developing a theory of mind. In S. BaronCohen, H. TagerFlusberg, & D.J. Cohen (Eds.), Understanding other minds: Perspectives from autism (pp. 335366). New York: Oxford University Press. Meltzoff, A. N., & Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 7578. Meltzoff, A. N., & Moore, M. K. (1994). Imitation, memory, and the representation of persons. Infant Behavior and Development, 17, 8399. Meltzoff, A. N., & Moore, M. K. (1995). Infants' understanding of people and things: From body imitation to folk psychology. In J. Bermúdez, A. J. Marcel, & N. Eilan (Eds.), The body and the self (pp. 4369). Cambridge, MA: MIT Press. Meltzoff, A. N., & Moore, M. K. (1997). Explaining facial imitation: A theoretical model. Early Development and Parenting, 6, 179192. Meltzoff, A. N., & Moore, M. K. (1998). Object representation, identity, and the paradox of early permanence: Steps toward a new framework. Infant Behavior and Development, 21, 201235. Moses, L. J. (1993). Young children's understanding of belief constraints on intention. Cognitive Development, 8, 125. Mumme, D. L., Won, D., & Fernald, A. (1994, June). Do one year old infants show referent specific responding to emotional signals? Poster presented at the meeting of the International Conference on Infant Studies, Paris, France. O'Neill, D. K. (1996). Twoyearold children's sensitivity to a parent's knowledge state when making requests. Child Development, 67, 659677. Perner, J. (1991). Understanding the representational mind. Cambridge, MA: MIT Press. Perner, J., Ruffman, T., & Leekam, S. R. (1994). Theory of mind is contagious: You catch it from your sibs. Child Development, 65, 12281238. Repacholi, B. M. (1998). Infants' use of attentional cues to identify the referent of another person's emotional expression. Developmental Psychology, 34, 1017 1025. Repacholi, B. M., & Gopnik, A. (1997). Early reasoning about desires: Evidence from 14 and 18month olds. Developmental Psychology, 33, 1221. Searle, J. R. (1983). Intentionality: An essay in the philosophy of mind New York: Cambridge University Press. Shultz, T. R. (1980). Development of the concept of intention. In W. A. Collins (Ed.), The Minnesota Symposium on Child Psychology (Vol. 13, pp. 131164). Hillsdale, NJ: Lawrence Erlbaum Associates. Shultz, T. R., Wells, D., & Sarda, M. (1980). Development of the ability to distinguish intended actions from mistakes, reflexes, and passive movements. British Journal of Social and Clinical Psychology, 19, 301310. Slaughter, V., & Gopnik, A. (1996). Conceptual coherence in the child's theory of mind: Training children to understand belief. Child Development, 67, 29672988. Stein, N. L., & Levine, L. J. (1989). The causal organization of emotional knowledge: A developmental study. Cognition and Emotion, 3, 343378. Taylor, M. (1996). A theory of mind perspective on social cognitive development. In R. Gelman & T. Au (Eds.), Handbook of perception and cognition: Vol. 13. Perceptual and cognitive development (pp. 283329). New York: Academic Press.
Page 41
Tomasello, M. (1995).Joint attention as social cognition. In C. Moore & P. . Dunham (Eds.), Joint attention: Its origins and role in development (pp. 103130). Hillsdale, NJ: Lawrence Erlbaum Associates. Tomasello, M., & Barton, M. E. (1994). Learning words in nonostensive contexts. Developmental Psychology, 30, 639650. Tomasello, M., Strosberg, R., & Akhtar, N. (1996). Eighteenmonthold children learn words in nonostensive contexts. Journal of Child Language, 23, 157176. Wellman, H. M. (1990). The child's theory of mind. Cambridge, MA: MIT Press. Wellman, H. M. (1993). Early understanding of mind: The normal case. In S. BaronCohen, H. TagerFlusberg, & D.J. Cohen (Eds.), Understanding other minds: Perspectives from autism (pp. 1039). New York: Oxford University Press. Wellman, H. M., & Banerjee, M. (1991). Mind and emotion: Children's understanding of the emotional consequences of beliefs and desires. British Journal of Developmental Psychology, 9, 191214. Wellman, H. M., & Woolley, J. D. (1990). From simple desires to ordinary beliefs: The early development of everyday psychology. Cognition, 35, 245275. Wittgenstein, L. (1953). Philosophical investigations (G. E. M. Anscombe, Trans.). New York: Macmillan. Yuill, N. (1984). Young children's coordination of motive and outcome in judgments of satisfaction and morality. British Journal of Developmental Psychology, 2, 7381.
