Advances in Child Development and Behavior Volume 30

ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR

Volume 30

Contributors to This Volume Karen Adolph

David C. Geary

Lorraine E. Bahrick

Erica E. Kleinknecht

Patricia J. Bauer

Robert Lickliter

L. Beckwith

Lonna M. Murphy

Thomas J. Berndt

Amanda Sheffield Morris

Melissa M. Burch

A. Rozga

Nancy Eisenberg

M. Sigman

ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR

edited by Robert V. Kail Department of Psychological Sciences Purdue University West Lafayette, Indiana

Volume 30

ACADEMIC PRESS An imprint of Elsevier Science Amsterdam Boston London New York Oxford Paris San Diego San Francisco Singapore Sydney Tokyo

This book is printed on acid-free paper. Q

Copyright 9 2002, Elsevier Science (USA). All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the Publisher. The appearance of the code at the bottom of the first page of a chapter in this book indicates the Publisher's consent that copies of the chapter may be made for personal or internal use of specific clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc. (www.copyright.com), for copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale. Copy fees for pre-2002 chapters are as shown on the title pages. If no fee code appears on the title page, the copy fee is the same as for current chapters. 0065-2407/2002 $35.00 Explicit permission from Academic Press is not required to reproduce a maximum of two figures or tables from an Academic Press chapter in another scientific or research publication provided that the material has not been credited to another source and that full credit to the Academic Press chapter is given.

Academic Press An imprint of Elsevier Science. 525 B Street, Suite 1900, San Diego, California 92101-4495, USA http://www.academicpress.com

Academic Press 84 Theobalds Road, London WC1X 8RR, UK http://www.academicpress.com International Standard Book Number: 0-12-009730-3 PRINTED IN THE UNITED STATES OF AMERICA 02 03 04 05 06 07 MM 9 8 7 6 5 4

3

2

1

Contents Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Learning to Keep Balance KAREN A D O L P H I. II. III. IV.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Learning to Keep Balance: Sway Model of Balance Control . . . . . . . . . . . . . . . . . . . . . . . . . Flexibility and Specificity of Motor Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . How Development May Constrain Motor Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 3 9 27 36

Sexual Selection and Human Life History D AVID C. GEARY I. II. III. IV. V. VI.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Natural Selection and Life History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sexual Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Life History and Sexual Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Human Developmental Sex Differences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

41 42 51 56 63 87 89

Developments in Early Recall Memory: Normative Trends and Individual Differences PATRICIA J. BAUER, MELISSA M. BURCH, AND ERICA E. K L E I N K N E C H T I. II. III. IV. V. VI.

Initiating the Study of Early Recall Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characterizing Recall Memory in the First Two Years of Life . . . . . . . . . . . . . . . . . . . . . . . . Individual Differences in Long-Term Recall: Children's Gender, Children's Language Proficiency, and Variability in Initial Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Individual Differences in Long-Term Recall: Children's Temperament Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Children's Temperament and Mothers' Language as Interacting Sources of Individual Differences in Long-Term Recall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

105 110 115 121 130 138 146

Contents

vi

Intersensory Redundancy Guides Early Perceptual and Cognitive Development LORRAINE E. B A H R I C K AND R O B E R T LICKLITER I. II. III. IV. V. VI.

Introduction: Historical Overview and Perspectives on Perceptual Development .... Amodal Relations and the Multimodal Nature of Early Experience . . . . . . . . . . . . . . . . . . Unimodal-Multimodal Dichotomy in Developmental Research . . . . . . . . . . . . . . . . . . . . . . Neural and Behavioral Evidence for Intersensory Interactions . . . . . . . . . . . . . . . . . . . . . . . Intersensory Redundancy Hypothesis: Toward an Integrated Theory of Perceptual Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Directions for Future Study of Perceptual Development . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

153 156 158 159 163 178 181

Children's Emotion-Related Regulation NANCY EISENBERG AND A M A N D A SHEFFIELD MORRIS I. II. III. IV. V. VI. VII. VIII.

Definition of Emotion-Related Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A Brief Review of Views of Emotion Regulation in Theories of Emotion . . . . . . . . . . . Conceptual Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Age-Related Trends in Emotion-Relevant Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurement of Emotion Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relation of Emotion-Relevant Regulation to Quality of Social Functioning . . . . . . . . . Relations of Dispositional Resiliency to Effortful and Reactive Control and Socioemotional Functioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

190 191 194 198 204 210 215 219 220

Maternal Sensitivity and Attachment in Atypical Groups L. BECKWITH, A. ROZGA, AND M. SIGMAN I. II. III. IV. V. VI.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Organization of Attachment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Maternal Sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Role of Child Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Atypical Groups of Mothers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

231 233 236 242 255 260 263

Influences of Friends and Friendships: Myths, Truths,

and Research Recommendations

THOMAS J. BERNDT AND L O N N A M. M U R P H Y I. II. III.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Influences of Friends' Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Influences of Friendship Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

275 277 294

Contents

IV. V.

Influences of Friends' Characteristics in Friendships Differing in Quality . . . . . . . . . . . Conclusions and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii

299 302 307

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

311

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

329

Contents of Previous Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

339

This Page Intentionally Left Blank

Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.

KAREN ADOLPH

Department of Psychology, New York University, New York, New York 10003 (1) LORRAINE E. BAHRICK

Department of Psychology, Florida International University, Miami, Florida 33199 (153) PATRICIA J. BAUER

Institute of Child Development, University of Minnesota, Minneapolis, Minnesota 55455 (103) L. BECKWITH

Department of Pediatrics, University of California at Los Angeles, Los Angeles, California 90024 (231) THOMAS J. BERNDT

Department of Psychological Sciences, Purdue University, WestLafayette, Indiana 47907 (275) MELISSA M. BURCH

Institute of Child Development, University of Minnesota, Minneapolis, Minnesota 55455 (103) NANCY EISENBERG

Department of Psychology, Arizona State University, Tempe,Arizona 85287 (189) DAVID C. GEARY

Department of Psychological Sciences, University of Missouri at Columbia, Columbia, Missouri 65211 (41) ERICA E. KLEINKNECHT

Institute of Child Development, University of Minnesota, Minneapolis, Minnesota 55455 (103) ROBERT LICKLITER

Department of Psychology, Florida International University, Miami, Florida 33199 (153) LONNA M. MURPHY

Department of Psychological Sciences, Purdue University, WestLafayette, Indiana 47907 (275)

x

Contributors

AMANDA SHEFFIELD MORRIS

Department of Psychology, University of New Orleans, New Orleans, Louisiana 70148 (189) A. ROZGA

Department of Psychiatry, University of California at Los Angeles, Los Angeles, California 90024 (231) M. SIGMAN

Department of Psychiatry, University of California at Los Angeles, Los Angeles, California 90024 (231)

Preface The amount of research and theoretical discussion in the field of child development and behavior is so vast that researchers, instructors, and students are confronted with a formidable task in keeping abreast of new developments within their areas of specialization through the use of primary sources as well as being knowledgeable in areas peripheral to their primary focus of interest. Moreover, journal space is often simply too limited to permit publication of more speculative kinds of analyses that might spark expanded interest in a problem area or stimulate new modes of attack on a problem. The serial publicationAdvances in Child Development and Behavior is intended to ease the burden by providing scholarly technical articles serving as reference material and by providing a place for publication of scholarly speculation. In these documented critical reviews, recent advances in the field are summarized and integrated, complexities are exposed, and fresh viewpoints are offered. These reviews should be useful not only to the expert in the area but also to the general reader. The series is not intended to reflect the development of new fads, and no attempt is made to organize each volume around a particular theme or topic. Manuscripts are solicited from investigators conducting programmatic work on problems of current and significant interest. The editors often encourage the preparation of critical syntheses dealing intensively with topics of relatively narrow scope but of considerable potential interest to the scientific community. Contributors are encouraged to criticize, integrate, and stimulate, but always within a framework of high scholarship. Although appearance in the volumes is ordinarily by invitation, unsolicited manuscripts will be accepted for review. All papers--whether invited or submitted--receive careful editorial scrutiny. Invited papers are automatically accepted for publication in principle, but usually require revision before final acceptance. Submitted papers receive the same treatment, except they are not automatically accepted for publication even in principle and may be rejected. I acknowledge with gratitude the aid of my home institution, Purdue University, which generously provided time and facilities for the preparation of this volume. I also thank David F. Bjorklund for his editorial assistance. Robert V. Kail

xi

This Page Intentionally Left Blank

LEARNING

TO KEEP BALANCE

Karen E.

Adolph

DEPARTMENT OF PSYCHOLOGY NEW YORK UNIVERSITY NEW YORK, NY 10003

I. INTRODUCTION A. THE IMPORTANCE OF B A L A N C E B. OVERVIEW II. LEARNING TO KEEP BALANCE: SWAY M O D E L OF B A L A N C E C O N T R O L A. REGION OF PERMISSIBLE P O S T U R A L SWAY B. L O C A L VARIABILITY AFFECTS THE SIZE OF THE SWAY REGION C. D E V E L O P M E N T A L CHANGES AFFECT DEFINING PARAMETERS III. FLEXIBILITY AND SPECIFICITY OF M O T O R L E A R N I N G A. SITTING AND C R A W L I N G AT THE EDGE OF GAPS B. C R A W L I N G AND WALKING DOWN SLOPES C. LEARNING TO DETECT THREATS TO BALANCE: THE VISUAL CLIFF AND OTHER FALLING TASKS IV. HOW D E V E L O P M E N T M A Y CONSTRAIN M O T O R LEARNING A. CONTENT OF EVERYDAY EXPERIENCE B. ENSURING FLEXIBILITY AND SPECIFICITY C. CONCLUSION REFERENCES

I. I n t r o d u c t i o n A. THE I M P O R T A N C E OF B A L A N C E

Motor development is a delicate balancing act. Newborn infants are slaves to the pull of gravity, but by 2 months babies can lift their heads from the crib mattress, balance their heads between their shoulders, and turn to look at an interesting event. By 6 months, they can balance in a sitting position anchored to the floor with their outstretched legs, lean forward to retrieve a fallen toy, and use their hands to play with objects or clap. By 8 months, infants can balance on hands and knees, crawl

ADVANCESIN CHILDDEVELOPMENT AND BEHAVIOR,VOL.30

Copyright 2002, Elsevier Science (USA). All rights reserved.

0065-2407/02 $35.00

2

Karen E. Adolph

across the living room floor, steer around furniture, and clamber over objects in their path. By 10 months they can pull themselves to a wobbly stand and walk sideways hanging onto the coffee table or couch for support. By the end of their first year, infants can balance on two feet; walk independently across the room; crouch down to peer under the table; stretch upward to pull books off the shelves; stop, start, and turn comers; and modify their step length and walking speed. These homely postural accomplishments are the stuff of motor development. Students' developmental textbooks are graced with a requisite chart of motor milestones. Parents' home videos and photo albums highlight the postural milestones of their infants' first year. In fact, the pioneering researchers of the 1930s and 1940s were so captivated by infants' dramatic transformation from worm to person and by the seeming regularity with which the metamorphoses occurred that the field of motor development was founded on normative descriptions of the ages and stages that characterize the various postural milestones. Although modem researchers no longer focus on cataloging infants' postural milestones, they agree that the most basic motor control problem is maintaining balance (Reed, 1982). Balance is not only important for relatively stationary positions such as sitting and standing, but also it provides the necessary stable base to support movements of the head, torso, or limbs. Everyone who has broken a rib or thrown out his or her back has experienced the centrality of posture for controlling movements--lifting an arm from the bed to grasp a glass of water becomes a painful lesson about the muscles involved in stabilizing the torso, and lifting a leg to put pants on or to climb stairs is nearly impossible. In the laboratory, researchers have shown that participants' abdominal muscles and back muscles fire prior to a reaching movement with the arms, indicating that postural stabilization is a primary part of the motor plan (Hofsten, 1993). Stabilizing the body and keeping it in balance are prerequisites for adaptive control of movement. B. OVERVIEW Two stories are woven throughout this chapter. The first story concerns the central issue in motor control of maintaining balance, but I approach the problem as a psychologist rather than as a biomechanist. ! argue that a complete understanding of balance control involves much more than a description of muscles, torques, and lever arms. Such understanding involves the basic psychological functions of perceiving the layout, discovering one's own capabilities, and forming plans to accomplish one's goals (Gibson & Pick, 2000; Gibson, 1958, 1979). Most important, I provide evidence that infants must learn to keep their bodies in balance and that this most basic and practical kind of knowledge represents an important and sophisticated psychological achievement. Learning to avoid falling down requires immense psychological flexibility for coping with moment to moment variations in infants' bodies, environments, and goals. In addition, I incorporate these arguments about balance control into a new account of a classic phenomenon--infants'

Learning to Keep Balance

3

avoidance on a visual cliff and provide a unifying framework for understanding a myriad of other tasks in which babies must avoid falling down. The second story concerns a central issue in developmental psychology--the relation between learning and development. I describe how infants' learning about balance control is nested in the context of ongoing developmental changes in their bodies, skills, and environments. Moreover, I argue that peripheral developmental changes may constrain the course of learning to ensure that infants acquire the requisite behavioral flexibility and specificity that they need to respond adaptively under continually changing local conditions. The chapter is divided into three sections. In the first section, I present a sway model of balance control proposed by me and my colleagues (Adolph, 2000; Adolph & Avolio, 2000; Adolph & Eppler, 1998, 2002, Adolph, Eppler, Marin, Weise, & Clearfield, 2000). According to our model, learning about balance control underlies adaptive responses to a cliff or any other risky ground surface or task that poses potential threats to balance. In the second section, I describe several experiments that provide evidence for the sway model. The weight of the evidence rests on the flexibility and specificity of infants' motor learning. On the sway model, infants' responses should be flexible enough to cope with changes in their bodies, skills, and task constraints but specific to each postural milestone in development. The final section addresses a central theme of the chaptermthe relation between learning and development. I will conclude with some suggestions for how developmental changes can help to ensure that infants' learning has just the right amount of flexibility and specificity to promote adaptive responding in a variable world.

II. Learning to Keep Balance: Sway Model of Balance Control A. REGION OF PERMISSIBLE POSTURAL SWAY

Keeping balance appears so effortless in adults' everyday actions and so difficult in babies' attempts to perform the same kinds of skills that a casual observer might think that adults have solved the problem of balance control but that babies have not. In fact, we never outgrow the problem of keeping balance. The process of imminent balance loss and recovery is continual and cyclical. Figure 1A shows a simple physical model of balance control in quiet, upright stance. The body is represented by a mass, M, along an inverted pendulum with the pivot point at the ankles. The dashed lines represent a three-dimensional cone-shaped region in which the body gently sways and teeters over its base of support. The angular displacement of the body as it sways back and forth is represented by tO. (The swaying motions are exaggerated in the figure to illustrate the point more clearly). To keep themselves from falling, infants must maintain their bodies within the cone-shaped region, termed the "cone of reversibility" (McCollum & Leen,

4

Karen E. Adolph

Fig. 1. Physical model of balance control in upright stance on (A) flat ground, (B) with increased mass and displaced center of mass, and (C) on a sloping surface. M is Mass; tO is angular distance body can sway without falling. Dashed lines represent the region of permissible postural sway.

1989) or the "region of permissible postural sway" (Adolph, 2000; Adolph & Eppler, 1998; Riccio, 1993; Riccio & Stoffregen, 1988; Stoffregen & Riccio, 1988). Infants fall if their bodies move outside the sway region without sufficient muscle torque to pull themselves back into position. Thus, a sway in one direction must be counteracted by a compensatory sway in the opposite direction. Such nearly imperceptible oscillations of the body are revealed when adults and infants maintain quiet stance on a force plate or in a "moving room" where the walls move to simulate the optic flow normally generated by postural sway (Bertenthal & Bai, 1989; Lee & Lishman, 1975; Nashner & McCollum, 1985; Stoffregen, Schmuckler, & Gibson, 1987). The size of the sway region, that is, the size of | depends on the amount of muscle torque infants can generate relative to the size of destabilizing torque. (Muscle torque is the amount of angular force generated by the body and destabilizing torque is the amount of force trying to pull the body over.) In the dynamic postures of locomotion, the balance control problem is similar, only now the sway region shifts its position in space as if undergoing a series of deliberate, controlled near-falls. In crawling and walking, for example, the base of support shifts forward in anticipation of catching the body as the moving limb swings forward from step to step (Breniere & Bril, 1988; Goldfield, 1989).

Learning to Keep Balance

5

B. LOCALVARIABILITYAFFECTSTHE SIZE OF THE SWAYREGION What makes balance control a psychological problem, not just a biomechanical one, is that the size of the sway region is continually changing. Figures 1B and 1C show two causes of change in the size of the sway region: changes in functional body dimensions and changes in the ground surface. In Figure 1B, the stick figure is loaded with additional mass and the location of the center of mass is raised along the lever arm. Both factors--increased mass and displacing the center of mass away from the pivot point--increase the size of destabilizing torques and decrease the size of the sway region before falling over. Adults experience such changes in their functional body dimensions when they walk carrying a heavy backpack. In Figure 1C, the stick figure is perched on a slope. Uptight balance is more precarious on a downward slope than on flat ground because the size of the sway region is smaller. The feet are at an angle, decreasing the base of support. Available muscle torque is diminished because leg and trunk muscles must do extra work to keep the body perpendicular to gravity and to curb forward momentum. The vertical distance that the body falls increases during each step. The critical point for my argument is that nearly everything changes the size of the sway region. The ground surface is typically variable in terms of slant, elevation, friction, rigidity, and obstacles. The body's functional dimensions are typically variable due to carrying loads or shifts in the location of the center of mass. Simply raising the arms, bending the knees, or tilting the head forward change the location of the center of mass! In addition, because the parts of the body are mechanically linked, movements themselves affect balance control. In fact, the size of the sway region fluctuates from step to step, with ever,,, irregularity in terrain, with every change in the location of the center of mass due to body movements or shifting a load, and so on. Infants would have to track these changes on a moment to moment basis so that keeping their bodies inside the region of reversible postural sway would require a continual process of calibration in real time. C. DEVELOPMENTALCHANGESAFFECTDEFININGPARAMETERS 1. Development Creates Local Variability The problem of maintaining balance is even more complicated because of ongoing developmental changes. One way that development affects balance is by creating local variability. In particular, developmental changes in infants' body dimensions, skill level, and exposure to environmental properties all affect the size of the sway region. Infants' bodies are ill suited for balance control. Their short bodies sway faster than adults' bodies and require quicker corrections, much like trying to balance a ruler upright on the palm of your hand versus trying to balance a yardstick

6

Karen E. Adolph

(Forssberg & Nashner, 1982; McCollum & Leen, 1989). Moreover, infants' bodies are top-heavy due to their relatively large heads and trunks and short legs (Palmer, 1944). Like a top-heavy bookcase threatening to tip, infants' top-heavy dimensions make them less stable during stance and locomotion (Adolph & Eppler, 2002). In addition to their precarious dimensions, the size and shape of infants' bodies are rapidly changing. The rapid growth and redistribution of body mass during the first 2 years of life rival the dramatic changes during the fetal period and puberty (Palmer, 1944). Moreover, growth changes in infancy are abrupt rather than continuous. Infants literally grow longer and heavier in the course of a day (e.g., 0.5-2.0 cm) separated by plateaus of no change for weeks at a stretch (Lampl, 1993; Lampl, Veldhuis, & Johnson, 1992). Thus, babies often wake up to a differently sized sway region with different constraints on maintaining balance. Changes in infants' body dimensions co-occur with dramatic improvements in their level of motor skill. Many infants begin crawling using idiosyncratic combinations of arm and leg movements while scraping or hopping along with their bellies on the floor. Later, they discover how to crawl on hands and knees or hands and feet. Across all of their crawling styles, infants display rapid improvements in their level of crawling skill. Crawling speed, for example, increases by 720% over the first 20 weeks of crawling (Adolph, Vereijken, & Denny, 1998), and the size of infants' crawling steps increases by 265%. Walking skill proceeds from a Charlie Chaplin (arms supinated upward and toes pointing outward) or Frankenstein gait (stiff legs and outstretched arms) to the bouncing walk of a typical toddler. The size of infants' walking steps increases by 137% over the first 16 weeks of walking and their base of support--the lateral distance between their feet--decreases by 150% (Adolph, Vereijken, & Shrout, 2002). Dramatic changes in skill levels imply changing ability to cope with destabilizing torque. Finally, changing bodies and skill levels co-occur with developmental changes in the environments in which infants travel. Each week after the onset of independent mobility, infants are exposed to new surfaces (grass, sand, cement, shag rug, linoleum, etc.) and obstacles (Chan, Lu, Marin, & Adolph, 1999; Chan, Biancaniello, Adolph, & Marin, 2000). Parents may implicitly take more mature body dimensions and more advanced skill levels into account when they allow their babies freer access to new features of the environment such as stairs, slopes, and playground climbers. As infants acquire more skill, they may be motivated and capable of exploring new aspects of the terrain. Thus, peripheral developmental changes--that is, changes which do not involve the central nervous system--are central to understanding the problem of balance control. Unstable body proportions, sudden growth changes, dramatic improvements in skill level, and exposure to an ever-widening array of environmental opportunities for interaction all result in changes in the size of the sway region. Variability in the size of the sway region, in turn, requires continual monitoring and updating for maintaining balance.

Learning to Keep Balance

7

2. Development Creates New Perception-Action Systems A second way that developmental changes affect balance control is by creating new perception-action systems with new defining control parameters. The series of postural milestones that characterize motor development differ in more than the size of the relevant sway regions; they differ in the location of the sway regions and the key parameters that define them. Figure 2 shows three major postural milestones in motor developmentmsitting, crawling, and walking. In a sitting posture, infants are anchored to the floor by their outstretched legs. In a crawling posture, they move forward while balancing on hands and knees. And, in a walking posture, they move forward while maintaining balance over their feet. Most infants sit at about 6 months of age, crawl at 8 months, and walk at 12 months. However, the ages and order of milestones are not important. The critical point is that infants do not acquire all of these skills at the same time. Sitting, crawling, and walking postures appear staggered over many months of development. The sway model proposes that each postural milestone in development is actually a different balance control system with different relevant control parameters. Sitting, crawling, and walking involve different regions of permissible postural sway for different key pivots around which the body rotates. The hips are typically the key pivot in a sitting posture, the wrists in a crawling posture, and the ankles in an uptight posture. Moreover, each balance control system involves different muscle groups for generating compensatory sway; different vantage points for viewing the ground ahead; different correlations between visual, vestibular, and somatosensory information; and different limbs are available for exploring the ground surface and obstacles. Sitting, crawling, and walking are quite different perception-action systems.

3. Consequencesfor Learning and Development New balance control systems and ongoing developmental changes make particular demands on learning. On the sway model, infants must learn to gauge their region of permissible postural sway. This learning entails two steps: first,

~,,/ ~r \..1

.

Sitting

Crawling

.

.

.

Walking

Fig. 2. Three postural milestones in infant development--sitting, crawling, and walking. Dashed lines represent the region of permissible postural sway around the key pivot in each posture. From "Development of Visually Guided Locomotion" by K. E. Adolph & M. A. Eppler, 1998, Ecological Psychology, 10(3-4) p. 314, Copyright 1998 by Lawrence Erlbaum Associates. Adapted with permission.

8

Karen E. Adolph

identification of control parameters, and second, calibration of those parameters in real time. Each time that infants acquire a new posture in development, they must learn to identify the relevant control parameters for keeping balance: What is the key pivot? What muscle groups control body sway around the pivot? What perceptual information is relevant? Only through experience fighting gravity in a new posture, can babies identify the location of the new sway region and the defining control parameters for the new balance control system. Once infants identify the relevant parameters, they must learn to calibrate the settings of the relevant parameters in real time. Thus, the second step of learning to keep balance is to tweak the parameters of the new balance control system to cope with the continually changing size of the new angle of permissible postural sway. A central prediction of the sway model is that learning should be both highly specific and extremely flexible. The dual processes of identification and calibration correspond to specificity and flexibility of learning, respectively. Transitions to new postures in development create specificity of learning. On the sway model, because developmental changes in posture create new balance control systems, learning to identify the relevant control parameters should be specific to each balance control system in development. At the same time, changes in the size of the sway region demand flexibility of learning. Within each postural milestone in development, learning should be flexible enough to cope with continual changes in body, skill level, and task--the everyday moment to moment changes in the size of the sway region. Thus, as shown in Figure 3, the sway model predicts that infants should display separate learning curves over days of experience with each posture and that there would be no transfer across developmental changes in posture. The adaptiveness of infants' responses in novel predicaments that challenge balance control should be related to the duration of their everyday experience keeping balance in the test posture. The discrepancy between infants' responses in each posture should depend on the relative timing of their onset and the slope of the learning curves. That is, novice sitters, crawlers, and walkers should display a high rate of errors; they

-"-.... Sitting

I -

oi L--,

L,IJ

-

.

.

Crawling

Walking

w

Experience (days) Fig. 3. Three learning curves in infant development corresponding to experience with each postural milestone. Errors represent frequency of infants falling due to misjudgments of their own abilities.

Learning to Keep Balance

9

should misjudge their abilities, lose balance, and fall over. In contrast, experienced sitters, crawlers, and walkers should respond adaptively to novel changes in their bodies, skills, and task constraints. With experience, responses should become increasingly scaled and fine-tuned to the relative degree of risk. In the weeks before learning reaches asymptote, infants should show more adaptive responses in familiar postures compared with unfamiliar ones.

HI. Flexibility and Specificity of Motor Learning In a series of experiments, we tested the predictions of the sway model by comparing infants' responses to risky ground surfaces in more versus less experienced postures (Adolph, 1997, 2000; Adolph, Eppler, & Gibson, 1993a; Leo, Chiu, & Adolph, 2000). We tested infants in novel tasks--stretching over gaps in the surface of support and descending slopes--so that we could observe how they decided in real time whether a surface posed a potential threat to balance. We used a graded array of surfaces--small and large gaps, shallow and steep slopes--presenting varying amounts of risk so that we could measure the adaptiveness of infants' decisions quite precisely. A. SITTINGAND CRAWLINGAT THE EDGE OF GAPS 1. Developmental Design and Rationale

In the "gaps" studies, we tested flexibility and specificity of infants' knowledge about balance control at the edge of a gap in the surface of support (Adolph, 2000). Infants were tested in two postures--sitting and crawling--in the same test session. In both postures, they were perched at the edge of an adjustable gap---a veritable cliff. Their task was the same in both postures. We encouraged them to span the gap by leaning forward while extending an arm toward an attractive toy. We disentangled infants' age from their sitting and crawling experience using an age-matched control design. All the infants were the same age, but their experience varied freely. The experimental design capitalized on the fact that infants display a period of overlap between sitting and crawling milestones. Typically, infants begin sitting many weeks before they begin crawling, meaning that they have much more experience with the sitting posture compared with the crawling posture. Nineteen infants participated. All were 9 months old ( 4-1 week) and all had more experience keeping balance in a sitting posture (M = 3.4 months of sitting experience) compared with a crawling posture (M -- 1.5 months of crawling experience). We reasoned that if experience maintaining balance promotes flexible, adaptive responding in novel tasks that challenge balance, then infants should respond adaptively in the novel gaps task by stretching their arms over safe gaps but refusing to attempt risky ones. If experience is specific to each posture in development, then

10

Karen E. Adolph

infants should show more adaptive responses in their more experienced sitting posture, even when tested on the same apparatus in the same task in the same session.

2. GapsApparatus The most serious threat to balance is a sheer drop-off because there is no floor to support the body. We tested infants' ability to respond adaptively to a drop-off using a modem variant of the classic visual cliff paradigm (Gibson & Walk, 1960; Walk, 1966; Walk & Gibson, 1961). In the classic experimental arrangement, babies are tested on a visual cliff rather than on a real one to ensure their safety. Infants begin on a centerboard dividing a large glass table. On the "deep" side, the apparatus looks like a sheer drop-off because the floor is visible far below the invisible safety glass. On the "shallow" side, the apparatus looks like solid ground because the floor tiles are attached to the underside of the glass. Mothers call to their infants from first one side, then the other. Infants must decide whether to crawl over the apparent drop-off or avoid going. Like the visual cliff, our gaps apparatus (shown in Figure 4) presented a dropoff in the surface of support. However, the gaps apparatus differed from the visual

C

..-.--...-

L....___._..__

Fig. 4. Gaps apparatus and procedure in (A) the sitting condition with movable stick in Experiment 1, (B) the crawling condition in Experiments I and 2, and (C) the sitting condition with movable plaO~orm in Experiment 2. Experimenter (shown)followed alongside infants to ensure their safety. Parents (not shown) stood at the far side of the landing pla(form and encouraged infants' efforts. From "Specificity of Learning: Why Infants Fall Over a Veritable Cliff" by K. E. Adolph, 2000, Psychological Science, 11(4), p. 292, Copyright 2000 by the American Psychological Society. Reprinted with permission.

Learning to Keep Balance

11

cliff in several important ways. First, the visual cliff is actually perfectly safe for locomotion because it is covered in safety glass. Infants determine this in the course of testing and, although they continue to be leery of crossing the glass by patting it, gazing down into the precipice, and clinging to the supporting walls of the apparatus, avoidance responses actually attentuate over repeated trials (Campos, Hiatt, Ramsay, Henderson, & Svejda, 1978; Eppler, Satterwhite, Wendt, & Bruce, 1997; Titzer, 1995). As a result, researchers are limited to one or two trials per infant and infants cannot be tested longitudinally. Findings must be reported in terms of the proportion of infants who avoided the apparent drop-off. In contrast, on the gaps apparatus, the floor had an actual hole with real consequences for balance control. Instead of safety glass, a vigilent experimenter spotted infants to ensure their safety, catching them as they fell. Infants found it aversive to feel themselves falling downward and thus were highly motivated to avoid gaps they perceived to be risky. As a result, we could present infants with 50-100 trials in a single test session. Using a crude but robust psychophysical procedure, we could draw response curves for each infant. A second important difference between the gaps apparatus and the visual cliff concerns the availability of multimodal perceptual information. On the visual cliff, visual and haptic information are in conflict. The visual cliff looks risky but feels safe. The discrepancy in perceptual information makes it difficult to interpret infants' responses. On the gaps apparatus, visual and haptic information are in agreement. Large gaps look risky, feel risky, and are risky. Finally, the visual cliffhas only two settings--the deep and the shallow sides. As with discrepant visual information, this factor also renders it difficult to interpret infants' avoidance responses. The deep side of the visual cliff would be impossibly risky, of course, without the safety glass. But so would many gradations of dropoff that present far less of a compelling visual display. For a crawling infant, a 3-ft precipice is no more risky than a 2-ft precipice: The probability'of crawling successfully is zero in both cases. If avoidance on the deep side reflects a motor decision about whether crossing is feasible, then infants should avoid smaller but equally risky increments as often as the larger more perceptually salient ones. To address this issue, the gaps apparatus was adjustable so that we could present infants with a nearly continuous gradation of gap sizes. The gaps apparatus consisted of a stationary starting platform (106-cm long • 76-cm wide • 86-cm high) and a movable landing platform (157-cm long x 76-cm wide • 86-cm high). In contrast to the visual cliff, we kept the vertical distance to the floor constant and varied the distance across the precipice. By rolling the landing platform along a calibrated track, we could create a gap between the two platforms varying from 0-90 cm in 2-cm increments. The largest gap size was similar to the standard visual cliff along all three dimensions. At the smallest gap distances, balance was trivial. At intermediate gap sizes, infants had to gauge the necessary forces required to span the gap. At the largest gap sizes, the distance exceeded infants' limit of permissible

12

Karen E. Adolph

postural sway. As on the visual cliff, avoidance was the appropriate response to impossibly large gaps. 3. Staircase Procedure to Normalize Risk Level

In the sitting condition (shown in Figure 4A), infants were placed in a sitting position at the edge of the starting platform with their legs dangling into the gap. They were encouraged to lean forward and extend their arms out over the gap to retrieve a toy. The lure was attached by velcro to the end of a stick protruding through a box which rested on the landing platform. An assistant moved the stick forward and backward to create gaps of 0-90 cm between the toy and the edge of the starting platform. Gap distance was varied by moving the stick rather than the landing platform to prevent pinching infants' legs in the gap and to keep them from propping their feet or free hand on the far side of the gap to aid in balance control. In the crawling condition (shown in Figure 4B), infants were placed on their hands and knees in the middle of the starting platform. They were encouraged to crawl to the edge of the gap, lean forward, and extend an arm toward the landing platform as they crawled over the gap. Toys on the landing platform provided the incentive to span the gap. In both conditions, parents waited at the far end of the landing platform and encouraged their babies to retrieve the toys. An experimenter (shown in Figure 4) followed closely alongside infants to ensure their safety but did not provide hands-on support unless infants fell into the gap. Both platforms were covered in soft carpet and the interior of the crevasse was lined with foam as an additional safety precaution. Because infants of the same age have widely varying body dimensions and levels of motor skill, we used a modified psychophysical staircase procedure to normalize the definition of safe and risky gaps relative to each infant's body size and skill in each condition (Adolph, 1995, 1997, 2000; Adolph & Avolio, 2000). The staircase procedure is a classic method in psychophysics for estimating a perceptual threshold using a minimal number of trials (Cornsweet, 1962). When estimating a perceptual threshold, researchers plot the function spanning the increments when the observer is 100% accurate to those when the observer is guessing at 50% chance levels. In this case, we modified the classic staircase procedure to estimate a "motor threshold." The function spanned the increments of gap size where infants spanned the gap successfully on 100% of the trials to those when success rates dropped to 0%. Trials were coded in real time as either successful attempts (safely spanned the gap), failed attempts (fell into the gap), or refusals (avoided the gap). For the purpose of estimating infants' gap thresholds, we treated failed attempts and refusals as equivalent, unsuccessful outcomes. Infants began with an easy baseline gap (10 cm in the sitting condition and 4 cm in the crawling condition). After successful trials, the experimenter increased the gap size by 6 cm. The experimenter repeated trials after one failure or refusal and decreased gap size by 4 cm

Learning to Keep Balance

13

after two consecutive unsuccessful trials. Easy baseline increments were interspersed throughout the protocol to maintain infants' motivation to span the gap. The procedure continued until converging on a gap threshold with a >67% success criterionmthe largest gap that infants spanned successfully at least two out of three times and less than two out of three times at the next 2-, 4-, and 6-cm increments. (Note that gap thresholds were not calculated using curve fitting or linear interpolation as is customary in perceptual psychophysics. Thus, with the >67% success criterion, infants sometimes displayed >67% successes at their estimated gap thresholds and success rates often dropped off sharply near to 0% successes on all larger increments of gap.) Gap thresholds varied widely between infants and conditions. The range in gap thresholds was 20-32 cm for sitting and 2-18 cm for crawling. In the sitting condition, the probability of spanning the gap successfully decreased from .96 at the gap threshold to .07 at gap distances 12 cm larger than threshold. In the crawling condition, the probability of success decreased from .86 at the gap threshold to .06 at gaps 12 cm larger. By definition, infants' gap thresholds marked the difference between safe and increasingly risky gaps. The wide range in gap thresholds highlights the importance of normalizing the definition of risk to individual babies' skill level in each condition to allow comparisons between sitting and crawling postures and between infants with different gap thresholds.

4. Flexibility and Specificity in Sitting and Crawling Postures After identifying the gap threshold, the experimenter tested the accuracy of infants' motor decisions by presenting them with two probe trials at each of several safe and risky gaps (gaps 6 cm smaller than gap threshold, 6 cm larger than gap threshold, 12 cm larger than gap threshold, 18 cm larger than gap threshold, and the largest 90-cm gap). In total, infants received 17-42 trials in the sitting condition and 21-38 trials in the crawling condition. The important psychological question was whether infants could detect potential threats to balance on risky gaps, that is, whether infants' motor decisions were scaled to their gap thresholds. We indexed the accuracy of infants' motor decisions with an "attempt ratio" by calculating the ratio of infants' attempts to span the gap to the total number of trials: (successes + failures)/(successes + failures + refusals). We calculated an attempt ratio for each infant at each increment of gap in each condition. The ratio must be >.67 at the gap threshold by definition but can vary freely from 0 to 1 at all other increments of gap size. Our logic was that infants would have a high attempt ratio on gaps they determined to be safe and a low attempt ratio on gaps they perceived to be risky.

a. Adaptive Responding in Experienced Posture. As predicted by the sway model, infants showed impressive flexibility in their experienced sitting posture by

14

Karen E. Adolph

adapting their responses to the constraints of the novel gaps task (see dashed line in Figure 5A). In their sitting posture, infants attempted to span safe gaps but refused to attempt increasingly risky ones. Attempt ratios were uniformly at zero on the largest 90-cm increment. Apparently, uneventful, everyday experience coping with balance in a sitting position facilitated coping with threats to balance at the edge of the gaps in the laboratory task. None of the infants had experienced a serious fall at home nor were their responses related to experiencing minor falls at home. Moreover, infants rarely fell on risky gaps (only 19% of risky sitting trials). At most risky sitting trials, infants refused to span the gaps outright; they were no more likely to refuse after falling on an earlier trial than after succeeding on an earlier trial.

b. Falling in Unfamiliar Posture. Also as predicted by the sway model, infants showed striking specificity of knowledge in their unfamiliar crawling posture by

i., ,.,,, i._

I-

0.75

.s

0.5

L-

Attempt Ratio (Experiment 1) ~---~,,-~. /r Crawl Sit

',

~

I'T

0 ~. 0.25 0

i

L_

0

B

[]

[]

i

,

|

i

i

i

i

-12

-6

0

+6

+12

+18

90

cm from Gap Boundary

Attempt Ratio (Experiment 2) '~ 0.75-

"-B'

=o o.5~,~

Sitk,, ".Tt~

0

0,, 0.25

Q-

0

. L ~~ m

j.

"~rT-~ i

-12

T

i

i

i

i

",

[]i

-6

0

+6

+12

+18

90

~-

cm from Gap Boundary Fig. 5. Attempt ratios in sitting and crawling conditions in (A) Experiment I and (B) Experiment 2. The data are plotted according to relative degree of risk. The zero point on the x axis represents each infant's gap threshold in each condition. Negative numbers on the x axis denote safe gaps (gaps smaller than the threshold) and positive numbers denote risky gaps (gaps larger than the threshold). Data are also included for the largest 90-cm gap. From "Specificity of Learning: Why Infants Fall Over a Veritable Cliff" by K. E. Adolph, 2000, Psychological Science, 11(4), p. 293, Copyright 2000 by the American Psychological Society. Adapted with permission.

Learning to Keep Balance

15

consistently misjudging constraints on keeping balance. Infants attempted impossibly risky gaps in the crawling condition on trial after trial, requiring rescue by the experimenter to prevent injury (solid line in Figure 5A). At every risky gap increment, infants were more likely to fall into the gap in their unfamiliar crawling posture compared with their more practiced sitting posture. Six of the 19 infants fell into thin air on every trial in the crawling condition, including the largest 90-cm gap. The remaining 13 infants showed some ability to discern threats to balance control by occasionally avoiding the larger gap distances. However, for this more discerning group of infants, failed attempts were significantly higher in the crawling condition than in the sitting condition at the gap threshold and at +6and + 12-cm increments. Because the data in Figure 5 were based on relative amount of risk, this analysis raises the possibility that posture-specific responding was merely a consequence of the fact that infants' gap boundaries for sitting were larger than for crawling, that is, that infants could lean farther forward in the sitting posture than in the crawling posture. If infants' responses were based simply on absolute gap size, such that they always attempted small gaps and refused large ones, then the finding of posture-specific responding would be spurious. However, examination of the attempt ratio at each absolute gap size shows that this was not the case. Every baby showed different attempt ratios at the same absolute gap size in sitting and crawling postures. The six foolhardy babies who fell into every risky gap in the crawling posture clearly responded differently to the same absolute gap size because they avoided the same gaps in the sitting condition. The subset of 13 more discerning infants also showed different attempt ratios to the same absolute gap sizes in sitting and crawling postures. These infants were actually less likely to attempt gaps between 14 and 32 cm in the crawling posture than the sitting posture. But, as shown in Figure 6, in the crawling posture, their attempt ratios still grossly overestimated their ability to span gaps; attempt ratios were significantly higher than the probability of success, even when the probability of success was zero. In contrast, in the sitting posture, infants' attempt ratios closely matched their ability to span gaps; the curves representing attempt ratios and probability of success were superimposed. Because the experimenter rescued infants after they began to fall, a second alternative explanation for the findings is that infants simply learned to rely on the experimenter to catch them. If infants expected the experimenter to catch them, or if they considered falling into the experimenter's arms to be a kind of game, then babies should have responded indiscriminately to all gap sizes in both postural conditions. However, they did not. Effects were the same regardless of condition order, such that infants were no more or less likely to attempt gaps depending on whether they were rescued in the first condition. Infants were most likely to avoid the largest gaps in the crawling condition, although those gaps appeared latest in the session after infants were rescued multiple times on smaller sizes of gaps. And all infants avoided risky gaps in the sitting condition, despite being rescued in the crawling condition.

16

Karen E. Adolph

.

1

Sitting

'~I - 0.75

f ' ~ b Success

= .9 0.5 "o

Attempt

=. 0.25 0

~-

0

. 2

.

. 8

.

.

.

14 20

" 26

32 38

|

44

50

o|

80 90

om

Crawling k.- 0.75 c

.o 0.5

tempt

Ill

=o25 ~ ' a o

o

IX.

0

Succes~'.M

k.

"I,

t~"t:l-- ~--n

i

i

2

a

i

i

14 20

i

26

o

i

i

i

i

32

38

44

50

i

/ F-

i

80 90

em Fig. 6. Attempt ratios of 13 infants who showed some sensitivity to risky gaps in the crawling condition (Experiment 1). The data are plotted according to absolute gap size in both conditions. Solid lines represent the probability of attempting the gap, and dashed lines represent the probability of success. Discrepancies between the curves represent trials where infants fell. From "Specificity of Learning: Why Infants Fall Over a Veritable Cliff" by K. E. Adolph, 2000, Psychological Science, 11(4), p. 293, Copyright 2000 by the American Psychological Society. Adapted with permission.

5. Replication We conducted a replication study to rule out a third alternative explanation: Infants' adaptive responding in the sitting posture may have been due to the fact that the landing platform was always 90-cm away. Possibly, view of the large precipice below the toy lure was responsible for their adaptive avoidance responses in that condition. To test this possibility, we tested a new group of infants using the same procedure but now we varied gap distance by moving the landing platform along a calibrated track (0-90 cm). As before, in the sitting condition, a toy was presented on the end of a stick but now the toy was always aligned to the edge of the landing platform, so that by moving the landing platform, we varied the distance of the lure to the baby (see Figure 4C). We increased the baseline gap size to 20 cm in the sitting condition to avoid pinching infants' legs inside the smallest gaps. The experimenter repeated trials on which infants propped their legs or free hand on the far side of the gap to aid in balance control. In the crawling condition, the toy was placed on the landing platform as before.