Page 43
Chapter 3— Intentional Relations and Triadic Interactions Chris Moore Dalhousie University The second half of infancy, from about 9 months to about 18 months of age, sees the advent of a particularly important set of social skills. During this period, infants start to engage with other persons in various interactions that involve a third object or event. An essential component of these "triadic" (Bakeman & Adamson, 1984) interactions is that, for the first time, infants become able to make use of another person's intentional action to find out about the world. In an essential way, these interactions are developmentally homologous with all later cultural activities involving a shared focus of attention, including, for example, the writing and reading of this volume. It is this ability to engage in triadic interactions that defines us as human because human knowledge is knowledge that is shared with others (e.g., Tomasello, Kruger, & Ratner, 1993). Even the earliest triadic interactions of late infancy clearly involve a sensitivity to the intentions of others. However, the nature of this sensitivity is still at issue. This chapter is about what the infant of about 12 months knows of the intentions of others. I do not say much more about the development of intentional action itself. As a number of authors have noted (e.g., Frye, 1981, 1991; Russell, 1996; Tomasello, 1995), it is probably no coincidence that infants show coordinated meansends sequences of action at about the time they start to respond appropriately to others' objectoriented action. In fact, part of my story is that the infant's own intentional orientations to objects and states of affairs are a necessary component of the first understanding of intention. However, my main
Page 44
goal is to elucidate how the infant at this age understands others' intentional action. To be clear from the outset, I should say something about what I mean by intentional. I am concerned with intentions in the broader sense of intentionality, and not limited to purposeful behavior. In this sense, intentional covers all psychological orientations to an object or state of affairs. Indeed, of special interest are those psychological orientations for which the object is real but displaced in space, of which the paradigm case in my opinion is looking at something. We have argued in recent years (e.g., Moore, in press; Moore & Corkum, 1994) that infants of about 12 months do not understand that others can look at (or attend to) things. Our claim has been that 12montholds interpret others primarily in terms of their actions and not in terms of their intentions (see also Povinelli, chap. 11, this volume). After giving my talk at the Toronto conference on which this book is based, I was accused of asserting that infants are behaviorists. That was not the first time the charge had been laid, nor has it been the last. I think this accusation occurs frequently because in many people's minds, either one allows that infants (or anyone) interpret action as intentional and mediated by mental states or one disallows the involvement of such mental properties in infants' interpretations and thereby consigns infants to being limited to behavioral interpretation. Because I have taken a skeptical stance in the evaluation of infants' social understanding, my account tends to get assimilated to the latter view. In fact, my position is that both viewing infants as having an understanding of others' intentions and viewing infants as little behaviorists are problematic and equally so in that they impute an individualistic form of interpretation to the infant. In contrast, my view is that at 12 months the infant's understanding of intention is grounded in interactive contexts, not in individuals. In the second half of the chapter, I consider some recent work from my laboratory using the novelty preference technique to examine infants' understanding of others. Although by no means conclusive, this work is, I believe, consistent with the theoretical account developed in the first part of the chapter. Intentional Relations In earlier work (Barresi & Moore, 1996), Barresi and I argued that the development of the understanding of intentionality is best seen as a progressive coordination of what we call intentional relations. The latter term is meant to denote the idea that all intentional activity is activity by an agent in relation to something. The "something" can be a real object, directly perceived, or it can be a mental object, which includes propositions.