Learning to Keep Balance

17

Seventeen 9-month-old infants participated. All were more experienced in the sitting posture (M = 3.4 months) than in the crawling posture (M = 1.8 months). With the new experimental set-up, we replicated all results in the first study. In their more experienced sitting posture, infants closely matched attempts to span the gap to the probability of spanning it successfully. But, in their less familiar crawling posture, infants attempted impossibly risky gaps on repeated trials and fell into the precipice (see Figure 5B). In the crawling posture, 8 of the 17 infants fell at every risky gap distance, including the largest 90-cm gap. B. CRAWLING AND WALKING DOWN SLOPES

1. Developmental Design and Rationale The facilitative effects of experience are not limited to maintaining balance in sitting and crawling postures or to avoiding a sheer drop-off at the edge of a gap. In an intense microgenetic study, we found that experience led to specificity of infants' knowledge about balance from crawling to walking postures when they were confronted with steep and shallow slopes (Adolph, 1997). Babies' task was to decide whether to descend the slopes using their typical crawling and walking method, or alternatively to descend in a sliding position or simply avoid going. As in the gaps experiments, we separated the effects of infants' experience and age, this time using a longitudinal design. We tested 15 infants from their very first week of crawling until several weeks after they began walking. Sessions were scheduled 3 weeks apart. An additional group of 14 infants were observed at matched session times (their first and tenth weeks of crawling and first week of walking) to control for the effects of repeated practice on laboratory slopes. Infants in the experimental group were between 5.5 and 9.9 months old at their first crawling session and between 13.6 and 16.1 months old at their 13th-week walking session. Infants in both the control group and the experimental group did not differ in age at the matched test sessions. Overall, infants contributed 219 test sessions. All the parents agreed to prohibit their infants from climbing up or down slopes (sloping yards, wheelchair ramps, playground slides, etc.) outside the laboratory.

2. Slopes Pla(orm and Procedure We constructed an adjustable slope by connecting three large wooden platforms with piano hinges (see Figure 7). The starting platform was always flat and maintained in a stationary position 71-cm high. The landing platform was always flat but could move up and down powered by a hydraulic pump built from a car jack. Moving the landing platform caused the middle section of the walkway to slope in 2 ~ increments from 0~ ~ The entire walkway was covered in soft carpet to provide a cushion and the sides of the walkway were lined with volleyball nets as a safety precaution.

18

Karen E. Adolph

,,

|

lie

J

Fig. 7. Walkway with adjustable slope. Experimenter (shown)followed alongside infants to ensure their safety. Parents (not shown) stood at the far side of the landing platform and encouraged infants' efforts. From "Learning in the Development of lnfant Locomotion" by K. E. Adolph, 1997, Monographs of the Society for Research in Child Development, 62 (3, Serial No. 251), p. 42, Copyright 1997 by the Society for Research in Child Development, Inc. Reprinted with permission.

Because bodies and skills varied widely between infants and sessions, we normalized the definition of safe and risky slopes relative to each infants' current level of crawling or walking skill. We used the same psychophysical procedure described for the gaps experiments to estimate infants' slope thresholds at each week of crawling and walking. The experimenter began with an easy baseline slope of 4 ~ increased the slant by 6 ~ after successes, decreased the slant by 4 ~ after two consecutive unsuccessful trials, and interspersed baseline trials throughout the protocol to maintain infants' motivation. Sessions ended with two trials at the steepest 36 ~ slope. On average, the total number of trials per session was 20.82. Overall infants in the experimental group experienced 102-369 trials descending slopes. Body size and skill level ranged widely between infants and across sessions, pointing to the importance of normalizing the definition of risk to individual infants' current capabilities. On average, from their first crawling session to their 13th walking session, infants' height increased by 12.5 cm, their weight by 2.9 kg, and their head circumference by 4.3 cm. Crawling speed increased by 32.48 crn/s from their first to their last weeks of crawling, and walking step length increased by 10.3 cm from their first to 13th week of walking. Moreover, changes in infants' body dimensions and locomotor skills translated into changing slope thresholds. Slope thresholds increased over weeks of belly crawling, hands and knees crawling, and walking, and decreased at the transitions from one form of locomotion to another. Figure 8 shows changing slope thresholds for a typical infant across weeks of crawling and walking. Thus, a risky slope one week could be perfectly safe the next week after locomotor skill improved. A safe slope for an experienced crawler could be risky for a novice walker, and so on.

Learning to Keep Balance

19

36 o

24 12

L_

~ i~

R 1

4

7 10 13 16 19

Crawling

1

4

7 10 13 16 19 22 25 28

Walking

Fig. & Change in slope thresholds for a typical infant in the experimental group across weeks of crawling and walking.

3. Flexibility and Specificity in Crawling and Walking Postures As predicted by the sway model, infants displayed both impressive flexibility and striking specificity of knowledge about balance control. To index the accuracy of infants' motor decisions, we calculated each baby's attempt ratio on safe slopes (shallower or equal to the slope threshold) and risky slopes (steeper than the slope threshold). Figure 9 shows infants' average error rates (attempt ratios) on risky slopesmwhen crawlers tried to descend in a crawling position and required rescue by the experimenter and when walkers attempted to descend in a walking position and fell into the experimenter's arms. Because of the nature of the staircase procedure, most of the trials on risky slopes were presented at increments within 8 ~ of the slope threshold. Thus, a low error rate represents an extremely high degree of accuracy.

100

Errors on Risky Slopes

75 I---

50

N

25 0

0

1 4 ~ 1'01'3 1~ 1'9 2'2

~ ,;, § 1'01'3

Weeks Crawling

Weeks Walking -

Experimental

0

Control

Fig. 9. Attempt ratios on risky slopes (errors) across weeks of crawling and walking. Infants in the experimental group were tested every 3 weeks. Infants in the control group were tested in their first and tenth weeks of crawling and in their first week of walking. From "Learning in the Development of lnfant Locomotion '" by K. E. Adolph, 1997, Monographs of the Society for Research in Child Development, 62 (3, Serial No. 251), p. 64, Copyright 1997 by the Society for Research in Child Development, Inc. Adapted with permission.

20

Karen E. Adolph

a. Experience Predicts Adaptive Responding. Figure 9 highlights several important findings. First, the decrease in infants' errors was related to the duration of their crawling and walking experience. In their first weeks of crawling and walking, infants' plunged over the brink of impossibly risky slopes on trial after trial. Over weeks of everyday locomotor experience, errors decreased steadily such that infants used their typical crawling or walking method on safe slopes but slid down or avoided risky ones. Of course, with each week of locomotor experience, infants also became one week older. Several findings point to experience rather than age as the critical determinant of adaptive responding. First, age varied widely at each session (average range = 4.2 months). A maturational component might be expected to show a fighter range in age. Second, infants displayed an increase in errors on risky slopes in their first weeks of walking after 22 or so weeks of crawling, despite their greater age in the walking sessions. And, third, when infants' test age and everyday locomotor experience were pitted directly against each other, experience proved to be the stronger predictor of adaptive responding. Infants' last week of crawling (the week prior to walking onset) provided the strongest test of age versus experience because it was the only session in which both factors varied freely. The duration of infants' crawling experience varied from. 16 to 8.4 months in the experimental group (one child crawled for only a few days prior to walking and others crawled for several months). Age ranged from 9.8 to 14.6 months. Not surprisingly, measures of age and experience were intercorrelated Jr(13) - .61, p < .02]. Older infants tended to have more locomotor experience than younger infants. To compare the unique contributions of age and experience, we conducted partial correlation analyses on attempt ratios on risky slopes. The partial correlation coefficient, r(12), between crawling experience and attempt ratios, controlling for the effects of test age, was -.58, p < .03. In contrast, the partial correlation coefficient, r(12), between test age and attempt ratios, controlling for the effects of crawling experience, was -.33, p -- .25. The fact that errors decreased with weeks of experience eliminates an alternative explanation, that infants simply relied on the experimenter to rescue them. On this alternative account, infants' behaviors might reflect their knowledge about stalwart, friendly adults rather than motor decisions about balance control. The same experimenter caught each infant at each session and most infants experienced dozens of rescues. However, if infants simply learned that the experimenter was going to catch them, they should have become more reckless across test sessions rather than more cautious as was clearly the case. b. Learning Takes a Long Time. A second important finding is that learning takes a really long time. As illustrated in Figure 9, on average, infants required more than 10 weeks of crawling and walking experience before their errors dropped from nearly 75% to below 50%. Most infants required nearly 20 weeks of locomotor experience with each posture before errors became very uncommon (around 10%).

Learning to Keep Balance

21

Moreover, serious falls outside the laboratory had no apparent impact on infants' behavior on the slopes. For example, one infant in the experimental group fell headfirst down a flight of stairs while pushing himself around in a mechanical baby walker 4 days before crawling onset and was rushed to the emergency room with a bloodied and bruised face. At his first test session a few days later, he went headfirst down risky slopes in the laboratory task. Similarly, in analyses of cross-sectional data, behavior on slopes was unaffected by serious falls at home in which infants incurred broken arms/legs, cut lips, or stitches (Adolph, 1995, 1997; Adolph et al., 1993a). Serious falls at home are also apparently unrelated to avoidance responses to a sheer drop-off on the visual cliff (Scarr & Salapatek, 1970). Just as serious falls at home did not facilitate learning in the novel laboratory task, neither did falling down slopes during the laboratory sessions. The strongest evidence for within-session learning over consecutive trials would be if failures on one trial prompted refusals on the very next trial, on the very same slope. However, we found the opposite effect. Although infants found falling to be aversive (they typically fussed after each fall), on 80% of paired trials where their first attempt resuited in falling, they attempted the same maladaptive crawling or walking method on the next trial at the same slope a few moments later. c. Infants Learn to Gauge Balance Control in Real Time. A third important finding illustrated in Figure 9 is that the decrease in errors points to immense psychological flexibility. The naturally occurring changes in infants' bodies and skills cause corresponding changes in the size of their region of permissible postural sway. Nevertheless, infants were able to make adaptive decisions about whether to descend slopes despite weekly changes in their bodies and locomotor skill levels. Additional evidence for flexibility comes from a cross-sectional study with relatively experienced 14-month-old walking infants (Adolph & Avolio, 2000). We experimentally manipulated the size of infants' sway region by loading them with shoulderpacks that varied in weight from trial to trial (120 g and 25% of infants' body weight). With the heavy shoulderpacks, the region of permissible postural sway was reduced by 30% and infants' slope thresholds showed a corresponding reduction of 4~ ~. Despite constant switching between trials of their shoulderpacks, infants could accurately gauge threats to balance as they walked down steep and shallow slopes. They treated the same absolute degree of slope as safe while wearing their light shoulderpacks but as risky while wearing their heavy shoulderpacks. On the sway model, both naturally occurring and experimentally induced changes in the sway region would require continual updating about infants' locomotor abilities relative to the degree of slant. Coping with moment-to-moment changes in the size of the sway region means that infants must learn to gauge threats to balance control in real time, in the context of their current abilities, goals, and the particular constraints of the task.

22

Karen E. Adolph

A related point about flexibility concerns the breadth of infants' generalization. Apparently, learning resulted from uneventful everyday experience maintaining balance in crawling and walking postures at home, not from experience descending slopes. None of the infants had experiences on slopes outside the laboratory. Moreover, the control infants (represented by the open symbols in Figure 9) who had no weekly experience descending laboratory slopes showed nearly identical decisions after comparable experience crawling and walking as the babies in the experimental group. A final point about flexibility concerns the consistency of infants' responses within sessions. At each test session, infants' attempt ratios were scaled to their slope thresholds. They were most likely to attempt risky slopes closest to their thresholds and least likely to attempt risky slopes most remote from their thresholds. In 89% of the 219 individual session protocols, infants' responses were entirely consistent on risky slopes, where attempt ratios were constant or steadily decreasing from the threshold slope to the steepest increment. Over weeks of crawling and walking experience, infants' attempt ratios decreased on slopes closer to their thresholds, until, finally, attempt ratios closely matched the conditional probability of success. Thus, learning reflected a process of"gearing in" to the limits of infants' abilities, a process of gradually differentiating the slopes which presented threats to balance from those that did not.

d. No Transfer from Crawling to Walking. The most striking result shown in Figure 9 was the specificity of infants' knowledge across developmental changes in posture. Learning curves across weeks of crawling and walking were separate and knowledge gained from crawling did not transfer to walking. The same infants who had responded adaptively for weeks in their old, familiar crawling posture, attempted to walk straight over impossibly steep slopes in their new, unfamiliar walking posture. Errors were just as high in the first weeks of walking as they were in the first weeks of crawling and learning was no faster the second time around. As a final test of the specificity of infants' knowledge about balance control, we tested infants in their first weeks of walking with six back-to-back trials at the steepest available 36 ~ slope: These included two trials at 36 ~ in their new uptight walking posture, then two trials in their old, familiar crawling posture, then two trials again in their new uptight posture. As in the gaps studies, this manipulation provided a way to control for infants' age while comparing their knowledge in more versus less experienced posture. Although both hands-and-knees crawling and walking were impossible at the steepest 36 ~ slope, infants' knowledge was posture specific. When new walkers faced the slope in their old crawling position, their decisions were extremely accurate, but when they faced the slope in their new uptight position, their decisions were fraught with errors. On average, the rate of falling when starting on hands and knees was .05, equally accurate as the rate of falling in the week prior to walking onset (M = .07 after 20 weeks of crawling). However, when new walkers began

Learning to Keep Balance

23

trials in an upright posture, their rate of falling was .30. Even after a reminder that the slope was risky in the crawling position and that several sliding strategies were available in their repertoires, babies continued to walk over the edge and fall in their final two walking trials. Moreover, many new walkers insisted on facing the obstacle in their new uptight posture, as though preferring to face the slope as hapless walkers rather than experienced crawlers. Despite being placed by the experimenter on their hands and knees, 10 infants in the experimental group and 7 infants in the control group stood themselves up and walked blithely over the edge and fell. Errors after standing themselves up were equally high (M = .38) as when infants began the trials in an upright posture (M = .30). Both the specificity and the flexibility of infants' behaviors argues against an alternative explanation that infants' responses were based on the absolute degree of slant rather than on the relative amount of risk. Because degree of risk has an ordinal nature--riskier slopes are steeper than safer slopes--infants might simply associate aversive consequences with the steepest slopes or learn to associate sliding and avoidance with the steepest slopes. That is, if babies simply learned to associate particular responses with absolute degree of slope, then they should have learned to attempt shallow hills and slide down steep ones regardless of changes in their level of locomotor skill and developmental changes in posture. However, they did not. Novice crawlers and walkers treated the steepest slopes as safe, regardless of their skill level: They plunged down 36 ~ and fell. Experienced crawlers and walkers geared their responses on the steepest slopes precisely to their skill level. Less proficient infants slid down or avoided going, more proficient infants used their typical crawling or walking method, and falls among experienced infants were exceedingly rare. 4. Summary

The slopes and gaps studies provide strong support for the sway model. In short, infants avoid falling by maintaining their bodies within a region of permissible postural sway. Because the size of the sway region changes from moment to moment, infants must learn to gauge their region of permissible postural sway in real time. Developmental changes in posture create new balance control systems. Each time infants acquire a new postural milestone, they must identify the relevant control parameters which define the new sway region and learn to calibrate its settings. As a result, adaptive responding in novel predicaments should increase with everyday experience in a given posture, and such responding should not transfer across developmental changes in posture. c. LEARNINGTO DETECTTHREATSTO BALANCE:THE VISUAL CLIFFAND OTHERFALLINGTASKS The sway model can account for developmental changes in infants' successes and failures in a wide range of motor tasks reported in the literature. The common

24

Karen E. Adolph

feature across tasks is that babies must avoid falling. Although most tasks were not conceived by their original designers in terms of balance control, I argue that the critical factor underlying infants' performance is their ability to gauge their region of permissible postural sway. 1. The Visual Cliff and Locomotor Experience

The best known "falling" task, of course, is the "visual cliff" (Gibson & Walk, 1960; Walk & Gibson, 1961). Since Gibson and Walk's classic studies, several researchers have found that human infants and other altricial animals require a protracted period of locomotor experience before they avoid crawling over the deep side of the visual cliff (Campos, Bertenthal, & Kermoian, 1992; Held & Hein, 1963; Rader, Bausano, & Richards, 1980; Richards & Rader, 1983; Walk, 1966; Walk & Gibson, 1961). In a particularly clever age-matched control design, Bertenthal, Campos, and Barrett (1984) showed that the duration of infants' everyday crawling experience predicts adaptive avoidance responses, independent of infants' age at testing or the age at which they began crawling. At the very same age at testing (7.5-8.5 months) only 35% of inexperienced crawlers (M = 11 days of crawling experience) avoided the apparent drop-off but 65% of more experienced infants (M -- 41 days) refused to crawl over the precipice. Several accounts have been proposed about what infants may learn during this period of experience that facilitates the coordination between perception and action. The early pioneers, Gibson and Walk (Gibson & Walk, 1960), proposed that infants must acquire depth perception to avoid the drop-off on the visual cliff. We now know that avoiding a cliff does not depend solely on depth perception because even newborns can differentiate disparities in depth (Campos, Langer, & Krowitz, 1970; Slater and Morison, 1985) and precrawling infants can use depth information adaptively to guide their movements in reaching tasks (Gordon & Yonas, 1976; McKenzie, Skouteris, Day, Hartman, & Yonas, 1993; Yonas, Granrud, Arterberry, & Hanson, 1986; Yonas & Hartman, 1993). The most widely cited modem explanation is that everyday crawling experience leads to fear of heights. This emotional response to depth information mediates adaptive avoidance responses (Campos et al., 1978, 1992). A different sort of explanation relies on associative learning about various kinds of surfaces and their consequences for locomotion. Infants may learn that cliffs are dangerous, for example, from experiences peering over the edges of sheer drop-offs and from many near-falls as vigilant parents grab them at the edge of the bed or the changing table (Thelen & Smith, 1994). Finally, others have proposed that experience leads to an appreciation of the properties of the ground surface for supporting the body (Bertenthal & Campos, 1990; Gibson & Schmuckler, 1989). In particular, experience crawling over solid ground might teach infants that locomotion is impossible without a substantial surface that they can see and feel beneath their hands and feet.

Learning to Keep Balance

25

According to the sway model, none of these accounts is sufficient for explaining how experience facilitates adaptive responses to depth information for a drop-off. If infants learn to avoid a drop-off because they are afraid of heights, know that cliffs are dangerous, or know that their bodies cannot be supported in empty space, then they should show similar adaptive avoidance responses to a drop-off regardless of the posture in which they are tested. However, our observations of infants at the edge of a sheer drop-off on gaps and at the brink of a graded drop-off on steep slopes showed that infants' learning is specific to each postural milestone in development. 2. Other Falling Tasks Although the visual cliff is the most widely known test paradigm, for more than a half century, researchers have devised novel and innovative tasks in which babies must avoid falling. Some tasks were designed to test biomechanical models of balance control by measuring babies' responses to unexpected disruptions of their balance. For example, infants were tested in sitting and standing positions on movable floors which jerked suddenly forward or backward, similar to the lurch of a subway car (Shumway-Cooke & Woollacott, 1985; Woollacott & Sveistrup, 1992; Woollacott, Hofsten, & Rosblad, 1988). Some tasks were designed to demonstrate the importance of vision or touch as a source of proprioceptive information about where the body is in space. Infants were observed trying to sit, stand, or walk in rooms where the walls jerked forward and backward to simulate the optical information generated by spontaneous forward and backward body sway (Bertenthal, Rose, & Bai, 1997; Lee & Aronson, 1974; Schmuckler & Gibson, 1989; Stoffregen et al., 1987). They were tested standing on special force platforms with their hands resting on a supporting bar to determine whether their body sway attenuates when they have something to grip onto (Metc~lfe & Clark, 2000). Other tasks were designed to test the proposal (Bernstein, 1996; Reed, 1982) that balance control underlies all perceptual-motor behaviors. Researchers measured babies' control of balance in the context of suprapostural goals such as reaching for objects at various distances and locations while keeping balance in a sitting position (McKenzie et al., 1993; Rochat, 1992; Rochat & Goubet, 1995; Rochat & Senders, 1991). Their postural responses were measured with electromyography as they engaged in reaching movements (Hofsten, 1993; Woollacott et al., 1988). More far afield, researchers designed tasks to test infants' perception of affordances for locomotion (Adolph, Eppler, & Gibson, 1993b; Gibson, Adolph, & Eppler, 1999; Gibson, 1979), to test their ability to manipulate environmental supports as tools to aid their locomotion, and to catalog developmental changes in common skills such as stair-climbing. Babies were observed coping with stance and locomotion on slippery and sticky surfaces (Lo, Avolio, Massop, & Adolph, 1999; Stoffregen, Adolph, Thelen, Gorday, & Sheng, 1997) and on squishy and rigid ones (Gibson et al., 1987). They were challenged to step over or under barriers

26

Karen E. Adolph

(Schmuckler, 1996; Van der Meer, 1997), to squeeze through wide and narrow apertures (Palmer, 1987), and to detour around obstacles in their path (Gibson & Schmuckler, 1989; Lockrnan, 1984; Schmuckler & Gibson, 1989). They were encouraged to cross wide and narrow bridges with and without handrails to hang onto (Berger & Adolph, 2002a, 2002b). They were asked to climb up and down stairs (Gesell & Thompson, 1938; Ulrich, Thelen, & Niles, 1990), ladders (McCaskill & Wellman, 1938), tall pedestals (McGraw, 1935), and steep slopes (Adolph, 1995, 1997; Adolph et al., 1993a; Eppler, Adolph, & Weiner, 1996; McGraw, 1935). They were even tested walking down slopes with weights attached to their shoulders (Adolph & Avolio, 2000) and filmed descending slopes while roller skating backward (McGraw, 1935)! The sway model provides a unifying psychological framework for understanding developmental changes in infants' performance across the range of falling tasks. Regardless of whether the tasks were originally designed to measure balance control, reaching, fear of heights, perception of surface affordances, tool use, or any other construct, the falling tasks share the common requirement of gauging in real time the current size of the sway region. That is, infants' success in each task depends on their ability to detect and respond appropriately to potential threats to balance. Moreover, every task shows a similar developmental progression. As in the visual cliff, gaps, and slopes studies, the adaptiveness of infants' responses depends on the duration of their everyday experience and their age at testing (Adolph, 1997). Older, more experienced infants show more mature muscle responses to a jerking floor, maintain balance more effectively in a moving room, are more likely to avoid crawling or walking down impossibly steep slopes, wend their way more successfully around barriers in their path, and so on. On the sway model, everyday experience should facilitate adaptive responding and experience should be specific to each posture in development. Unfortunately, few investigators have tried to understand the separate effects of age and experience in this myriad of tasks or have tested infants across developmental changes in posture. However, the available data suggest that experience is the stronger predictor of adaptive responding and that learning does not transfer across developmental changes in posture. As on the visual cliff, gaps, and slopes, experience is a better predictor than age or walking skill of infants' responses when confronted by high and low barriers in their path (Kingsnorth & Schmuckler, 2000; Schmuckler, 1996). More experienced infants are better able to judge safe heights for stepping over the barriers. As in the gaps and slopes studies, experienced crawlers avoided crossing the visual cliff when facing it from their familiar crawling position. However, the same infants went straight over it when supported uptight in mechanical baby walkers (Rader et al., 1980). Similarly, infants' ability to execute reaching detours around barriers increased over weeks of sitting, but the same infants had to relearn how to execute detours to negotiate barriers when

Learning to Keep Balance

27

they were tested while in a crawling posture (Lockman, 1984). Finally, infants' ability to maintain balance in a moving room or on a jerking floor apparently does not transfer from sitting to standing postures (Bertenthal et al., 1997). In sum, the sway model provides a unifying framework for understanding how infants come to respond adaptively in a wide range of tasks in which the penalty for error is falling down. Common to each task is behavioral flexibility--the ability to generate new solutions to new challenges to balance control on a momentto-moment basis. Consistent with the predictions of the sway model, extended periods of learning appear to occur during everyday experience maintaining balance in each posture, and learning does not transfer across developmental changes in posture.

IV. How Development May Constrain Motor Learning A. CONTENT OF EVERYDAY EXPERIENCE

In the previous sections, I described how adaptive responding in novel laboratory tasks depends on the duration of infants' everyday locomotor experience. Moreover, the results of several studies showed that extended periods of experience---on the order of several weeks or months--are required before infants display consistently adaptive responses in novel laboratory tasks that challenge balance control. In this section, I ask how infants might learn about balance control in the course of everyday experience. That is, what sorts of everyday experiences could promote learning that is flexible enough to cope with novel challenges to balance but specific enough to ensure that infants identify the relevant parameters of each new balance control system in development? Although many researchers have engaged in theorizing about learning mechanisms related to infants' locomotor experiences (e.g., Acredolo, 1988; Adolph, Vereijken, et al., 1998; Bertenthal, Campos, & Kerrnoian, 1994; Gibson, 1988; Piaget, 1954; Richards & Rader, 1983; Thelen & Smith, 1994; Walk, 1966), ironically, the theorizing to date has been largely unconstrained by empirical facts about infants' everyday locomotor experiences. In my own work, such as the gaps and slopes studies described earlier, in the work of the other investigators who observed infants in "falling" tasks, and elsewhere in the literature where investigators posited a role of locomotor experience in infants' development, experience has been treated similar to age: Experience is simply the number of days elapsed between the onset of a skill and the test date. There are a number of serious problems inherent in quantifying locomotor experience in terms of elapsed time. The most serious problem is that elapsed time is not an explanatory variable. Just as age differences alone cannot explain developmental progress, experience differences cannot address underlying learning

28

Karen E. Adolph

mechanisms without a description of the critical experience-related changes that occurred during the elapsed time between onset and test dates. The only purported mechanism that has been examined empirically is one-trial learning from serious falls, that is, some sort of fast-mapping between perceptual information for disequilibrium and the aversive consequences of falling (Bertenthal et al., 1984). On this account, overall duration of locomotor experience may predict adaptive responding because the longer the children travel around independently, the more likely they are to have experienced a serious fall. An analogy might be that long-time drivers respond more adaptively than student drivers to novel challenges because the former have had more opportunities to learn from memorable and aversive events such as steering out of a skid or avoiding a rear-end collision. As described previously, the existing data do not support such a mechanism. Very few parents report that their infants have incurred serious falls during everyday locomotion and those infants that do fall are no more likely to exhibit adaptive responses in novel "falling" tasks in the laboratory than infants who have not experienced serious mishaps (Adolph, 1995, 1997; Scarr & Salapatek, 1970). With the exception of parents' reports about falling accidents, to my knowledge, there are no published data that describe the actual content of human infants' everyday motor experience--how frequently infants locomote, where they go and how often, how far they travel, and what surfaces and paths they traverse. A second problem with quantifying locomotor experience in terms of elapsed time concerns the way that duration of experience is calculated. To determine an onset day, the researcher must set a definitional criterion (crawls a distance of 90 cm on hands and knees, walks a distance of 305 cm without holding onto parents or furniture for support, etc.). The problem is that motor skills typically appear in a graded fashion, where earlier appearing behaviors are previews of the later appearing target skills (e.g., Gesell, 1946; McGraw, 1945). Thus, infants may acquire a significant amount of relevant experience prior to passing the criterion for the onset day. For example, infants may crawl or walk slightly less than the criterial distances or exhibit precursory approximations to the final skill (e.g., crawl on belly, pivot in circles, balance on hands and knees, and so on, prior to crawling to criterion on hands and knees). Such smaller bits of experience with earlier appearing approximations to a target skill have proven to be powerful predictors of performance with the target skill weeks later (Adolph, Vereijken, et al., 1998). A third problem with quantifying locomotor experience in terms of elapsed time concerns the developmental trajectory of motor skill acquisition across days. By equating duration of experience with the number of days between onset and test days, researchers inherently assume that developmental trajectories are abrupt rather than gradual or variable. That is, infants who pass criterion for a target skill on Monday should perform the skill on Tuesday and Wednesday and each successive day between onset and test. However, if developmental trajectories do not always turn on and off like a faucet so that each day is like the last, then experience is not

Learning to Keep Balance

29

necessarily meted out in daily doses. That is, an infant who passes criterion for a skill on Monday may not demonstrate the skill again until Friday or may pass criterion only every other day, and so on. Finally, the duration and content of infants' experience is typically estimated from parents' retrospective reports. With the exception of a few longitudinal studies where infants were tracked prospectively (Adolph, 1997; Bertenthal et al., 1984; Eppler et al., 1997; McGraw, 1935), parents must rely on their memories and incidental records to report onsent dates and special experiences such as serious falls, exposure to stairs and slides, and so on. Our data suggest that with careful probing, structured interview methods, and support from parents' "baby books" and calendars, parents' reports of onset dates are relatively accurate; for example, duration of experience calculated from retrospective reports is well correlated with objective measures of locomotor skill such as step length and step velocity. However, retrospective reports are subject to smoothing errors and cannot be independently verified. It is difficult for parents to remember which day their infants crawled 90 cm or walked 305 cm during everyday locomotion at home when they are queried weeks or months later in the laboratory. To redress the problems in understanding experience in terms of elapsed time, we have designed several convergent prospective diary methods to describe what infants' everyday locomotor experience really entails. Because the diaries are prospective rather than retrospective we hope to minimize errors in parents' reports. Our aim is to compile a rich, archival database of everyday experience in a sample of healthy, typically developing infants and to thereby set boundary conditions on possible learning mechanisms.

1. Daily Frequency of Infant Locomotion We devised a "checklist diary" method to test the daily frequency of infants' locomotor experience (Adolph, Biu, Pethkongathan, & Young, 2002). We train parents to check off 56 motor skills, events, and instances of passive locomotion on a daily basis. Each motor skill is defined according to a strict, standard criterion. For example, forms of locomotion include belly crawling, hands-and-knees crawling, walking, and running a minimum of 305 cm (10 feet) consecutively, and cruising (walking sideways holding onto furniture for support) three consecutive steps (the length of a standard couch). Precursors to bonafide locomotion include taking one or two forward steps on belly, hands and knees, or feet; pivoting in circles on belly; and so on. The events include serious falls (resulting in bruises, bumps, breaks, or blood); serious illnesses or family trips that may have prevented infants from engaging in normal gross motor activities (bed-bound, traveling in car for long stretches of the day, etc.); and classes that may have presented infants with special motor experiences (e.g., Gymboree). To date, five infants in the New York City area have completed participation in the checklist diary study. Parents began the diaries when their infants were 1.2-4.0

30

Karen E. Adolph

months old (all prior to rolling or sitting independently) and stopped keeping the diaries when their infants were 14.6-19.2 months old. During monthly visits to the laboratory, we tested infants on the list of skills, queried parents' about their diary entries, and refreshed parents' criteria for noting each skill. Preliminary data suggest that the practice of quantifying locomotor experience in terms of days between onset and test days may indeed promote an erroneous understanding of infants' everyday experience. In every case where infants demonstrated a target locomotor skill, they displayed precursory approximations to the skill prior to its onset date (and, in the case of crawling, they sometimes demonstrated precursors without ever passing criterion for the target skill). For example, they managed to maintain static postures or to take a couple of faltering steps prior to traveling the criterion distances. Both precursory skills and bonafide locomotion showed abrupt (1 day) transitions in only 18 of 70 possible cases. In all of the other cases, skills appeared gradually such that infants demonstrated the skill for a day or two but failed to demonstrate it on succeeding days, sometimes stretching the on-off pattern over several weeks. Similarly, crawling and cruising disappeared gradually (for five of eight possible cases) such that infants stopped performing the skill for a day or two but sometimes performed it on succeeding days. Across skills, transitional periods ranged from 1 to 107 days. Even the periods of presumed stability that flank transitional periods were marred by occasional appearances or disappearances of the target skill. Thus, the developmental trajectory of infants' motor skills is not necessarily stage like, "turning on" or "off" like a faucet from one day to the next; the frequency is not necessarily distributed regularly across days, and more days between onset and test days does not guarantee more days of experience. Rather, the daily frequency of locomotor skills is typically messy and uneven, both in the transitional periods when skills first appear in or disappear from infants' repertoires and during the flanking periods when skills are relatively stable in their presence or absence. In addition, in accordance with previous research, we found that infants' locomotor experience was generally uneventful. Only one infant (my own!) experienced a serious fall which required medical attention. Neither the serious fall nor the everyday bumps that all the infants experienced appeared to hinder their progress at locomotion. The only circumstance that kept infants from locomoting was preventing their access to the floor. Several families went on vacations during which infants were cooped up on long car tides, penned in a tiny camping tent or play space, or otherwise denied access to the floor or to large enough floor spaces to pass criterion for locomotion. One child had a serious medical illness which kept him bed-bound for several weeks.

2. The Nature of lnfants' Travels Our "telephone diary" method is designed to obtain detailed information about the content of infants' crawling experience on a minute-to-minute basis (Chan et al.,

Learning to Keep Balance

31

1999, 2000). With the telephone diary method, we train parents over a 4- to 6week period to track their crawling infants' active and passive locomotion from the time babies wake up until the time they go to bed. Parents telephone throughout the day to a time-stamped answering machine about where their babies went and how they got there. A researcher calls the parents at the end of each day to verify the information and to fill in missing gaps in the protocols. Then, from detailed blueprints of the infants' homes, we reconstruct their crawling paths, the distance that they traveled, the locations that they visited, and so on. To date, seven infants have participated in versions of the telephone crawling diary. My own daughter began testing as a prelocomotor infant at 2 months and completed her last diary entry 129 days after she began crawling to criterion (10 feet consecutively). Three infants began testing on their first day of crawling between 6.5 and 12.3 months of age and completed testing 13.7-16.4 weeks after crawling onset. Three infants began testing 53-100 days after crawling onset (between 10 and 10.5 months of age) and completed testing 3.1-4.1 weeks later. Our preliminary data show that prior to crawling onset, passive locomotion occupied only 7.3% of the day, on average, and time on the floor occupied only 15.9% of the day. However, after crawling onset, infants acquired massive amounts of varied experiences with independent locomotion and balance control. They spent, on average, 16.8% of their walking day in passive locomotion (being carried or wheeledin strollers) and an additional 40.8% of their day on the floor (crawling, cruising, and playing). In terms of raw numbers, infants spent 5.1 hr per day, on average, in active balance and locomotion and 2.6 hr per day in passive locomotion. Although infants spent most of their floor time in the common family spaces (kitchen/family/eating room areas), each day they managed to travel through most of the open floor spaces in most of the rooms of their homes, visiting 5.7-12.3 different surfaces, on average, en route. On average, infants crawled 27.1-42.6 m/hr and covered a total daily distance of 60.4-187.8 m. Given that crawlers average 17 steps per meter while crawling continuously over the laboratory floor, the preliminary telephone diary data suggest that infants experienced 1028-3198 crawling steps per day. Nearly all of infants' forays (97.2%) were one-way trips punctuated by playing or stopping to rest.

3. Quantifying Steps Our "step counter diary" method is designed to quantify infants' walking steps directly and automatically (Adolph, Avolio, Barrett, Mathur, & Murray, 1998). We outfit walking infants with tiny foot switches in their shoes. A Gaitrite paper-thin pressure-sensitive pad is slipped into the insole of their shoes and attached to a tiny micromemory chip and battery clipped to the outside of their shoes. Parents keep a daily checklist diary of the rooms and floor coverings infants travel over and of visits to outside surfaces. At the end of the testing period, we download the data into a computer and calculate the number of walking steps per time block. Preliminary

32

Karen E. Adolph

data with three 14-month-old toddlers using a beta-test version of the step counter showed that infants were on the floor more than 50% of a typical 12-hr walking day and crossed most of the floor coverings in most of the rooms in their homes. Each infant registered between 500 and 1500 steps per hour. The normal cadence for a 14-month-old toddler is, on average, 190 steps per minute while walking continuously over the laboratroy floor, and each step length averages 30 cm. Thus, the preliminary step counter data indicate that infants' walking experience occurs in fits and starts, with bursts of activity separated by rest periods where they stand still or play. By the end of the day, they may have traveled a total distance of 2700 m.

4. How Experience Promotes Learning In the course of everyday experience with each posture in development, infants execute various exploratory movements, glean the resulting perceptual information, practice compensatory sway responses, and acquire a repertoire of alternative strategies when maintaining balance is impossible. How, then, does learning work? The primary aim of the diary studies was to describe the actual content of everyday locomotor experience in human infants. The hope was that a rich set of descriptive data, obtained with convergent methods, might constrain theorizing about how experience promotes adaptive motor control. Despite the preliminary nature of the data, the small sample sizes, and large individual differences, four findings seem clear and important. First, in accordance with previous studies, our diary data showed that infants' locomotor experience is generally happy and uneventful. Infants rarely experience serious falls or incur frightening episodes with locomotion or balance. Second, infants' experience occurs in massive daily doses compounded over the days prior to and following their official onset days. Third, infants gain experience with stationary balance and locomotion under wildly variable conditions in terms of surfaces, places, and events. And fourth, during the course of each day, experience with locomotion is always interspersed with frequent rest periods. From day to day, experience with locomotion typically alternates between practice days and rest days, especially during the acquisition period. How might these findings help us to understand possible learning mechanisms? The finding that experience is generally positive speaks against one-trial learning. Indeed, the data suggest the opposite: Infants' learning about balance control may occur over many thousands of trials during the course of each day. Even the most elaborate laboratory training studies cannot begin to approximate the massive amount of experience that infants obtain in everyday locomotion. To put the sheer magnitude of infants' locomotor experience into perspective, each day crawling infants practice keeping balance for more than 5 hr; during that time, they can travel the lengths of two football fields and take more than 3000 crawling steps. Walking infants practice keeping balance for more than 6 hr per day; during that time, they can travel the lengths of 29 football fields and take 9000 walking steps. In addition to the time that they are in transit, crawlers, cruisers, and walkers

Learning to Keep Balance

33

gain hours of daily experience maintaining balance in stationary positions. Each little crawling and walking step, each forward lean and reach, each time infants roll over or sit up is a tiny "trial" with balance control. These daily experiences are compounded over the total protracted period of experience, beginning with infants' partial approximations of a new postural milestone, stuttering through the transition period when they first meet criterion for the skill proper, and continuing through the period when the skill is relatively stable in their repertoires. Our preliminary data suggest that infants' locomotor experience should be tightly measured in epochs. Similar to the immense amounts of daily practice which promote expert performance in musicians, athletes, typists, and chess players (Bryan & Harter, 1897, 1899; Ericsson & Charness, 1994; Ericsson, Krampe, & Tesch-Romer, 1993), there are hundreds of thousands of trials spread over several weeks or months before babies show adaptive responses in novel laboratory tasks which challenge balance control. Infants' massive accumulated practice is not synonymous with dull, rote, drudgery. Babies are not performing the same movements over and over in response to the same goals and impediments in the same environmental context, as if on a "blocked" practice regimen in the laboratory where participants are presented with the same problem on successive trials. Rather, infants' cope with balance control in an astounding variety of events, places, and surfaces--even in the familiar setting of their homes and play yards. The changeability and unpredictability of everyday experience resembles an exaggerated version of "variable" or "random" practice, where environmental conditions such as stimulus increment and trial order vary from one attempt to the next. In the laboratory, variable and random practice are detrimental for performance during acquisition of a skill but lead to better performance during transfer tasks under novel conditions. Conversely, blocked practice is beneficial for performance during acquisition but leads to decrements during transfer (Gentile, 2000; Schmidt, 1988). A classic explanation for the difference between blocked and variable/random practice is that the former leads to repeating a particular solution over and over on successive trials whereas the latter leads to a process of continually generating new solutions or generating old solutions anew (Gentile, 2000; Schmidt, 1988). The change in context during variable/random practice may actually prevent infants from merely repeating solutions by causing interference so that the current solution exits working memory and infants are required to construct or reconstruct a solution on successive attempts. Similarly, changing the context from moment to moment may dissuade infants from relying on simple associations between stimuli and responses and force them instead to track the complex constellation of relevant factors for calibrating their current region of sway. Put simply, variety of experience promotes "learning to learn," rather than learning particular solutions (Bernstein, 1967, 1996; Harlow, 1949, 1959). Such variety of experience might be exactly the kind of practice that infants need to achieve adaptive responding when challenged with novel threats to balance.

34

Karen E. Adolph

The rest periods between bouts of locomotion during each day and the rest days between practice days when skills are displayed may function to increase contextual interference. That is, by stopping to play or socialize between bursts of locomotion, infants change the motivational and biomechanical context of their activity. During the next crawling foray or the next stationary posture, infants must solve the problem of balance control all over again. Alternatively, as in laboratory studies of distributed and massed practice, infants' intermittent experiences with locomotion within and across days may provide them with time to consolidate learning or to allow fatigue or flagging motivation to dissipate (Schmidt, 1988). In sum, the descriptive diary data suggest that infants' everyday opportunities for learning involve the sort of training regimen that promotes expert performance, adaptability, and transfer in laboratory studies of skill acquisition: large amounts of variable, distributed practice. B. ENSURING FLEXIBILITY AND SPECIFICITY

The final section of this chapter addresses a long-standing problem in developmental psychology, understanding the relation between learning and development. The problem of balance control provides a particularly apt illustration of how peripheral developmental changes can constrain the course of learning. By peripheral developmental changes, I refer to ongoing changes that are not linked to the target behavior via a neural prescription in the central nervous system (Thelen & Smith, 1994). As conceived within the context of the sway model, peripheral developmental changes play a central role in ensuring that learning is sufficiently flexible to cope with a variable world and sufficiently specific that infants learn what they need to know. On the sway model, sitting, crawling, and walking are different balance control systems: A hip-torso system, a prone system, and an uptight system. Infants must learn to identify the relevant control parameters that are specific to each of these systems--the relevant muscle groups to control sway, the relevant sensory input to anticipate disruptions of balance, the relevant effectors to generate perceptual information, and so on. How might peripheral developmental changes ensure the necessary specificity to enable infants to identify the relevant balance control parameters? As infants acquire the ability to assume new postures in development, they are forced to search out the new defining parameters because the old parameters simply do not work. If they rely on their old balance control system to guide them, babies will fall over the cliff, or into the gap, or down the slope. The creation of new balance control systems forces infants to learn about the new control parameters. On the sway model, maintaining balance in any posture requires infants to continually monitor the current size of the sway region. Infants must learn to recalibrate their movements, with infinitely sensitive adjustments for each potential

Learning to Keep Balance

35

disruption of balance. Rather than learning facts or rules ("cliffs are dangerous," "descend stairs by crawling backward," "I'm a poor crawler, . . . . my body is fat/thin/tall/short"), infants must learn how to gauge threats to balance in real time. Similar to Harlow's (Harlow, 1949, 1959) notion of "learning to l e a r n , " flexibility in motor control requires much more than simple associative pairing. How might peripheral developmental changes ensure the necessary flexibility to allow infants to cope with unlimited, continual variability? One avenue for constraining the course of learning is by maintaining variability in infants' bodies, skill levels, and environments. In a sense, infants are prevented from learning particular facts or rules for keeping balance because as soon as a fact would be consolidated, the circumstances have again changed: Babies' bodies have grown, their skill level has improved, and they have been exposed to a new array of surfaces and events. A second avenue in which peripheral developmental changes might constrain the course of learning is by extending the learning period. Individual infants' experiences differ widely in terms of opportunities for learning. Chubbier, less muscular babies must work harder to locomote through the environment. Less skillful infants gain access to a subset of the locations and events available to more skillful infants. Infants in small New York City apartments have smaller arenas to travel in compared with infants in large suburban homes. Infants in cold climates are bundled into more clothing and have less frequent access to outdoor surfaces. Nonetheless, all healthy babies eventually learn to keep balance and respond adaptively to novel challenges. On the sway model, the specific experiences do not matter but variable practice across different local conditions does. By extending the learning period so that infants spend weeks or months in each posture, infants are ensured distributed practice keeping balance over time, surfaces, and events. C. CONCLUSION

In this chapter, I have presented the sway model in an effort to provide a unifying framework for understanding infants' performance on the visual cliff and dozens of other tasks in the literature on perceptual-motor development. In addition, I had loftier aims. I have argued that keeping balance is not merely a biomechanical problem for movement scientists and sports enthusiasts. It is a manifestly psychological problem that is central to understanding motor control. I have also argued that studies of infants falling can speak to the general problem in developmental psychology of the relation between learning and development. Balance control provides a rich example of how peripheral developmental changes can constrain learning without building knowledge anywhere into the system. Finally, balance control highlights the importance of psychological flexibility, one of the most wondrous achievements of all.