Page 45
Thus, the idea of intentional relations is supposed to capture not only the most complex propositional attitudes that are usually the domain of philosophers but also the simplest intentional acts. Significantly, our goal was to unite the philosophical notion of the mental as intentional with the commonsense notion of purposeful action. Although there is clearly a difference between mental activity and behavioral activity, casting both as forms of intentional relation recognizes their common roots. As adults, we have an individualistic conception of intentional relations. Our commonsense or folk psychology is individualistic in the sense that agents, commonly persons, are taken to be individual centers of intentional activity capable of taking up a range of intentional relations with real and represented objects. Importantly, one of those agents is the self. The fact that the self is recognized to be an individual agent like others around us is selfevident and yet arguably one of the most significant features of commonsense psychology. When it comes to agenthood, self and other are understood to be essentially equivalent. We understand that other people have conscious experience of the real and imaginative worlds in the same way that we do and at the same time, we understand that we are objective agentive entities of the same kind as the others we see around us. In our earlier work, Barresi and I (Barresi & Moore, 1996; Moore & Barresi, 1993) used the understanding of selfother equivalence as a starting point for our theoretical investigation into the origins of intentional understanding. Understanding this equivalence, we argued, must present a problem for a naive observer of intentional activity because the information available to an observer about their own intentional activity is qualitatively very different from the information available to that observer about anyone else's intentional activity. To elaborate, we argued that information that is available to an observer about another agent's intentional activity emphasizes the agent's action. Correspondingly, the object is less salient, in particular when the intentional activity is in relation either to a mental object or to a real object that is displaced in space. The observer cannot know directly the object of another's intentional relation (the intentional object) and consequently it will have to be inferred in some way. In contrast, the information that is immediately available to an observer about his or her own intentional activity is primarily focused on the intentional object. The agent (self) and any action involved are less salient. In this way, thirdperson information and firstperson information tend to emphasize different components of the intentional relation. Thirdperson information emphasizes the agent and his or her activity. Firstperson information tends to emphasize the phenomenal characteristics of the object and one's attitude towards it. The epistemic problem for a naive observer of intentional activity is how to recognize that both first and thirdperson information about that
Page 46
activity correspond to the same kind of thing, namely an intentional relation between an agent and an object. We (Barresi & Moore, 1996) argued that there must be some way for the young infant to construct the equivalence of intentional relations of self and other from the available input and suggested that interactive contexts in which intentional relations between self and other are shared would provide the right kind of experience from which to construct such an understanding. Gopnik and Meltzoff (1994) also pointed to selfother equivalence as an epistemic problem for understanding intentionality, although their proposed solution to its development is somewhat different from ours (see Moore, 1996). Our account indicates that there is a period in development during which infants can participate in shared intentional relations without being able to attribute an intentional relation to an individual agent, either self or other. The suggestion is that this period comes in the second half of infancy from about 9 months of age. As noted earlier, it is at about this age that infants start to show a variety of behaviors that allow them to interact with another person in relation to some third object. I adopt the term triadic (cf. Bakeman & Adamson, 1984) to refer to these interactions because it captures in a relatively theoretically free manner the important issue— that there are now three things involved: infant, adult, and third object. Commonly cited examples of triadic interactions are joint visual attention or gaze following, social referencing, and prelinguistic communicative acts such as pointing (see Moore & Corkum, 1994). Although some have interpreted these phenomena as revealing an early form of theory of mind (e.g., Bretherton, 1991; Wellman, 1993), I think it is fair to say that such an interpretation is not warranted on the basis of the phenomena alone. Elsewhere (Moore & Corkum, 1994), we have provided a skeptical account of infants' joint visual attention, whereby infants will turn to look in the same direction as an interactive partner. We argued that such gaze following does not require the attribution to young infants of an understanding of attention. Instead, infants may understand that others' gaze behavior is predictive of where interesting sights will occur. The empirical literature on the topic is fully consistent with such a view (see Moore, in press, for a review). Baldwin and Moses (1996) similarly considered the phenomenon of social referencing, or checking of a caregiver's emotional expression in times of uncertainty, and concluded that its origins may lie in the attachment system and not in intentional understanding (see also Povinelli, chap. 11, this volume). If infants at the end of the first year do not understand others in terms of intentional relations, then how do they represent others? We have argued that, at this age, infants process others' activity in terms of thirdperson information, or information about the spatiotemporal properties of behavior. They are able to engage in triadic interactions because the third