36

Karen E. Adolph

ACKNOWLEDGMENTS This research was supported by NICHD Grant #HD33486 to Karen Adolph. I gratefully acknowledge Anthony Avolio, Marion Eppler, Ann Gentile, Eleanor Gibson, and Esther Thelen for their help in understanding the data; members of my Infant Motor Development Laboratory at Carnegie Mellon University and New York University for their help in data collection, coding, and analyses; and the members of Esther Thelen's Infant Motor Development Laboratory at Indiana University for their help in data collection.

REFERENCES Acredolo, L. E (1988). Infant mobility and spatial development. In J. Stiles-Davis, M. Kritchevsky, & U. Bellugi (Eds.), Spatial cognition: Brain bases and development (pp. 157-166). Hillsdale, NJ: Erlbaum. Adolph, K. E. (1995). A psychophysical assessment of toddlers' ability to cope with slopes. Journal of Experimental Psychology: Human Perception and Performance, 21, 734-750. Adolph, K. E. (1997). Learning in the development of infant locomotion. Monographs of the Society for Research in Child Development, 62 (3, Serial No. 251), pp. 1-162. Adolph, K. E. (2000). Specificity of learning: Why infants fall over a veritable cliff. Psychological Science, 11, 290-295. Adolph, K. E., & Avolio, A. M. (2000). Walking infants adapt locomotion to changing body dimensions. Journal of Experimental Psychology: Human Perception and Performance, 26, 1148-1166. Adolph, K. E., Avolio, A. M., Barrett, T., Mathur, P., & Murray, A. (1998, April). Step counter: Quantifying infants' everyday walking experience. International Conference on Infant Studies, Atlanta, GA. Adolph, K. E., Biu, O., Pethkongathan, S., & Young, J. (2002, April). Transitions in development: The trajectory of motor skill acquisition. International Conference on Infant Studies, Toronto, Canada. Adolph, K. E., & Eppler, M. A. (1998). Development of visually guided locomotion. Ecological Psychology, 10, 303-321. Adolph, K. E., & Eppler, M. A. (2002). Flexibility and specificity in infant motor skill acquisition. In J. Fagen (Ed.), Progress in infancy research (Vol. 2, pp. 121-167). Norwood, NJ: Ablex. Adolph, K. E., Eppler, M. A., & Gibson, E. J. (1993a). Crawling versus walking infants' perception of affordances for locomotion over sloping surfaces. Child Development, 64, 1158-1174. Adolph, K. E., Eppler, M. A., & Gibson, E. J. (1993b). Development of perception of affordances. In C. Rovee-Collier & L. P. Lipsitt (Eds.),Advances in infancy research (Vol. 8, pp. 51-98). Norwood, NJ: Ablex. Adolph, K. E., Eppler, M. A., Marin, L., Weise, I. B., & Clearfield, M. W. (2000). Exploration in the service of prospective control. Infant Behavior and Development, 23, 441-460. Adolph, K. E., Vereijken, B., & Shrout, P. E. (2002). What changes in infant walking and why. Manuscript under review. Adolph, K. E., Vereijken, B., & Denny, M. A. (1998). Learning to crawl. Child Development, 69, 1299-1312. Berger, S. E., & Adolph, K. E. (2002a). Infants use handrails as tools in a locomotor task. Manuscript under review. Berger, S. E., & Adolph, K. E. (2002b, April). Infants' handrail use as a tool for balance control. International Conference on Infant Studies, Toronto, Canada. Bernstein, N. (1967). The coordination and regulation of movements. Oxford: Pergamon Press. Bernstein, N. (1996). Dexterity and its development. In M. L. Latash & M. T. Turvey (Eds.), Dexterity and its development (pp. 3-244). Mahwah, N.J.: Erlbaum.

Learning to Keep Balance

37

Bertenthal, B. I., & Bai, D. L. (1989). Infants' sensitivity to optical flow for controlling posture. Developmental Psychology, 25, 936-945. Bertenthal, B. I., & Campos, J. J. (1990). A systems approach to the organizing effects of self-produced locomotion during infancy. In C. Rovee-Collier & L. P. Lipsitt (Eds.),Advances in infancy research (Vol. 6, pp. 1-60). Norwood, NJ.: Ablex. Bertenthal, B. I., Campos, J. J., & Barrett, K. C. (1984). Self-produced locomotion: An organizer of emotional, cognitive, and social development in infancy. In R. N. Emde & R. J. Harmon (Eds.), Continuities and discontinuities in development (pp. 175-210). New York: Plenum Press. Bertenthal, B. I., Campos, J. J., & Kermoian, R. (1994). An epigenetic perspective on the development of self-produced locomotion and its consequences. Current Directions in Psychological Science, 3, 140-145. Bertenthal, B. I., Rose, J. L., & Bai, D. (1997). Perception-action coupling in the development of visual control of posture. Journal of Experimental Psychology: Human Perception and Performance, 23, 1631-1643. Breniere, Y., & Bril, B. (1988). Pourquoi les enfants marchent en tombant alors que les adultes tombent en marchant? [Why does the child walk in falling whereas the adult falls in walking?] Comptes Rendus, Academy of Science, Paris, 307, 617-622. Bryan, W. L., & Harter, N. (1897). Studies in the physiology and psychology of the telegraphic language. Psychological Review, 4, 27-53. Bryan, W. L., & Harter, N. (1899). Studies on the telegraphic language: The acquisition of a hierarchy of habits. Psychological Review, 6, 345-375. Campos, J., Hiatt, S., Ramsay, D., Henderson, C., & Svejda, M. (1978). The emergence of fear on the visual cliff. In M. Lewis & L. Rosenblum (Eds.), The development of affect (pp. 149-182). New York: Plenum. Campos, J. J., Bertenthal, B. I., & Kermoian, R. (1992). Early experience and emotional development: The emergence of wariness of heights. Psychological Science, 3, 61-64. Campos, J. J., Langer, A., & Krowitz, A. (1970). Cardiac responses on the visual cliff. Science, 170, 196-197. Chan, M., Lu, Y., Matin, L., & Adolph, K. E. (1999). A baby's day: Capturing crawling experience. In M. A. Grealy & J. A. Thompson (Eds.), Studies in perception and action V (pp. 245-249). Mahwah, NJ: Erlbaum. Chan, M. Y., Biancaniello, R., Adolph, K. E., & Marin, L. (2000, July). Tracking infants' locomotor experience: The telephone diary. International Conference on Infant Studies, Brighton, England. Cornsweet, T. N. (1962). The staircase-method in psychophysics. American Journal of Psychology, 75, 485-491. Eppler, M. A., Adolph, K. E., & Weiner, T. (1996). The developmental relationship between exploration and action on sloping surfaces. Infant Behavior and Development, 19, 391-405. Eppler, M. A., Satterwhite, T., Wendt, J., & Bruce, K. (1997). Infants' responses to a visual cliff and other ground surfaces. In M. A. Schmuckler & J. M. Kennedy (Eds.), Studies in Perception and Action IV (pp. 219-222). Mahway, NJ: Erlbaum. Ericsson, K. A., & Charness, N. (1994). Expert performance: Its structure and acquisition. American Psychologist, 49, 725-747. Ericsson, K. A., Krampe, R. T., & Tesch-Romer, C. (1993). The role of deliberate practice in the acquisition of expert performance. Psychological Review, 100, 363-406. Forssberg, H., & Nashner, L. M. (1982). Ontogenetic development of postural control in man: Adaptation to altered support and visual conditions during stance. Journal of Neuroscience, 2, 545-552. Gentile, A. M. (2000). Skill acquisition: Action, movement, and neuromotor processes. In J. Carr & R. Shepard (Eds.), Movement science: Foundations for physical therapy in rehabilitation (2nd ed.). New York: Aspen Press. Gesell, A. (1946). The ontogenesis of infant behavior. In L. Carmichael (Ed.), Manual of child psychology (pp. 295-331). New York: Wiley.

38

Karen E. Adolph

Gesell, A., & Thompson, H. (1938). The psychology of early growth including norms of infant behavior and a method of genetic analysis. New York: Macmillan. Gibson, E. J. (1988). Exploratory behavior in the development of perceiving, acting and the acquiring of knowledge. Annual Review of Psychology, 39, 1-41. Gibson, E. J., Adolph, K. E., & Eppler, M. A. (1999). Affordance. In E Keil (Ed.), Encyclopedia of the cognitive sciences. Cambridge, MA: MIT Press. Gibson, E. J., & Pick, A. D. (2000). The ecological approach to perceptual learning and development. New York: Oxford University Press. Gibson, E. J., Riccio, G., Schmuckler, M. A., Stoffregen, T. A., Rosenberg, D., & Taormina, J. (1987). Detection of the traversability of surfaces by crawling and walking infants. Journal of Experimental Psychology: Human Perception and Performance, 13, 533-544. Gibson, E. J., & Schmuckler, M. A. (1989). Going somewhere: An ecological and experimental approach to development of mobility. Ecological Psychology, 1, 3-25. Gibson, E. J., & Walk, R. D. (1960). The "visual cliff." Scientific American, 202, 64-71. Gibson, J. J. (1958). Visually controlled locomotion and visual orientation in animals. British Journal of Psychology, 49, 182-194. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Goldfield, E. C. (1989). Transition from rocking to crawling: Postural constraints in infant movement. Developmental Psychology, 25, 913-919. Gordon, E R., & Yonas, A. (1976). Sensitivity to binocular depth information in infants. Journal of Experimental Child Psychology, 22, 413-422. Harlow, H. E (1949). The formation of learning sets. Psychological Review, 56, 26-39. Harlow, H. E (1959). Learning set and error factor theory. In S. Koch (Ed.), Psychology: A study of a science (pp. 492-533). New York: McGraw-Hill. Held, R., & Hein, A. (1963). Movement-produced stimulation in the development of visually guided behavior. Journal of Comparative and Physiological Psychology, 56, 872-876. Hofsten, C. (1993). Prospective control: A basic aspect of action development. Human Development, 36, 253-270. Kingsnorth, S., & Schmuckler, M. S. (2000). Walking skill versus walking experience as a predictor of barrier crossing in toddlers. Infant Behavior and Development, 23, 331-350. Lampl, M. (1993). Evidence of saltatory growth in infancy. American Journal of Human Biology, 5, 641-652. Lampl, M., Veldhuis, J. D., & Johnson, M. L. (1992). Saltation and statis: A model of human growth. Science, 258, 801-803. Lee, D. N., & Aronson, E. (1974). Visual proprioceptive control of standing in human infants. Perception and Psychophysics, 15, 529-532. Lee, D. N., & Lishman, J. R. (1975). Visual proprioceptive control of stance. Journal of Human Movement Studies, 1, 87-95. Leo, A. J., Chiu, J., & Adolph, K. E. (2000, July). Temporal and functional relationships of crawling, cruising, and walking. International Conference on Infant Studies, Brighton, England. Lo, T., Avolio, A. M., Massop, S. A., & Adolph, K. E. (1999). Why toddlers don't perceive risky ground based on surface friction. In M. A. Grealy & J. A. Thompson (Eds.), Studies in Perception and Action V (pp. 231-235). Mahwah, NJ: Erlbaum. Lockman, J. J. (1984). The development of detour ability during infancy. Child Development, 55, 482-491. McCaskill, C. L., & Wellman, B. L. (1938). A study of common motor achievements at the preschool ages. Child Development, 9, 141-150. McCollum, G., & Leen, T. K. (1989). Form and exploration of mechanical stability limits in erect stance. Journal of Motor Behavior, 21, 225-244. McGraw, M. (1935). Growth: A study of Johnny and Jimmy. New York: Appleton-Century.

Learning to Keep Balance

39

McGraw, M. (1945). The neuromuscular maturation of the human infant. New York: Columbia University Press. McKenzie, B. E., Skouteris, H., Day, R. H., Hartman, B., & Yonas, A. (1993). Effective action by infants to contact objects by reaching and leaning. Child Development, 64, 415-429. Metcalfe, J. S., & Clark, J. E. (2000). Sensory information affords exploration of posture in newly walking infants and toddlers. Infant Behavior and Development, 23, 391-405. Nashner, L. M., & McCollum, G. (1985). The organization of human postural movements: A formal basis and experimental synthesis. Behavioral and Brain Sciences, 8, 135-172. Palmer, C. E. (1944). Studies of the center of gravity in the human body. Child Development, 15, 99-163. Palmer, C. E (1987, April). Between a rock and a hard place: Babies in tight spaces. Poster presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Piaget, J. (1954). The construction of reality in the child. New York: Basic Books. Rader, N., Bausano, M., & Richards, J. E. (1980). On the nature of the visual-cliff-avoidance response in human infants. Child Development, 51, 61-68. Reed, E. S. (1982). An outline of a theory of action systems. Journal of Motor Behavior, 14, 98134. Riccio, G. E. (1993). Information in movement variability about the qualitative dynamics of posture and orientation. In K. M. Newell & D. M. Corcos (Eds.), Variability and Motor Control (pp. 317-357). Champaign, IL: Human Kinetics. Riccio, G. E., & Stoffregen, T. A. (1988). Affordances as constraints on the control of stance. Human Movement Science, 7, 265-300. Richards, J. E., & Rader, N. (1983). Affective, behavioral, and avoidance responses on the visual cliff: Effects of crawling onset age, crawling experience, and testing age. Psychophysiology, 20, 633-642. Rochat, E (1992). Self-sitting and reaching in 5- to 8-month-old infants: The impact of posture and its development on early eye-hand coordination. Journal of Motor Development, 24, 210-220. Rochat, E, & Goubet, N. (1995). Development of sitting and reaching in 5- to 6-month-old infants. Infant Behavior and Development, 18, 53-68. Rochat, E, & Senders, S. J. (1991). Active touch in infancy: Action systems in development. In M. J. Weiss & E R. Zelazo (Eds.), Biological constraints and the influence of experience (pp. 412-442). Norwood NJ: Ablex. Scarr, S., & Salapatek, E (1970). Patterns of fear development during infancy. Merrill-Palmer Quarterly, 16, 53-90. Schmidt, R. A. (1988). Motor control and learning: A behavioral emphasis. Champaign, IL: Human Kinetics. Schmuckler, M. A. (1996). Development of visually guided locomotion: Barrier crossing by toddlers. Ecological Psychology, 8, 209-236. Schmuckler, M. A., & Gibson, E. J. (1989). The effect of imposed optical flow on guided locomotion in young walkers. British Journal of Developmental Psychology, 7, 193-206. Shumway-Cooke, A., & Woollacott, M. H. (1985). The growth of stability: Postural control from a developmental perspective. Journal of Motor Behavior, 17, 131-147. Slater, A., & Morison, V. (1985). Shape constancy and slant perception at birth. Perception, 14, 337344. Stoffregen, T., Adolph, K. E., Thelen, E., Gorday, K. M., & Sheng, Y. Y. (1997). Toddlers' postural adaptations to different support surfaces. Motor Control, 1, 119-137. Stoffregen, T. A., & Riccio, G. E. (1988). An ecological theory of orientation and the vestibular system. Psychological Review, 95, 3-14. Stoffregen, T. A., Schmuckler, M. A., & Gibson, E. J. (1987). Use of central and peripheral optical flow in stance and locomotion in young walkers. Perception, 16, 113-119.

40

Karen E. Adolph

Thelen, E., & Smith, L. B. (1994). A dynamic systems approach to the development of cognition and action. Cambridge, MA: MIT Press. Titzer, R. (1995, March). The developmental dynamics of understanding transparency. Paper presented at the meeting of the Society for Research in Child Development, Indianapolis. Ulrich, B., Thelen, E., & Niles, D. (1990). Perceptual determinants of action: Stair-climbing choices of infants and toddlers. In J. E. Clark & J. Humphrey (Eds.), Advances in Motor Development Research (Vol. 3). New York: AMS. Van der Meer, A. L. H. (1997). Visual guidance of passing under a barrier. Early Development and Parenting, 6, 149-157. Walk, R. D. (1966). The development of depth perception in animals and human infants. Monographs of the Society for Research in Child Development, 31 (5, Serial No. 107), 82-108. Walk, R. D., & Gibson, E. J. (1961). A comparative and analytical study of visual depth perception. Psychological Monographs, 75 (15, Whole No. 519). Woollacott, M., & Sveistrup, H. (1992). Changes in the sequencing and timing of muscle response coordination associated with developmental transitions in balance abilities. Human Movement Science, 11, 23-36. Woollacott, M. H., Hofsten, C., & Rosblad, B. (1988). Relation between muscle response onset and body segmental movements during postural perturbations in humans. Experimental Brain Research, 72, 593-604. Yonas, A., Granrud, C., Arterberry, M., & Hanson, B. (1986). Distance perception from linear perspective and texture gradients. Infant Behavior and Development, 9, 247-256. Yonas, A., & Hartman, B. (1993). Perceiving the affordance of contact in four- and five-month-old infants. Child Development, 64, 298-308.

S E X U A L S E L E C T I O N A N D H U M A N LIFE H I S T O R Y

David C. Geary DEPARTMENT OF PSYCHOLOGICAL SCIENCES UNIVERSITY OF MISSOURI AT COLUMBIA COLUMBIA, MISSOURI 65211

I. INTRODUCTION II. N A T U R A L SELECTION AND LIFE HISTORY A. N A T U R A L SELECTION B. LIFE HISTORY III. SEXUAL SELECTION A. M A T I N G OR PARENTING? B. I N T R A S E X U A L COMPETITION C. INTERSEXUAL CHOICE IV. LIFE HISTORY AND SEXUAL SELECTION A. I N T R A S E X U A L COMPETITION B. INTERSEXUAL CHOICE C. PHENOTYPIC PLASTICITY V. H U M A N D E V E L O P M E N T A L SEX DIFFERENCES A. SEXUAL SELECTION DURING H U M A N EVOLUTION B. SEX DIFFERENCES IN LIFE HISTORY C. SEX DIFFERENCES IN D E V E L O P M E N T A L ACTIVITY VI. CONCLUSION REFERENCES

I. Introduction What is the evolutionary raison d'etre of lifetimes and effort? w R . D. Alexander, The biology ofmoral systems (1987, p. 38)

Sexual selection and life history are firmly established disciplines in evolutionary biology, and associated theory and research are focused on determining the ultimate and proximate causes of sex differences and developmental patterns, respectively 41 ADVANCESIN CHILDDEVELOPMENT AND BEHAVIOR,VOL.30

Copyright 2002, Elsevier Science (USA). All rightsreserved. 0065-2407/02 $35.00

42

David C. Geary

(Andersson, 1994; Charnov, 1993; Darwin, 1871; Roff, 1992). Research in human developmental science in general and human developmental sex differences in particular has not been informed by this wealth of empirical and theoretical work, with a few exceptions (Archer, 1992; Bjorklund & Pellegrini, 2002; Bogin, 1999; Freedman, 1974; Hill & Kaplan, 1999; Kenrick & Luce, 2000). As a redress, the general focus here is on relating human developmental sex differences to sexual selection and life history (Geary, 1999). In particular, I describe theory and evidence regarding the view that many human life history traits and developmental sex differences have evolved as a result of various forms of social competition (Alexander, 1987, 1989; Geary & Flinn, 2001). To provide the necessary background and to introduce human developmental scientists to relevant work in evolutionary biology, in the next three sections I review research on life history, sexual selection, and their relation in nonhuman species. Then I relate the basic patterns and principles described in the first three sections to human developmental sex differences, with a focus on the relation between the social competition inherent in the dynamics of sexual selection and the evolution of sex differences in human life history traits (e.g., adult size, maturational patterns) and in developmental activity (e.g., play).