Page 47
person information provided by the other's intentional activity directed at objects has become linked in familiar and arousing interactive structures with the infant's own intentional objectoriented activity. The thirdperson information provided by the observation of the other's activity is linked to the corresponding firstperson information provided by the infant's coordinated response. To take again the example of gaze following, infants turn in the same direction as someone else, not because they understand that the other is looking at something, but because they are able to participate in an interactive structure whereby the observation of another's head turn leads to an expectation that an interesting sight will appear to the side, and to a consequent head turn to look to the side. This account of gaze following fits the data on gaze following. For example, Butterworth and others (e.g., Butterworth & Cochran, 1980; Morissette, Ricard, & Gouin Decarie, 1995) showed that when infants first start to follow gaze to objects outside of the immediate visual field, they do not seek out the target of the other's gaze but only turn as far as necessary to see something of interest. At about the same time as they start to follow gaze to objects outside the visual field, infants also start to direct others' attention towards novel or interesting objects using gestures such as pointing (see Carpenter, Nagell, & Tomasello, 1998). Again, however, this behavior can be understood in terms of the thirdperson information provided by the activity of the other being linked within familiar and arousing interactive structures with the firstperson information provided by the infant's own intentional orientation to the novel or interesting event. Here, the infant's own observation of something of interest leads him or her to perform a gesture with the expectation that the adult will then provide some interesting feedback (Moore, 1998; Moore & Corkum, 1994). In such interactive episodes, the infant is presented with corresponding first and thirdperson information about the intentional relation, and it is the combination of these two forms of information that provides the first form of intentional understanding in that an agent is thereby connected to an intentional object. The critical point for present purposes, however, is that the intentional relation is not understood to be a property of either the self or the interactive partner. Why This Is Not a Behavioral Account of Infant Social Understanding It is important to point out here that the claim that infants process others' activity in terms of thirdperson information does not commit one to a behaviorist view of infant social understanding. As I noted earlier, it is sometimes thought that either infants understand the intentional nature
Page 48
of others' activity or they are little behaviorists. If infants only process information about others in thirdperson terms, then is that not a behavioral view? To see why not, consider what it means to have a behavioral understanding of human activity. To construe human action behaviorally means that the observer understands there to be a relation between the action of the other person and the objective state of affairs. This relation is deemed to be a causal one, with the objective state of affairs causing the person's behavior, and one that is unmediated by that person's mental states. Important, the objective state of affairs is understood to be affecting the other person independently of its effects on the observer. Thus, behavioral understanding resembles adult intentional understanding in that it is individualistic. It differs from intentional understanding in that it eschews mentalistic variables. Our claim is that infants do not understand that the other is in a relation, intentional or otherwise, to the objective state of affairs. The other's activity is not seen to be related to an object in either a behavioral or an intentional manner. The activity of another agent is linked to an object or state of affairs only through the connection with the infant's own firstperson intentional orientation to that object or state of affairs. Thus, the other is not understood to have intentional orientations to objects independently of the infant's own orientations to those objects. Equally, whereas the infant has intentional orientations to objects, the self is not understood to be an intentional agent. It is the thirdperson information provided by the observation of the other's activity combined with the firstperson information provided from the infant's own intentional orientation to the object or state of affairs that constitutes the first form of understanding of intentional relations. One might say that the 12 monthold recognizes intentional relations but only to the extent that she can share such a relation with an interactive partner. For the 1yearold, intentional relations exist