II. Natural Selection and Life History I begin by describing the basic mechanisms of natural selection along with the basic principles of life history and sexual selection. Then I meld the principles of life history and sexual selection to provide the theoretical foundation for interpreting research on human developmental sex differences. A. NATURALSELECTION The fundamental observations and inferences that led to Darwin's and Wallace's (1858; Darwin, 1859) insights regarding natural selection and evolutionary change are shown in Table I. Of particular importance are individual differences, which largely result as a consequences of sexual reproduction (Hamilton & Zuk, 1982; Williams, 1975). The process of natural selection occurs when heritable variability in a trait, such as age of reproductive maturity, covaries with variability in survival or reproductive outcomes (Price, 1970). As an example, if age of maturation is heritable and early maturing individuals survive to maturity and thus reproduce more successfully than later maturing individuals, then after many generations the mean age of maturation for this population will shift downward (see Reznick & Endler, 1982). The strength of selection pressures can vary such that individual differences in some traits strongly influence the probability of survival or reproduction (e.g., to next breeding season), whereas other traits are only weakly related to or are

Sexual Selection and Human Life History

43

TABLE I Darwin's and Wallace's Observations and Inferences Observations

Inferences

1. All species have such high potential fertility that populations should increase exponentially. 2. Except for minor annual and rare major fluctuations, population size is typically stable. 3. Natural resources are limited and in a stable environment they remain constant. 1. More individuals are born than can be supported by available resources, resulting in competition for those resources that covary with survival prospects. 1. No two individuals are exactly the same; populations have great variability. 2. Much of this variability appears to be related to inheritance that is passed on from parents to offspring. 1. Prospects for survival are not entirely random but covary with inherited characteristics. The relation between these characteristics and differential survival is natural selection. 2. Over generations, natural selection leads to gradual change in the population, that is, microevolution, and production of new species, that is, macroevolution or speciation.

Note: Observations and inferences are based on Darwin and Wallace (1858), Darwin (1859), and Mayr (1982). Although genetics were not yet understood, Darwin inferred that traits were passed on from parent to offspring through, among other things, what was known about the effects of selective breeding (artificial selection) on the emergence of various domestic species.

unrelated to survival or reproductive prospects. If strong selection is maintained across many generations, then heritable variability should be reduced to zero, and it has been for some traits (e.g., all genetically normal humans have two legs, a heritable trait that shows no variability across individuals). However, for a variety of reasons many of the traits that covary with survival and reproductive outcomes show heritable variability and are thus subject to evolutionary change (see Roff, 1992, for a discussion of why heritable variability is maintained). Mousseau and Roff (1987) conducted a comprehensive review of the heritable variability of the morphological, behavioral, physiological, and life history phenotypes (i.e., measurable traits) that covary with survival and reproductive outcomes in wild,

44

David C. Geary

outbred animal populations. The analysis included 1120 heritability estimates-the proportion of variability across individuals that appears to be due to genetic variability--across 75 invertebrate and vertebrate species. Although there was considerable variation--across species, contexts, and phenotypes--in the magnitude of the heritability estimate, their analysis indicated that "significant genetic variance is maintained within most natural populations, even for traits closely affiliated with fitness" (Mousseau & Roff, 1987, p. 188). The median heritability estimates were .26 for life history traits (e.g., age of maturation), .27 for physiological traits (e.g., cardiovascular capacity), .32 for behavioral traits (e.g., mating displays), and .53 for morphological traits (e.g., body size), values that are similar to those found in human populations (Plomin, DeFries, McClearn, & McGuffin, 2001). Kingsolver and colleagues (2001) reviewed field studies of the relation between the types of traits analyzed by Mousseau and Roff (1987) and survival and reproductive outcomes in wild populations (see also Endler, 1986). Across species and triats, the median effect size indicated that being one standard deviation above (e.g., late maturation) or below (e.g., early maturation) the mean was associated with a 16% increase in survival (e.g., surviving to next breeding season) or reproductive (e.g., number of offspring) fitness. If the heritability of any such trait was only .25, "then selection of this magnitude would cause the trait to change by one standard deviation in only 25 generations" (Conner, 2001, p. 216), or in 12-13 generations with a heritability of .50. The basic point is that the principles of natural selection have been empirically evaluated in many species and for many different traits. Many of these traits both show heritable variability and covary with survival and reproductive outcomes, the conditions needed for natural selection and thus evolutionary change to occur (see Table I). B. LIFE HISTORY

As aptly described by Alexander, "lifetimes have evolved to maximize the likelihood of genic survival through reproduction" (Alexander, 1987, p. 65), and the focus of life history research is on the suite of phenotypic traits that defines the species' maturational and reproductive pattern (Charnov, 1993; Roff, 1992). A suite of traits must be considered because of the trade-offs involved in the expression of one phenotype versus another (Williams, 1957). The trade-offs are commonly conceptualized in terms of a competitive allocation of resources (e.g., calories) to somatic effort or reproductive effort, as shown in Figure 1 (Alexander, 1987; Reznick, 1985, 1992; Williams, 1966). Somatic effort is traditionally defined as resources devoted to physical growth and to maintenance of physical systems during development and in adulthood (see West, Brown, & Enquist, 2001), although growth also involves the accumulation, as in increases in body size, of reproductive potential. Reproductive effort is expended during adulthood and is distributed

Sexual Selection and Human Life History

45

Life History Somatic Effort . _

_

.

Infancyand Juvenility Growth

.

Reproductive Effort .

.

_

Developmental Activity

Life Span Maintenance (Survival)

Adult ReproductiveYears Mating

Parenting

Nepotism

Fig. 1. Components of life history. Development activity refers to social, behavioral and cognitive activities during juvenility that promote survival and increase reproductive potential (see Figure 6). Nepotism refers to activities that promote the somatic or reproductive efforts of kin, such as nephews and nieces.

among mating, parenting, and in some species nepotism, that is, investment in kin other than offspring (Emlen, 1995; Hamilton, 1964). In addition to these traditional components of life history, Figure 1 includes developmental activity as a feature of somatic effort during infancy and juvenility. Some developmental activities will promote survival during development (e.g., predator avoidance), whereas others are analogous to the relation between physical growth and the accumulation of reproductive potential. The latter developmental activities result in the refinement of behavioral (e.g., practicing mating displays), cognitive (e.g., birdsong), and physical (e.g., improving cardiovascular capacity) competencies that will later influence reproductive prospects (Geary & Bjorklund, 2000), and presumably result in somatic changes (e.g., modification of neural systems supporting birdsong) during infancy and juvenility. In other words, the results of many developmental activities are incorporated into the developing soma--for instance, distribution of slow and fast muscle fibers as related to physical activity (Byers & Walker, 1995)--and facilitate later reproductive activities. Lifetimes are thus conceptualized as involving the accumulation of reproductive potential--captured by growth and developmental activity during infancy and juvenility--and then the expenditure of this potential on reproductive effort in adulthood (Alexander, 1987). The clearest examples of this view of lifetimes are found in many species of Insecta where distinct morphs are associated with different life history stages. An illustration is provided in Figure 2 for the tomato hornworm moth (Manduca quinquemaculata), where the behavior of the larvae (caterpillars) is focused on somatic effort--to avoid predation and to growmbut the behavior of the adult (moth) is focused on reproductive effort. In fact, the caterpillar morph cannot reproduce and in some species of Insecta the adult morph does not eat; that is, the sole function of the moth or butterfly is to reproduce (Alexander, 1987). Although life history traits will sometimes fall into more than one of the categories shown in Figure 1--sex hormones, for instance, may influence growth as

46

David C. Geary

Fig. 2. Two life history stages of the tomato hornworm moth (Manduca quinquemaculata). To the left is the larval stage during which the caterpillar's behavior is focused on somatic effort, that is, avoiding predation and growth. To the right is the adult stage during which the moth's behavior is focused on reproductive effort. United States Department of Agriculture, public domain illustrations.

well as allocation of reproductive effort into mating or parenting--these categories nonetheless provide a useful heuristic for conceptualizing trade-offs. Within finite lifetimes, trade-offs between the different components of somatic and reproductive effort can be conceptualized in terms of the relative size of the corresponding rectangles in Figure 1. As examples, traits that facilitate predator avoidance (e.g., dull coloration, a feature of somatic effort) may be negatively correlated with mating success (e.g., attracting mates), or traits that facilitate mating activities (e.g., testosterone) may negatively affect health (Folstad & Karter, 1992). Thus the selective advantage for expressing one trait often has an associated cost in terms of other selection pressures or in terms of the expression of other traits (Williams, 1966). Because of these trade-offs, selection pressures do not commonly operate such that a single trait is optimally expressed (e.g., having the brightest possible plumage within physiological constraints). Rather, selection will result in an evolved combination of traits that minimize costs (e.g., predation risks) and enable--within trade-off and ecological (e.g., food) constraints--producing the optimal number of offspring that are likely to survive to adulthood and reproduce themselves (Roff, 1992). A full discussion of the complexity of life history trade-offs is beyond the scope of this treatment (see Charnov, 1993; Roff, 1992; Stearns, 1992), but in the pages that follow I describe some of the most basic of these trade-offs as well as the related issue of phenotypic plasticity.

1. Lifetime Pattern of Reproduction Reproductive activity takes on two general forms: in semelparity all reproductive potential is spent in one breeding episode, but in iteroparity reproductive potential is allocated across more than one breeding episode. Semelparity is a more risky strategy because reproduction during poor ecological conditions could result in extremely high offspring mortality rates, with no opportunity to reproduce under

Sexual Selection and Human Life History

47

more favorable conditions. Semelparity is, however, favored when adult mortality is high and thus the probability of surviving to the next breeding season is low. Under these conditions, individuals that devote minimal resources to somatic effort in adulthood and maximal resources to reproductive effort will produce more offspring than individuals that do not. In contrast, iteroparity is favored when juveniles and adults are likely to survive from one breeding season to the next (e.g., due to low predation risks) and juveniles are unlikely to reproduce successfully (Roff, 1992; Wittenberger, 1979). For these species, the current reproductive effort is balanced against the costs of this effort with respect to survival and future reproductive potential. As a result, during each breeding season iteroparous species invest more in maintenance and less in reproduction than semelparous species (Roff, 1992). A comparison of female Pacific salmon (Salmo oncorhynchus) and female Atlantic salmon (S. salar) illustrates how selection pressures can influence the evolution of a semelparous or iteroparous reproductive strategy. Female Pacific salmon experience intense competition for suitable nesting sites and must guard these sites after depositing their eggs (de Gaudemar, 1998). The intensity of the competition favors expending all resources in one reproductive episode. Females that do not incur the costs of competition will not obtain a suitable nesting site or will have their site destroyed by other females, and thus they will not reproduce at all. The result of this competition has been the evolution of a life history strategy such that resources that could be used for maintenance and survival--females die at the end of the first breeding episode--are expended on behavioral competition for nesting sites and on the development of eggs. The latter results in the production of several-fold more eggs than the iteroparous Atlantic salmon (Roff, 1992). In contrast, competition among females for suitable nesting sites is less intense in Atlantic salmon. In this species, females devote more resources to maintenance and less to reproduction during each breeding season and thus survive to reproduce over many breeding seasons (Roff, 1992). The advantage of distributing reproduction over several seasons is to counter year-to-year fluctuations in predation or other risks (e.g., lack of food) to offspring survival. Although female Atlantic salmon produce fewer eggs during any single season than do female Pacific salmon, the number of viable offspring produced during the reproductive life span of these two species is comparable. There are also variations in life history traits (e.g., age of maturation) across semelparous species and across iteroparous species. These cross-species differences are understandable in terms of differences in selection pressures in each species' specific ecology. As an example, consider Reznick and Endler's (1982) study of the influence of predation on growth and reproductive patterns in iteroparous guppies (Poecilia reticulata). Here, three populations of the same species were studied under three patterns of predation risk: high risk (predators feeding on large adults), medium risk (predators feeding on juveniles), and low risk (few predators). When risk was high, females matured more rapidly and were smaller

48

David C. Geary

as adults, two factors that lowered their risk of being eaten by predators before reproducing. In addition, they allocated more reproductive effort to initial breeding episodes, producing two to three times as many offspring in these breeding episodes than did females in less risky environments. In locales where predation was less severe and adult mortality rates were lower, individuals grew more slowly, attained a larger adult size, and females allocated their reproductive effort over more breeding episodes. Follow-up studies revealed that these differences in life history pattern were due to a combination of genetic differences between these populations and phenotypic plasticity (discussed later, Reznick & Bryga, 1987, 1996; Reznick, Shaw, Rodd, & Shaw, 1997; Rodd, Reznick, & Skolowski, 1997). In sum, Reznick and colleagues' studies empirically demonstrate systematic relations between predation risks and life history and reproductive parameters in three populations of the same species of guppy and thus illustrate the conditions that could ultimately lead to the emergence of different species of guppy with different life history and repoductive traits (Darwin, 1859).

2. Reproductive Costs Reproduction involves costs associated with mating (e.g., finding mates), producing gametes and offspring (e.g., eggs), and for many species parental care (Roff, 1992). Mechanisms underlying the cost~enefit trade-offs involved in reproducing may be genetic or social/environmental, or they may represent a genotype by environment interaction (Reznick, Nunney, & Tessier, 2000). Social costs include those incurred during intrasexual competition over mates and are described later. Genetic trade-offs arise when the same gene or genes affect two or more life history traits (Williams, 1957). In many species, reproducing earlier in life is associated with a shorter life span (Reznick, 1992). The same genes that promote early reproduction have the negative consequence of accelerating the onset of senescence and reducing the life span. Life span is also influenced by more proximal reproductive costs, such as producing eggs, competing for mates, and caring for offspring, which can compromise the physical health and oftentimes the survival prospects of parents (Clutton-Brock, 1991; Steams, 1992). The underlying physiological mechanisms governing these cost/benefit trade-offs are not fully understood, but include the energetic demands of reproduction (e.g., parental care) and associated hormonal changes (Sinervo & Svensson, 1998). For example, the development of male secondary sexual characteristics needed to compete with other males (e.g., antlers) or to attract females (e.g., a bright plumage) requires an increase in testosterone levels which in turn can compromise the immune system and survival prospects of unhealthy males (Folstad & Karter, 1992; Saino & Mr 1994; Saino, Mr & Bolzern, 1995). Similarly, in the female collard flycatcher (Ficedula albicollis) large brood sizes are associated with a reduced production of antibodies for a common parasite; the result is increased infection rate and mortality rate (Nordling, Andersson, Zohari, & Gustafsson, 1998).

Sexual Selection and Human Life History

49

3. Growth and Development All other things being equal, evolution should result in a life history pattern in which females produce many, fast maturing offspring, that have an increased probability of surviving to reproduce (Williams, 1966). The fact that many species do not show this life history pattern indicates that the associated trade-offs are costly. These trade-offs include smaller and less competitive offspring that in turn suffer high mortality rates (Stearns, 1992). Across species of plant, insect, fish, reptile, and mammal, offspring that are larger at time of hatching or birth have increased survival rates due, in part, to decreased predation risk and decreased risk of starvation (Roff, 1992). The trade-off is that females of these species produce fewer offspring than do females of related species that produce many smaller offspring. Thus fast maturation and large numbers of offspring are associated with low-quality offspring (i.e., high mortality risks and low competitiveness). High-quality--larger and more competitiveuoffspring come at a cost of fewer offspring produced during a reproductive life span. Many factors will influence whether a species tends toward a low-quality/high-quantity or high-quality/lowquantity reproductive pattern, including age-specific mortality risks (e.g., through predation), population stability or expansion, and intensity of competition with conspecifics (Mac Arthur & Wilson, 1967; Steams, 1992; Roff, 1992). Species that produce fewer and larger offspring also tend to have slower rates of growth, higher levels of parental care, and longer life spans in comparison to related species that produce smaller but more offspring (Roff, 1992; Shine, 1978, 1989; Stearns, 1992). This life history pattern is more common in iteroparous than in semelparous species and is associated with relatively low juvenile mortality rates and a low probability of reproducing at an early age (Roff, 1992). Low juvenile mortality is related to larger size at hatching or birth as well as to parental protection and provisioning (Clutton-Brock, 1991; Shine, 1978). As described later, a low probability of reproducing at an early age can result from reproductive competition with more mature individuals in the population. In this situation, delayed maturation can improve reproductive prospects through, for instance, an increase in body size. Large body size enables females to give birth to larger and thus more competitive offspring, and for males it facilitates malemale competition in adulthood (Carranza, 1996; Steams, 1992). In some species, developmental activity during the maturational period enables improvements in survival- and reproduction-related behavioral/cognitive competencies. Slow maturation and growth thus allows for the accumulation of more reproductive potential, through physical development and developmental activity, than is possible with faster maturing species. 4. Phenotypic Plasticity Phenotypic plasticity refers to the potential for the modification of survival- and reproduction-related phenotypes in response to social and ecological (e.g., food) conditions, but within genetically based constraints (Roff, 1992). The potential

50

David C. Geary

to modify the expression of life history traits presumably evolved as an adaptation to variability across seasons and generations in the ecologies in which the species evolved. Phenotypic plasticity enables a more optimal expression of life history traits as these relate to survival and reproductive demands in the local ecology. The mechanisms associated with plasticity include hormonal and/or other endocrine responses as well as ecological conditions (e.g., water availability) that affect the physical and behavioral condition of the individual (McNamara & Houston, 1996; Sinervo & Svensson, 1998). Phenotypic plasticity has been empirically demonstrated in a wide range of plant species (Fenner, 1998; Sultan, 2000) as well as in a diversity of other species ranging from plankton to primates (Alberts & Altmann, 1995; McLaren, 1966; McNarnara & Houston, 1996; Miaud, Guyrtant, Elmberg, 1999; Roff, 1992). In all of these species, phenotypic plasticity is expressed within the constraints of norms of reaction (Stearns & Koella, 1986). Norms of reaction represent a genotype whose phenotypic expression varies with ecological conditions, but only within a genetically constrained range. Consider field voles as one example (Microtus agrestis; Ergon, Lambin, & Stenseth, 2001). In this species, populations residing in different locales vary significantly in two life history traits, adult body mass and timing of yearly reproduction. On one hand, if the population differences reflect genetic variance then individuals transplanted from one population to the other will show the body mass and reproductive timing of their natal group. On the other hand, if the population differences reflect variation in local ecologies, such as quality and availability of food, then, in the season following transplantation, body mass and reproductive timing of transplanted individuals should be the same as that of the local community. In fact, the life history traits of transplanted individuals were indistinguishable from those of the local community and differed significantly from those of their natal community. Regardless of natal community, individuals living in richer ecologies developed a higher wintering body size and as a result were able to reproduce earlier. Individuals living in poorer ecologies needed to devote added time to foraging and growth--somatic effortuand thus experienced a delay in the onset of reproductionJreproductive effort. Phenotypic plasticity in growth and reproductive timing has also been demonstrated for many other species, including humans (Steams & Koella, 1986), as well as for many other life history traits (Roff, 1992). For some species, crossgenerational plasticity has been demonstrated, whereby the ecological conditions experienced by the mother influence life history trade-offs in offspring (Hofer, 1987). For example, offspring of nutrient-deprived plants allocate more growth-related resources to root production, whereas offspring of light-deprived plants allocate more resources to leaf production (Sultan, 2000; see also Alekseev and Lampert, 2001, for an analogous mechanism in the crustacean Daphnia). In mammals, maternal condition during pregnancy and during offspring suckling can

Sexual Selection and Human Life History

51

have long-term reproductive consequences. Healthy mothers give birth to heavier offspring and they provide more milk, both of which promote early growth and this, in turn, is associated with larger adult size and higher breeding success (CluttonBrock, 1991). As an example involving social dynamics, testicular maturation and achievement of social dominance are accelerated in male baboons (Papio cynocephalus) borne to high-ranking females, thereby enhancing the males' reproductive prospects (Alberts & Altmann, 1995). 5. Conclusion

From plankton to primates, considerable empirical evidence supports the position that age of maturation, reproductive pattern, number of offspring, extent of parenting, and length of the developmental period are evolved features of the species' life history (Williams, 1966; Roff, 1992; Steams, 1992). The accompanying suite of developmental and reproductive traits essentially involves the respective accumulation and expenditure of reproductive potential, within the constraints imposed by external conditions, such as parasites and predators, and social competition, including sexual selection (Alexander, 1987).

III. Sexual Selection Sexual selection refers to the processes associated with mating competition with members of the same sex and species (intrasexual competition) and the processes associated with choosing mates (intersexual choice; Darwin, 1871). Sexual selection is related to sex differences in hundreds of species and most typically includes male-male competition over access to mates and female choice of mating partners (Andersson, 1994). As I describe in the first part of this section, the dynamics of sexual selection turn on the degree to which members of each sex allocate their reproductive effort to competing for mates or investing in parenting. In the second and third parts of this section, I illustrate the evolutionary influences of intrasexual competition and intersexual choice, respectively. A. MATINGOR PARENTING? As shown in Figure 1, reproductive effort is distributed between mating (e.g., time spent searching for mates), parenting, and occasionally nepotism. Nepotism is less central to the later discussion of human life history and is not considered further (see Emlen, 1995); this is not to say that humans do not engage in significant levels of nepotism--they do (see Geary & Flinn, 2001; Pasternak, Ember, & Ember, 1997). The distribution of reproductive effort across mating and parenting is, however, central to the later discussion, and it turns on the extent of each sexes' parental effort or parental investment (Trivers, 1972; Williams, 1966).