in the interaction and are not a property of, or descriptive of, individuals. This, then, is a constructivist developmental story in that the interactive structures that contain the essential components for intentional understanding form the basis on which more mature, individualistic forms of understanding develop. Our claim has been that the individualistic form of understanding depends on the capacity for imagination or secondorder representation, which develops around the middle of the second year. Only when the child can imagine the component of the intentional relation that is not given in immediate experience (i.e., firstperson component for others and thirdperson component for self) will the notion of intentional relations that are descriptive of persons be possible. Imagination means that when the child observes another person engaged in intentional activity, she can imagine the corresponding firstperson information com
Page 49
ponent and attribute it to the other. At the same time, when the child herself acts intentionally, she can imagine herself from a thirdperson perspective. In this way a uniform understanding of self and other becomes possible because the child can represent both as agents acting in relation to intentional objects. Notice that although this account has some similarities to the simulation view of social understanding (e.g., Harris, 1992), it differs in the sense that not only is self used as a model for others, but also others are used as a model for self. Empirical Approaches So, one part of our claim is that 1yearolds represent others' activity in terms of thirdperson information, not in terms of intentional relations, but how can one examine infants' understanding of others' intentional activity? Some years ago, Corkum and I (Moore & Corkum, 1994) proposed that the novelty preference paradigm could be used to shed some light on these issues. We were particularly intrigued by the use of the novelty preference method by Cohen and Oakes (1993; Oakes & Cohen, 1990) to study infants' representation of causal relations. Like intentional relations, causal relations involve a relation between an agent and an object, albeit one that is based on the physical rather than psychological properties of the agent. In an ingenious series of experiments, Cohen and Oakes showed that 10 monthold infants are sensitive to the characteristics of the object serving as the agent and the type of action, but not to the characteristics of the object serving as the recipient of the causal event. Our idea (see also Woodward, 1995) was that if 12montholds understand others mainly in terms of thirdperson information, then, when observing an agent engaged in an intentional relation with an object, they should pay more attention to the agent and his or her activity and less attention to the overall relation including the object, especially for intentional activity operating at a distance. Take the example of someone pointing at an object. Figure 3.1a represents the case of a person pointing at one of two objects, indicating attention or interest in that object. From a purely thirdperson perspective, the relevant information comes from the spatial properties of the agent's action, namely the direction of the pointing gesture. Imagine showing an infant this scene repeatedly until habituation occurs and then substituting either of Fig. 3.1b or Fig. 3.1c in which the position of the two objects has been switched. Both of these figures show the same agent pointing at one of the two objects. In Fig. 3.1b, the agent continues to point to the same location but now it is towards the other object. In Fig. 3.1c, the agent continues to point to the same object as before, thereby changing the spatial properties of the pointing gesture. According to the logic of novelty
Page 50
Fig. 3.1. Schematic representation of habituation and dishabituation pointing events.
preference, if the infant processed the original scene in terms of the intentional relation, ''Agent is attending to Object A," then there should be greater recovery of looking to Fig. 3.1b. In contrast, if the infant processed the original scene in terms of the agent's activity alone, then there should be greater recovery of looking to Fig. 3.1c. We have now conducted a series of three experiments using this logic (see Table 3.1) and although the results did not come out exactly as anticipated, I believe they do shed some light on the issues discussed. TABLE 3.1 Overview of the Three Experiments Using Novelty Preference Paradigm to Examine Infants' Representation of Intentional Action
Experiment
Format
Habituation Stimulus
Dishabituation Stimuli
1
Live
Adult 1 points at Toy A
Adult 1 points at Toy A Adult 1 points at Toy B
2
Live
Adult 1 points at Toy A
Adult 2 points at Toy A Adult 2 points at Toy B
3
Video
Adult 1 points at Toy A
Adult 1 points at Toy A Adult 1 points at Toy B
Page 51