52

David C. Geary

Parental investment is any cost (e.g., time) associated with raising offspring that reduces the parents' ability to produce or invest in other offspring (Trivets, 1974). Given that some level of parental investment (even if it only involves producing eggs) is necessary for the reproduction of both parents, the nature of the investment provided by females and males creates the basic dynamics of sexual reproduction and sexual selection. If one sex provides more than his or her share of parental investment, then members of that sex become an important reproductive resource for members of the opposite sex (Dawkins, 1989; Trivers, 1972). The reproductive success of members of the lesser investing sex is more strongly influenced by the number of mates that can be found than by investing in the well-being of individual offspring, whereas the reproductive success of members of the more highly investing sex is more strongly influenced, in most cases, by investment in offspring than in finding mates. In most species, the sexes differ in the degree to which the reproductive effort is allocated to competition for access to mates or to parental investment (Andersson, 1994; Trivers, 1972; Williams, 1966). These differences, in turn, are related to the potential rate of reproduction and to social and ecological influences on mating opportunities, in particular, the operational sex ratio (OSR; Clutton-Brock & Vincent, 1991; Emlen & Oring, 1977; Krebs & Davies, 1993). Reproductive rates and the OSR are related but described separately.

1. Reproductive Rates A sex difference in potential rate of reproduction can create a sex difference in relative emphasis on mating or on parenting. Most generally, the sex with the higher potential rate of reproduction invests more in mating effort than in parental effort, whereas the sex with the lower rate of reproduction invests more in parental effort than in mating effort (Clutton-Brock & Vincent, 1991). This pattern arises because members of the sex with the higher potential rate of reproduction can rejoin the mating pool more quickly than can members of the opposite sex. Under these conditions, individuals of the sex with the faster rate of reproduction will typically have a higher lifetime reproductive success if they rejoin the mating pool and compete for mates than if they parent (Parker & Simmons, 1996). For species with internal gestation and obligatory postpartum female care (e.g., suckling in mammals), the rate at which females can produce offspring is considerably lower than the potential rate of reproduction of conspecific males (CluttonBrock, 1991). In addition, internal gestation and the need for postnatal care results in a strong bias in mammalian females toward parental investment and results in a sex difference in the benefits of seeking additional mates (Trivers, 1972). Males can benefit, reproductively, from seeking and obtaining additional mates, whereas females cannot. In other words, males that compete for additional mates typically have more offspring than do males that do not compete and instead invest in parenting. Thus, the sex difference in reproductive rate, combined with offspring that can

Sexual Selection and Human Life History

53

be effectively raised by the female, creates the potential for a large female-male difference in the mix of mating and parenting, and this difference is realized in 95-97% of mammalian species (Clutton-Brock, 1989). In these species, females can effectively provide the majority of parental care and do so. Female care, in turn, frees males to invest in mating effort, which typically takes the form of male-male competition over access to mates or for control of the resources (e.g., territory) that females need to raise their offspring.

2. Operational Sex Ratio The OSR is defined as the ratio of sexually active males to sexually active females in a given breeding population at a given point in time and is related to the rate of reproduction (Emlen & Oring, 1977). In a population where the number of sexually mature females equals the number of sexually mature males--an actual sex ratio of l : l - - a n y sex difference in the rate of reproduction will skew the OSR. As noted, mammalian males have a faster potential rate of reproduction than conspecific females, which typically results in more sexually receptive males than sexually receptive females in most populations. This biased OSR creates the conditions that lead to intense male-male competition over access to a limited number of potential mates. Although these patterns are most evident in mammals, they are also found in many species of bird, fish, and reptile (Andersson, 1994). And they are not limited to males: When females have a faster rate of reproduction than males (e.g., when males care for eggs), female-female competition is often more salient than male-male competition (e.g., Reynolds, 1987). The sex difference in potential reproductive rate and a skewed OSR appear to be the ultimate sources of the male focus on mating effort and the female focus on parental effort in the vast majority of mammalian species (Emlen & Oring, 1977; Parker & Simmons, 1996). The biology of internal gestation and suckling are not the only factors that influence the potential rate of reproduction and the OSR in mammals; social and ecological factors are sometimes important as well. As an example, male callitrichid monkeys (Callithrix) have a higher potential rate of reproduction than conspecific females do. However, shared territorial defense, female-on-female aggression that drives away the males' potential mating partners, and other factors negate this physiologically based sex difference and result in a more balanced OSR, monogamy, and high levels of paternal investment (see Dunbar, 1995). B. INTRASEXUAL COMPETITION

Intrasexual competition over mates, whether male-male competition or femalefemale competition, will result in the evolutionary emergence of sex differences for those traits that facilitate this competition (Andersson, 1994; Darwin, 1871). Studies of intrasexual competition have revealed that the associated sex differences

34

David C. Geary

can be physical, behavioral, or cognitive (including neural) and typically only affect those features actually involved in the competition (Andersson, 1994; Geary, 1998). One of the more common expressions of intrasexual competition involves physical threats and fights over access to mates or for control of the territory that members of the opposite sex need to raise offspring (e.g., nesting spots). A common result is that physically larger, healthier, and more aggressive individuals (typically males) monopolize the reproductive potential of members of the opposite sex (typically females). The accompanying individual differences in reproductive success--some individuals have many offspring, others have few or none--result in the evolution of sex differences in physical size and aggressiveness. The polygynous ruff (Machetes pugnax) provides one example of such competition among males. As described by Darwin, males are considerably larger and more aggressive than females and physically compete for sexual access to females. These physical and behavioral sex differences evolved in the ruff, and many other species, because of the reproductive advantages associated with larger size and pugnacity in males (Darwin, 1871). Sometimes the competition is more behavioral or cognitive (e.g., spatial cognition, as in searching for mates) than physical (Gaulin & Fitzgerald, 1986, 1989; Gilliard, 1969). In these situations, behavioral and cognitive traits that facilitate intrasexual competition will evolve in the same way that physical traits evolve (see Geary, 1998, for elaboration). When males parent (e.g., incubate eggs), females may compete more intensely for mates than males; that is, physical female-female competition is more intense than male-male competition. In these species, females are larger and more aggressive than males (see Reynolds, 1987). c. INTERSEXUALCHOICE The sex that invests more in parenting tends to be more choosy with regard to mating partners than the other sex (Trivers, 1972). Because females tend to invest more in parenting than males, female choice is predicted to be and is more common than male choice. Male choice is predicted for species with paternal investment, although this prediction has not been as thoroughly tested as female choice. In any case, female choice has been studied most extensively in birds, although it is also evident in insects, fish, reptiles, and mammals, including humans (Andersson, 1994; Buss, 1994). Several examples of male traits that have been shaped by female choice are shown in Figures 3 and 4; in some species these traits may also be involved in male-male competition, as in dominance displays. These traits often involve elaborate physical displays, as in the crest along the back and tail of the male crested newt (Triturus cristatus; Figure 3), the dorsal fin of the male dragonet (Callionymus lyra; Figure 3), and the comb of the male hoopoe (Upupa epops; Figure 4). In many species, males are often more elaborately colored than females. The comb of the male hoopoe is a bright orange, and the male dragonet has a brilliant yellow body of varying shades, whereas the female dragonet is a "dingy reddish-brown" (Darwin, 1871, Vol. II, p. 8).

Sexual Selection and Human Life History

55

Fig. 3. Indicators of male fitness shaped by female choice for selected species of amphibian and fish. To the left are male (top) and female (bottom) Triturus cristatus (from The Descent of Man, and Selection in Relation to Sex, Part II, p. 24, by C. Darwin, 1871, London: John Murray). To the right are male (top) and female (bottom) Callionymus lyra (from The Descent of Man, and Selection in Relation to Sex, Part II, p. 8, by C. Darwin, 1871, London: John Murray).

Traits such as those shown in Figure 3 are indicators of the physical, genetic, or behavioral fitness (e.g., ability to provide) of the male. These traits are honest indicators of male fitness, as they commonly are not expressed in unfit males (Zahavi, 1975). As an example, in some species of bird the coloration of male plumage covaries with physical health, in particular, resistance to infection by local

Fig. 4. The size and coloration of the comb of the male hoopoe (Upupa epops) are indicators of male fitness shaped by female choice (from A History of British Birds, p. 50, by F. O. Morris, 1891, London: Nimmo ).

56

David C. Geary

parasites (see Hamilton & Zuk, 1982). In red jungle fowl chicks (Gallus gallus), males infected with a parasitic worm (Ascaridia galli) grow more slowly than their healthy peers, and, in adulthood, their sexually selected characteristics are more adversely affected by the infection than are other physical characteristics. For instance, the comb of affected males is smaller and duller than that of unaffected males but many other physical characteristics (e.g., ankle length) do not differ across these groups. Furthermore, females substantially prefer unaffected males (by two to one), and preference is related to sexually selected features, such as comb length, but not other features, such as ankle length (Sheldon, Meril~i, Qvarnstrrm, Gustafsson, & Ellegren, 1997; Zuk, Thornhill, & Ligon, 1990). Related studies have demonstrated that unfit males cannot tolerate the hormonal changes needed for the expression of sexually selected traits. Experimentally increasing male testosterone levels to induce the expression of secondary sexual characteristics results in increased mortality in unhealthy males, perhaps due to suppression of immune functions (Folstad & Karter, 1992). In series of field experiments, Siano and colleagues assessed the effect of testosterone implants on mortality rates in the barn swallow (Hirundo rustica; Saino et al., 1995; Saino, Bolzern, & Moiler, 1997). In this species, female choice is influenced by the length and symmetry of the male's tail features (MOiler, 1994). Testosterone implantation suppresses the immune system in males with shorter tail feathers more severely than in males with longer tail feathers and results in increased parasite loads and higher mortality rates in shorter tailed than in longer tailed males. The pattern indicates that males with long tail feathers can support high testosterone levels--and thus more effectively compete for mates--without compromising their immune system, suggesting that their immune system is well adapted to local parasites or that they are in better general physical condition than are males with short tail feathers. The pattern also illustrates an important life history trade-off for shorter tailed males. Resources (e.g., calories) that could be used for the development of a sexually selected trait must be diverted to immune functions. The cost is an inability to attract mates during the current mating season, and the benefit is increased survival prospects and thus an opportunity to mate in subsequent seasons.

IV. Life History and Sexual Selection Sexual selection and life history have been linked theoretically and empirically, although the extent of the interrelation is not fully understood (Andersson, 1994). Predation risks, for example, may indirectly influence the opportunities for intrasexual competition or intersexual choice to operate and thus evolve (Partridge & Endler, 1987; Winemiller, 1992). More relevant to the current treatment is the prediction that social dynamics inherent in intrasexual competition and intersexual choice can influence and be influenced by life history traits (Hrglund &

Sexual Selection and Human Life History

57

Sheldon, 1998; Kokko, 1997; Svensson & Sheldon, 1998). In fact, competition among members of the same species should, in theory, favor life history traits that include fewer, more competitive offspring (Mac Arthur & Wilson, 1967), which favors the evolution of iteroparity, a longer developmental period, higher levels of parental investment, and other traits that support social competition (Roff, 1992). Empirically, many of the hormonal mechanisms that influence the expression of life history traits (e.g., maturational timing) and trade-offs (e.g., mortality risks) also influence expression of secondary sexual characteristics involved in intrasexual competition and intersexual choice (Sinervo & Svensson, 1998). The full extent of the relation between sexual selection and life history remains to be determined. For now, in the parts that follow I illustrate how intrasexual competition and intersexual choice appear to relate to the evolution and phenotypic expression of life history traits and trade-offs. In the final part, I address the issue of phenotypic plasticity. A. INTRASEXUAL COMPETITION

Studies of the relation between intrasexual competition and life history have focused largely on males (Stearns, 1992), presumably because male-male competition is more common than female-female competition (Darwin, 1871; Andersson, 1994). 1 The relation between male-male competition and a few life history traits (e.g., age of maturation) have been studied in a variety of mammalian and bird species (e.g., Clinton & Le Boeuf, 1993; Harvey & Clutton-Brock, 1985; McElligott & Hayden, 2000; Rohwer, Fretwell, & Niles, 1980; Wiley, 1974) as well as in some other species (Stamps, 1995). I first illustrate the relation between physical male-male competition and sex differences in life history traits and then consider the relation between behavioral competition and sex differences.

1. Physical Competition Males of many species of insect, reptile, fish, bird, and mammal show little or no parental investment and compete intensely for access to females (Andersson, 1994; Clutton-Brock, 1989; Darwin, 1871). One result is that only a minority of males reproduce, thereby creating strong selection pressures for the evolution of traits that support competitive ability (e.g., Clinton & Le Boeuf, 1993; McElligott & Hayden, 2000; Plavcan & van Schaik, 1997a). Among these traits are physical size and aggressiveness such that larger, more aggressive males are 1When it occurs, female-female competition should relate to life history traits in many of the same ways as male-male competition. Social dynamics in polyandrous shorebirds, for instance, include intense female-female competition for access to males to brood their eggs. Although the issues have not been thoroughly studied, females of these species show some of the same life history patterns common for males of other species in which male-male competition is intense (Reynolds, 1987; Reynolds & Szrkely, 1997).

58

David C. Geary

typically more competitive than smaller, less aggressive males. In theory, these conditions could result in the evolution of growth rates such that males grow faster than females and achieve large size at progressively younger ages. However, the calories needed to achieve large size and the lost opportunity to practice fighting before reproductive maturity place formidable constraints on the evolution of such a life history pattern. A more common pattern is for males to grow more slowly and mature later than females and to engage in play fighting during juvenility (Smith, 1982; Stearns, 1992). As an example, the mating dynamics of primates are consistently related to sex differences in maturational patterns (e.g., duration, growth spurt) and physical size (Leigh, 1995). Polygynous species with physical male-male competition are characterized by consistent sex differences, whereby males grow more slowly and evidence both a longer period of rapid growth (i.e., the growth spurt), and a longer overall developmental period than females. The result is larger males than females. In these species, both males and females are physically aggressive, but male-onmale physical aggression is related to competition for mates and is more severe and deadly than female-on-female aggression, which is related to competition for food (Smuts, 1987). Further evidence that the sex differences in life history pattern are related to male-male competition comes from comparisons of evolutionarily related (i.e., having a recent common ancestor) monogamous and polygynous species. Intrasexual competition is less intense in monogamous species and thus the selective advantages for physical size and aggressiveness are considerably less relative to polygynous species (Clutton-Brock, Harvey, & Rudder, 1977). Among monogamous species of primate, the sexes rarely differ in adult size or maturational pattern (Leigh, 1995). For mammalian species in which physical male-male competition is found, the development period of males can range from moderately longer (e.g., 2.8 vs 3.5 years in the patas monkey, Erythrocebus patas) to more than twice as long as that of females (e.g., 3.0 vs 8.0 years in the northern elephant seal, Mirounga angustirostris; Le Boeuf & Reiter, 1988; Harvey & Clutton-Brock, 1985; Stearns, 1992). Males may weigh slightly more than females (e.g., 19% heavier in colobus monkeys, Colobus angolensis) or can weigh more than double that of females (e.g., 120% heavier in mandrills, Mandrillus sphinx, another monkey; Harvey & Clutton-Brock, 1985). Intrasexual competition and accompanying reductions in parental investment (Trivers, 1972) are also related to average length of the life span (Allman, Rosin, Kumar, & Hasenstaub, 1998). The lesser investing sex shows more intense intrasexual competition and has a shorter life span on average than does the more highly investing sex, whether the latter is female or male. The sex difference in life span appears to be a consequence of the higher nutritional demands needed to grow larger, the injuries associated with male-male competition, and the immunosuppressive effects of testosterone (Clinton & Le Boeuf, 1993;

Sexual Selection and Human Life History

59

Clutton-Brock, Albon, & Guinness, 1985; Folstad & Karter, 1992). The advantage of physical size also creates a selective advantage for larger offspring and an accompanying increase in the size of females, but with the trade-off of fewer offspring (Carranza, 1996; Roff, 1992). The overall pattern is consistent with predictions noted earlier regarding social competition and life history evolution (Mac Arthur & Wilson, 1967; Roff, 1992). The features of primate life history and particularly that of humans (described later) and related species (e.g., chimpanzees, Pan troglodytes) appear to be especially in keeping with this hypothesis. Across species of primate, larger and fewer offspring are associated with the predicted life history patterns of longer interbirth intervals, higher levels of maternal investment, larger brains, longer developmental periods, and longer maximum life spans (Allman, McLaughlin, & Hakeem, 1993; Harvey & Clutton-Brock, 1985). Across these species, the length of the developmental period and brain size covary positively with the species' social complexity, the intensity of intrasexual competition, and with foraging complexity (Allman et al., 1993; Dunbar, 1993; Joffe, 1997; Sawaguchi, 1997), suggesting that a long maturational period is not simply about producing a larger body size, at least in primates. Presumably, a long developmental period and a large brain enable the practice and refinement of sociocompetitive and foraging skills~accumulation of reproductive potential~before engaging in actual (potentially life threatening) competition and unsupported (by parents) food acquisition.

2. Behavioral Competition Intrasexual competition with a strong behavioral component can result in alternative reproductive strategies and life histories for one or both sexes or in an exaggeration of the behavioral and associated life history traits. The different reproductive strategies and life histories of smaller jack and larger hooknose male salmon (Oncorhynchus kisutch) provide a clear example of the former (Gross, 1985). Hooknose males mature later and compete physically for access to eggs laid by females, whereas jack salmon are specialized to hide among rocks and furtively spawn while hooknose males are fighting. The smaller and earlier maturing jacks are just as reproductively successful, on average, as larger hooknose males, and thus early maturity and furtive mating represents a successful life history strategy for males of this species (Gross, 1985). Studies of bowerbirds provide some the best examples of how intrasexual competition and intersexual choice can involve behavioral competition and result in the evolution of behavioral sex differences (Gilliard, 1969). In most of these species, the principal focus of competition and choice is the bower, a structure made of tree boughs and vines shown in Figure 5 (Darwin, 1871). A female bowerbird's choice of mating partners is strongly influenced by the complexity and symmetry of the male's bower as well as by the number of decorations around the bower. Males thus compete with one another through bower building and through the destruction of

60

David C. Geary

Fig. 5. Bower building and behavioral male-male competition in the bowerbird (Chlamydera maculata) (from The Descent of Man, and Selection in Relation to Sex, Part II, p. 70, by C. Darwin, 1871, London: John Murray).

their competitors' bowers (Borgia, 1985a, 1985b; Collis & Borgia, 1992), although courtship displays, calls, and physical fights among males are also components of male-male competition and female choice (Borgia & Coleman, 2000). As with species in which physical male-male competition is prevalent, sex differences in the life history traits of one well-studied bowerbird, the satin bowerbird (Ptilonorhynchus violaceus), are evident, including differences in growth patterns and developmental activity. Female satin bowerbirds begin to reproduce at 2 years of age, whereas males do not produce sperm until they are 5 years of age and do not achieve an adult-male plumage until they are 6- or 7-years-old (Vellenga, 1980). Even then, most males that hold bowers do not mate until 10 years of age, if they mate at all (Borgia, personal communication, August 24, 2001). During development, young males watch older males at their bower and imitate bower building and courtship displays when the older male leaves the bower (e.g., to feed; Collis & Borgia, 1992). Young males also engage in play fighting, which provides the experience needed for dominance-related encounters in adulthood. Although the degree to which bower building is genetically or experientially based is not yet certain, the delayed maturation of male satin bowerbirds almost

Sexual Selection and Human Life History

61

certainly provides an opportunity to practice and refine the bower construction and physical competition skills that will be needed in adulthood. In this circumstance, the later reproductive advantage of bower building and other competition-related skills will--when combined with low juvenile mortality risks--provide a selective advantage for delayed maturation in males. In other words, delayed maturation enables males through developmental activities to accumulate the reproductive potential needed to compete with other males and to attract female mates in adulthood. B. INTERSEXUALCHOICE The relation between intersexual choice--typically female choice--and sex differences in life history traits has received less attention than the relation between intrasexual competition and life history. One difficulty is that many traits that influence female choice are also related to male-male competition, as in dominance displays (e.g., birdsong often has this dual function; Borgia & Coleman, 2000). An important window into the likely influence of female choice on the evolution of life history traits in males comes from studies of lekking species (Htglund & Alatalo, 1995; Wiley, 1974). Leks can be areas in which males gather together during the mating season to strut, display plumage, or engage in other activities that function to attract mating partners, or they can be more dispersed areas in which single males display. In both situations, females will survey a number of males and will then mate with one or a few of them. The result is a minority of males copulate with most of the females that visit the lek. The satin bowerbird provides one example of a lekking species. The delayed maturation of male bowerbirds and the sex difference in developmental activities (i.e., bower building) are likely the evolutionary result of male-male competition and female choice, although the relative contribution of these different components of sexual selection are not known. Whatever the relative contribution, female choice has contributed to the evolution of sex differences in the life history of bowerbirds. Peacocks (Pavo cristatus) provide another example of a lekking species but one in which male-male competition is minimal and female choice largely determines which males reproduce and which do not (Htglund & Alatalo, 1995; Petrie, 1994; Petrie, Halliday, & Sanders, 1991). Unlike peahens, peacocks develop large tail trains with varying numbers of eyespots. Males display unfolded trains to females, and females choose mates on the basis of the length of the train and the number of eyespots (Petrie et al., 1991). Train length and number of eyespots are reliable predictors of the growth and survival rate of the males' offspring and are thus honest (i.e., cannot be faked; Zahavi, 1975) indicators of the genetic quality of the male (Petrie, 1994). As with bowerbirds, sex differences in life history traits are found. Peahens begin to reproduce at 2 years of age, whereas males do not develop their full trains until 3 years of age and do not establish a lekking display site until 4 years

62

David C. Geary

of age (Manning, 1989; Petrie, personsal communication, August 28,2001). Some males successfully mate at age 4, whereas other males do not mate until later years and some not at all. In this species, the development of the sexually selected traits that females use in mate choice decisions has resulted in accompanying changes in the life history of males, including a lengthening of the developmental period and an increase in size at maturity (for related discussion see Brooks & Kemp, 2001; Wiley, 1974). C. PHENOTYPIC PLASTICITY

Phenotypic plasticity is a common feature of life history traits. That is, the ontogenetic expression of life history traits is influenced by ecological conditions within the constraints of reaction norms (Stearns, 1992). Plasticity of life history traits is often conceived in terms of nonsocial factors, such as predation, food availability, or rainfall (e.g., Stearns & Koella, 1986; Sultan, 2000). Life history traits that have been shaped by sexual selection are also likely to show phenotypic plasticity, but the ontogenetic expression of these traits should be more strongly influenced by social competition and dynamics than by nonsocial conditions (Rohwer et al., 1980; Selander, 1965). In fact, the relation between many ecological variables, such as food availability, and the plastic expression of life history traits may be moderated by social competition. The age of reproductive maturity in many species is influenced by access to high-quality foods (Stearns & Koella, 1986), which in turn is often influenced by social competition. In many species of primate, coalitions of related females compete for access to high-quality food sources, such as fruit trees (Wrangham, 1980). Socially dominant coalitions gain access to these foods, and the combination of dominance and better nutrition is related to a number of indices of reproductive maturity and success, including age at first conception and number of offspring surviving to maturity (Silk, 1993). The dynamics of sexual selection also appear to influence the ontogenetic expression of many life history traits. As with bowerbirds, males are often physiologically able to reproduce many years before they actually reproduce (Wiley, 1974). The reproductive delay can be due to competition with older and more dominant males or a female preference for older males (Brooks & Kemp, 2001; Selander, 1965). In many lekking species, for instance, males must establish and defend a display territory, and older males typically have an advantage over younger males in competition for these sites (Wiley, 1974). In some species, the physiological stress associated with male-male competition and the behavioral subordination to more dominant males, as well as other forms of social competition, can delay physiological maturation or reduce reproductive potential (e.g., reduce size of testes) in adult males (Dixson, Bossi, & Wickings, 1993; Walter & Dittami, 1997; Waiters & Seyfarth, 1987). As noted earlier, testicular maturation in male baboons is related to the social rank of their mother (Alberts & Altmann, 1995).