General Method In all three experiments, participants ranged in age from 12 to 14 months. Forty infants participated in each of the first two experiments with average ages of 13;1 (i.e., 13 months, 1 day) and 13;6, respectively. Thirtysix infants of average age 13;7 participated in Experiment 3. At each age, infants were divided into two equal presentation order groups. The experiments were conducted in a laboratory room enclosed in plain brown curtains. The infant sat in a high chair pulled up to a table, 60 cm high, 90 cm long, and 60 cm wide. In the middle of the table was a slot into which a cloth screen (60 cm wide × 40 cm high) was folded. This screen was attached via fishing line and a pulley system to a handle that the experimenter, who was present in the room with the infant, manipulated. The screen thereby could be raised and lowered by the experimenter at the appropriate points in the session. The stimuli were presented on the other side of the screen from the infant so that when raised, the screen obscured the infant's view of the stimulus materials. The stimulus included an adult who produced the pointing gesture and two toys placed at just over arm's length in front of the adult. The toys, a small stuffed dog and a similarly sized plastic car, were placed on small platforms so that they were about 4 inches off the table. A novelty preference paradigm was employed. In the first phase, infants were presented repeatedly with the habituation stimulus. Infant looking was coded by an observer in an adjacent room from a monitor. A different observer was used for each experiment. The observer could communicate with the experimenter through the use of wireless intercom. Looking time was measured using MacXHab 1.4 software, ©1995 (J. Pinto, Stanford University, 1995) running on a Macintosh Powermac 7100 computer. Whenever the infant was judged to be looking at the display, the observer depressed a computer key and released it when the infant looked away. For the purposes of scoring infant looking, the display included the adult and the two toys. If the key was released for 2 seconds, the computer signaled that the trial was over and the observer relayed this information to the experimenter. The experimenter then raised the screen for 2 seconds and lowered it again to begin the next trial. The computer calculated looking time automatically and signaled when the habituation criterion (a 50% decrement in mean looking time over three trials compared to the first three trials) was reached. At this point, the observer signaled to the experimenter that the criterion had been reached. The experimenter raised the screen and switched to the dishabituation stimuli. There were 6 dishabituation trials comprising three different toy trials and three same toy trials in alternating sequence. Order of the trials was counterbalanced. For different toy trials the experimenter pointed at the second toy, which
Page 52
was now in the same location as the first toy had been during habituation trials. In this way, the pointing gesture was exactly the same as before, but it was now towards a different toy. For the same toy trials the experimenter pointed at the same toy as before, but now it was in a different location requiring a change in the spatial organization of the pointing gesture. Dishabituation trials proceeded and were coded in the same manner as habituation trials. Reliability In order to check reliability, all subjects' looking behavior was coded from videotape by a new coder who was naive to the experimental hypotheses. Reliability coding was done in the same way as live coding. Rather than generate an arbitrary reliability statistic for coding acceptability, we wanted to determine if the patterns of results were affected by who did the coding. To this end, for each experiment, we analyzed all of the data for the live coder and for the reliability coder in overall repeated measures analyses of variance (ANOVAs), treating coder as a withinsubjects variable. These analyses all showed a main effect of coder with the looking times being longer for the reliability coder. In addition, for Experiment 2 only, the coder variable interacted significantly with the other variables. Further analyses showed that the effects of the other variables were stronger for the reliability coder than for the live coder. For Experiments 1 and 3, the coder variable did not show interaction effects with any of the other variables. Thus, although the reliability coder tended to record longer looking times in general, the other effects were robust independently of who did the coding or whether it was done live or from replay. In what follows, only the results from the live coder are reported. Experiment 1 In the first experiment, we essentially conducted the experiment sketched in the introduction to this section. Infants sat opposite an adult experimenter who stood in full view and facing the infant. On each trial the experimenter said, "Oh look," and produced a point with arm extended at one of two toys. Throughout the habituation phase the experimenter pointed at the same object on the same side. The particular object used in the habituation phase as well as the side on which it was located was counterbalanced across subjects. Once the habituation criterion had been reached, the experimenter switched the locations of the toys and the experiment moved into the dishabituation phase. Over six trials, the experimenter alternated the location of her point, thereby presenting different toy and same toy trials in alternating sequence. For half of the infants, the
Page 53
experimenter started the dishabituation phase with a different toy trial and for the other half, this phase was started with a same toy trial. To determine whether infants looked longer at the dishabituation stimuli, a twoway repeated measures ANOVA was carried out on the results. For each presentation order, the mean looking time of the last three habituation trials was compared with the looking times for the first different toy trial and for the first same toy trial. This analysis yielded only a significant effect of trial, F(2, 76) = 28.62, p
>
EuropeanAmerican 5.25 .89 52 3.31 .49 30 1.28 .31 30 1.86 .12 29
>
>