Sexual Selection and Human Life History

63

In short, the presence of dominant males often results in the inhibition of competitive behavior in younger males and can delay the onset of reproductive maturity and inhibit the expression of traits associated with female choice. Note also that nonsocial influences on the expression of life history traits, such as food availability in the absence of social competition, can strongly influence the later ability to compete for mates or attract mating partners (e.g., Nowicki, Peters, & Podos, 1998).

V. Human Developmental Sex Differences A comprehensive understanding of human developmental sex differences can only be achieved through consideration of the evolution and proximate functions of life history traits (Bjorklund & Pellegrini, 2002; Bogin, 1999; Geary & Bjorklund, 2000; Hill & Kaplan, 1999; Kaplan, Hill, Lancaster, & Hurtado, 2000; Kenrick & Luce, 2000; Lancaster & Lancaster, 1987). Current models of human life history focus on the complexity of the foraging demands in traditional societies (Kaplan et al., 2000) or on social competition (Alexander, 1987, 1989; Geary & Flinn, 2001). In fact, both forms of selective pressure are likely to have been important. In many traditional societies, men provide--through hunting--the majority of calories and protein consumed by their social group (Ember, 1978; Kaplan et al., 2000). The acquisition of hunting skills requires many years of practice and experience, which has been interpreted as a selective pressure for an increase in the length of the developmental period during which hunting (and foraging for females) skills are practiced (Kaplan et al., 2000). The proceeds of hunts are also related to male-male competition (social status) and female choice of mating partners (Hill & Hurtado, 1996), components of sexual selection. Moreover, the extraordinary hunting and foraging skills of humans in traditional societies, and presumably during human evolution, contribute to the ecological dominance of most human groups, which in turn significantly alters the pattern of selective pressure, as noted by Alexander (1989, p. 458): The ecological dominance of evolving humans diminishedthe effects of "extrinsic" forces of natural selection such that within-speciescompetitionbecame the principle "hostileforce of nature" guidingthe long-termevolutionof behavioralcapacities,traits, and tendencies. With the achievement of ecological dominance, natural selection becomes a struggle with other human beings for access to and control of the social (e.g., competition for mates), biological (e.g., food), and physical (e.g., territory) resources that covary with survival and reproductive outcomes (Geary, 1998). Elaborating on Alexander (1989), Geary and Flinn (2001) argued that the resulting social competition was the primary selective pressure driving the coevolution of a suite of

David C. Geary

64

human characteristics, including several life history traits described in Table II. In this view, the hunting and foraging demands described by Kaplan et al. (2000; Hill & Kaplan, 1999) are intertwined within a broader suite of social-competitive selection pressures, including intrasexual competition and intersexual choice. My goal in the remainder of this chapter is to extend the social competition model to provide a theoretical frame for understanding sex differences in human life history and for organizing psychological research on sex differences in developmental activities, such as play patterns; implications for understanding sex differences in human cognition and cognitive development are described elsewhere (Geary, 1998, 2002). In the first two parts, I provide sketches of sexual selection in humans and sex differences in life history traits, respectively. In the third part, I describe the framework for and research on developmental activities. The behavioral and social activities described in all parts are more strongly influenced by inherent and implicit processes than by conscious choice. In other words, as with other species, children and adults are inherently biased to engage in activities that covaried with survival and reproductive outcomes during human evolution, whether or not they are consciously aware of the proximate function (e.g., to attract mates) of the activities (Geary, 1998).

TABLE II Unique and Unusual Traits Related to Social Competition and Human Life History

L Large brain and complex social competencies 1. The human neocortex is 35-60% larger than expected for a primate of the same overall body and brain size (Rilling & Insel, 1999). 2. The neocortex apparently is larger than that of other primates in those areas that support social competencies that are unique to humans (Rilling & Insel, 1999), that is, theory of mind (Baron-Cohen, 1995) and language (Pinker, 1994). II. High levels of paternal investment 1. Paternal investment is only evident in 3-5% of mammalian species (Clutton-Brock, 1989). 2. Even for these species, humans are unique in that paternal investment occurs in a social context of large multimale-multifemale communities and where most adult members of these communities reproduce (Alexander, 1990; Geary, 2000). IlL Long developmental period and adult life span 1. Relative to other mammals and primates, children have a very long developmental period characterized by slow development during middle childhood and high dependency on adult caregiving (Bogin, 1999). 2. Relative to other great apes, humans have a very long adult life span, and low juvenile and adult mortality rates (Allman et al., 1993; Hill et al., 2001). IV. Menopause 1. Menopause may enable women to heavily invest in their later born children or in grandchildren (Hawkes et al., 1998; Williams, 1957). Adapted from Geary and Flinn (2001, p.6).

Sexual Selection and Human Life History

65

A. SEXUALSELECTIONDURINGHUMANEVOLUTION Unlike most other mammals, both women and men invest in parenting, although men on averge do not invest as much as women (Geary, 2000). In natural environments, paternal investment supports the intensive parenting and long developmental period that is characteristic of humans and appears to allow parents to successfully raise more children than would otherwise be the case (see Table II; Geary & Flinn, 2001; Lancaster & Lancaster, 1987; Kaplan et al., 2000). In addition, paternal investment complicates the dynamics of sexual selection, resulting in male choice and female-female competition, on top of the standard components of male-male competition and female choice. Intensive parenting and the complexity of sexual selection create six interrelated classes of social relationship that involve conflict and competition, as described in Table III. These forms of social relationship capture the within-species competition noted by Alexander (1989) and are likely to have been (and continue to be) important forces in human cognitive, brain, and social evolution, as well as potent influences on evolutionary change in human life history and developmental activity. Extended spousal and parent-child relationships create conditions that favor the evolution of complex sociocognitive competencies, as both forms of relationship involve not only cooperation but also manipulation and deception as related to attempts to gain access to resources that covary with survival and reproductive outcomes (e.g., Geary, 2000; Trivers, 1974). Further discussion of these relationships is beyond the scope of this article, but here I summarize work on intrasexual competition and intersexual choice. 1. Intrasexual Competition

On the basis of the fossil record, males apparently were larger than females throughout hominid evolution, suggesting physical male-male competition and a sex difference in parental investment (e.g., McHenry, 1991). The fossil record does not provide insights into the intensity or form of this male-male competition (Plavcan & van Schaik, 1997b), but patterns throughout human history and in extant cultures suggest competition involving male coalitions and one-on-one competition for status and dominance within coalitions (Chagnon, 1988; Geary, 1998; Geary & Flinn, 2002; Horowitz, 2001; Wrangham, 1999). To illustrate, for forest dwelling Ache (hunter-gatherer society, Paraguay) coalitional warfare with non-Ache accounted for 36% of all adult male deaths, and an additional 8% of men died during status-oriented club fights with other Ache men (Hill & Hurtado, 1996). This pattern of coalitional and within-coalition male-male competition is common in traditional societies (Keeley, 1996) and has evolutionarily significant survival and reproductive consequences. Dominant men in dominant coalitions typically have more wives and more surviving children than do other men (e.g., Chagnon, 1988). In addition to facilitating male-male competition, men's coalitions also enable them to achieve ecological dominance, that is, to efficiently extract resources

66

David C. Geary

TABLE III Forms of Social Conflict and Competition Intrasexual competition L Male-male competition 1. In traditional societies, men form coalitions that compete for control of mating dynamics (e.g., exchange of brides) and for control of the resources that covary with survival and reproductive outcomes in the local ecology (Chagnon, 1988; Geary, 1998). Men also form dominance hierarchies within the in-group coalitions and compete for position (and thus influence) in the hierarchy. Competition is often physical and deadly (Keeley, 1996).

II. Female-female competition 1. Women compete for access to resources, including access to resource-holding or socially influential men. Relative to men, this competition is less physical (Campbell, 1999) and involves subtle manipulation of social relationships, with the goal of organizing these relationships to maximize the woman's access to resources that covary with survival and reproductive outcomes in the local ecology (Geary, 2002). Intersexual choice I. Male choice 1. Paternal investment leads to the prediction that men will be selective in their mate choices (Trivers, 1972), and this is the case. Men's mate choices are influenced by fertility cues (e.g., age) as well as by indicators of women's social and maternal competence (Geary, 1998).

II. Female choice 1. Women's mate choices are influenced by men's social and parental competence. More so than men, women also focus on men's social status, including material resources, social influence, and cues to their ability to acquire and maintain these resources (Buss, 1989, 1994). Women are also sensitive to their ability to influence potential mates and thus gain access to their resources. Family conflict L Spousal 1. Spouses, of course, cooperate in raising children, but extended maternal and paternal investment also results in strong potential for conflicts of interest (Kaplan et al., 2000; Svensson & Sheldon, 1998). Conflicts are predicted to center on (1) extent of maternal versus paternal investment, (2) resource control (e.g., spending on children or status-oriented objects), and (3) marital fidelity.

IL Parent-offspring and sibling 1. Across species, conflicts of interest are endemic to parent-offspring relationships (Trivers, 1974). Parents, of course, invest time and resources to promote the well-being of offspring, but offspring often press for additional resources, sometimes with accompanying morbidity and mortality costs to parents (Westendorp & Kirkwood, 1998). The long developmental period of humans results in an extended parent-child relationship and thus the potential for extended conflicts over parental allocation of resources. In the context of these relationships, children are predicted to attempt to secure from parents (and to a lesser degree kin) resources that facilitate (1) growth and maintenance and (2) reproductive potential. 2. Siblings will also compete for parental resources.

from the environment through hunting and to maintain a comparatively risk-free (e.g., low predation risk) territory for their social group (Tiger, 1969). These male coalitions consist of related men and define the basic social structure of the group. Women tend to emigrate into the group of their husband, although in many societies they maintain ties to their kin (Pasternak et al., 1997; Seielstad,

Sexual Selection and Human Life History

67

Minch, & Cavalli-Sforza, 1998). In any case, male coalitions provide protection from other coalitions, and provision their wives and children, although women also provide food through gathering (Ember, 1978; Geary & Flinn, 2001; Kaplan et al., 2000). Within these groups, women form a small network of relationships with other women and provide social and emotional support to one another (Taylor et al., 2000). Women also work to organize social relationships within the wider group to divert additional resources to themselves and to their children (Geary, 2002; Geary & Flinn, 2002). One result is conflicts of interest with other women. The resulting female-female competition focuses on disrupting the social network of their female competitors (often co-wives), as well as competing for the attention and resources of men within the group. This form of interpersonal competition has been termed relational aggression by Crick and her colleagues (Crick, Casas, & Mosher, 1997). The function of male-male (achieving status and cultural success; Irons, 1979) and female-female (control of interpersonal relationships; Geary, 2002) competition is the same from one society or historical period to the next: to gain access to and control of the resources that covary with survival and reproductive outcomes in the local group and ecology. However, the form of intrasexual competition shows considerable phenotypic plasticity, varying with cultural mores (e.g., prohibitions against polygyny) and social conditions, such as the operational sex ratio (Flinn & Low, 1986; Geary, 1998, 1999; Low, 1989). In some societies, men achieve cultural success by killing other men, whereas in other societies they achieve cultural success by obtaining an education and securing a high-income job. The underlying motive is always the same, however: to achieve dominance over other men and thus increase access to culturally important resources and, through this, reproductive opportunity (P6russe, 1993). Patterns of intrasexual choice also show phenotypic plasticity, as illustrated by cross-cultural variability in women's preference for husbands who have or have not killed other men (preference varies with whether or not killing confers social status; Geary, 1998). 2. Intersexual Choice

Patterns of intersexual choice can influence the form and intensity of intrasexual competition. As an example, men's focus on physical attractiveness (e.g., fertility cues) in choosing a spouse intensifies female-female competition in this area, with women competing by highlighting or manipulating (e.g., through makeup or padded bras) these cues (Buss & Shackelford, 1997). Similarly, women's preference for culturally successful and resource-holding men intensifies male-male competition for control of culturally important resources, especially in cultures where female choice is not suppressed (Geary, 1998). The reproductive preferences of men and women conflict directly in other ways (Buss & Schmitt, 1993). As with other mammalian males, men can reproduce without paying the cost of parental investment, but women do not have this option. The result is men, on average, are more interested in casual sexual relationships than are women and

68

David C. Geary

will sometimes manipulate women into sexual relationships and then abandon them if a pregnancy occurs (Oliver & Hyde, 1993; Symons, 1979). Men face a different issue. When they invest in children, the paternity of these children is never certain. Although firm estimates are not yet available, perhaps 10-15% of children are sired by a man who is not the women's husband and putative father (Geary, 2000). In these situations, the man is being manipulated by his spouse into raising the children of another man. B. SEX DIFFERENCES IN LIFE HISTORY

A relation between heritable variability in life history traits (e.g., parenting) and survival and reproductive outcomes in human populations has been inferred in many analyses (Geary, 2000; Kaplan et al., 2000), but, unfortunately, rarely evaluated in genetic studies of reproductive fitness. In one such twin study, the relation between lifetime reproductive fitness (number of surviving children) and three life history traits--age at menarche, at first reproduction, and at menopause-was assessed for Australian women over the age of 45 (Kirk et al., 2001). Age at first reproduction (early 20s vs late 20s) was significantly related to reproductive fitness (controlling for educational level and religious affiliation), with earlier reproduction resulting in more children during the reproductive life span. The covariation between age of first reproduction and reproductive fitness was due, in part, to shared genetic influences, indicating that a subset of genes influences both age of first reproduction and lifetime reproductive fitness in the Western women assessed in this study. Similar studies have not yet been conducted for all of the life history traits described in Table II, although individual differences in many of these traits (e.g., life span) have been shown to be related to both heritable and environmental factors (e.g., Herskind et al., 1996). In any event, these traits suggest that selection favored a life history pattern whereby humans invested heavily in a small number of offspring, presumably so that these offspring acquire sophisticated social, behavioral, and cognitive skills, that is, acquired reproductive potential. An unresolved issue is the selection pressures that contributed to the evolution of this suite of life history traits. As stated, theoretical models tend to focus on ecological (e.g., food acquisition, predation) pressures, social pressures (e.g., between group competition), or some combination. To the extent that social competition contributed to the evolution of human life history, sex differences in the associated traits should vary in ways consistent with sexual selection and other forms of social competition. In Table IV, I present a series of predictions regarding how sexual selection might have influenced the evolution of sex differences in human life history. Many of the empirical studies cited therein and described in the following sections are consistent with the predictions, but definitive tests will require the type of analysis conducted by Kirk et al. (2001). Before proceeding to the discussion of sex differences, I provide a general outline of human life history in the first part.

Sexual Selection and Human Life History

69

TABLE IV Predicted Influence of Sexual Selection on Human Life History Women's life history L Female-female competition 1. Female-female competition is highly social, involving the manipulation and disruption of the relationships of other girls and women as related to mating dynamics and resource control (Crick et al., 1997). Based on the relations among social competition, length of the developmental period, and brain size in primates (Joffe, 1997), human female-female competition is predicted to favor delayed maturation and developmental activities that involve the practice and refinement of competition-related social competencies. II. Male choice 1. Due to menopause and women's declining fertility beginning in the late 20s (Menken et al., 1986), men base mate choice decisions, in part, on physical indicators of fertility. These indicators, such as large eyes, are correlated (across species) with youth. Male choice is thus predicted to favor earlier maturation in women and thus retainment of these cues, although selection for earlier maturation is balanced by the conflicting benefits of larger size and thus delayed maturation (Kirk et al., 2001; Stearns & Koella, 1986).

Men's life history L Male-male competition 1. Men are predicted to have and evidence the same life history pattern found in other mammalian species with physical male-male competition. Included among these traits are, in comparison to women, a shorter life span, slower growth and longer developmental period, larger adult size, higher mortality rate at all ages, and higher levels of risk taking and intrasexual violence (e.g., Allman et al., 1998; Wilkins, 1996; Wilson & Daly, 1985). 2. In relation to other mammals, the previously noted pattern has likely been mitigated by paternal investment, resulting in an evolutionary reduction in the magnitude of the sex differences in many life history traits (e.g., maximum life span). 3. Coalitional male-male competition is necessarily a complex and highly social activity (Geary & Flinn, 2002). As with women, social competition should favor delayed maturation and developmental activities that involve the practice and refinement of associated competencies. II. Female choice 1. Female choice is related, in part, to the physical and social cues associated with male-male competition and should thus intensify these aspects of male-male competition, such as the tendency toward behavioral risk taking (Buss, 1989; Kelly & Dunbar, 2001). 2. Female choice is also related to physical indicators of health (e.g., Gangestad, Bennett, & Thornhill, 2001), and the expression of these indicators may be costly for many men (Shackelford & Larsen, 1997). For instance, the androgens that result in the development of physical traits (e.g., masculine jaw) associated with female choice are predicted to compromise the immune system (Folstad & Karter, 1992) and thus the physical health and development of some males (Geary, 1998).

1. Pattern of Human Life History The focus of human life history research conducted by biologists and anthropologists is largely on the growth component of somatic effort shown in Figure 1 as well as reproductive correlates. These studies reveal that in traditional societies and in preindustrial Europe, a typical life history pattern involves women achieving menarche at about 15-16 years of age, marrying soon thereafter, and having their first child between 18 and 22 years of age (Kaplan et al., 2000; Korpelainen, 2000;

70

David C. Geary

Stearns & Koella, 1986). For women, peak fertility is achieved in the mid- to late 20s and gradually declines to near zero by age 45 (Menken, Trussell, & Larsen, 1986; Wood, 1994), resulting in a 25-year reproductive span. During this span, a common interbirth interval is 2-4 years and a common pattern is for women to have four to six children before the onset of menopause, although one to three of these children do not survive to adulthood (Blurton Jones, 1986; Hill & Kaplan, 1999; Lummaa, 2001). Women and men who survive to age 15 years will, on average, live to their mid-50s (e.g., Hill & Hurtado, 1996). In many traditional societies, socially dominant men will have several wives, but often do not marry their first wife until their 20s or later; many other men will never marry (e.g., Borgerhoff Mulder, 1990; Chagnon, 1988; Murdock, 1949). Relative to current industrial society, a 30-40% mortality rate before the age of 15 and an average life span of 55 for those who survive to adulthood seems rather dismal. However, in comparison to other apes and mammals, the human pattern represents comparatively low infant and child mortality risks (Lancaster & Lancaster, 1987) and a comparatively large number of children surviving to adulthood relative to other apes (Hill et al., 2001). The achievement of comparatively low infant and child mortality risks and a comparatively large number of surviving children is not likely to be achievable without men's parental investment, at least in traditional societies, whether the tendency to provide this investment evolved primarily for the provisioning of women and dependent children (Kaplan et al., 2000) or to facilitate a broader suite of sociocompetitive competencies (Alexander, 1989; Geary, 2000; Geary & Flinn, 2001; but see Hawkes, O'Connell, Blurton Jones, Alvarez, & Charnov, 1998). As with other species, both heritable and environmental factors appear to contribute to individual differences in human life history traits (de Bruin et al., 2001; Kirk et al., 2001), and apparently within the constraints of norms of reaction (Stearns & Koella, 1986). As an example, Kirk et al. found that individual differences in women's life history traits were influenced, in part, by genetics, with heritability estimates ranging from .51 for age of menarche to .21 for age at first reproduction. Stearns and Koella demonstrated that well nourished and healthy girls achieve menarche earlier than other girls, indicating an important environmental component to the expression of this life history trait. At the same time, there are genetic constraints such that age of menarche is not typically achieved earlier than age 11 years, with a mean of 13 years, even in well-nourished and healthy populations (Kirk et al., 2001). 2. Sexual Selection and Women's Life History Whatever the primary selective advantage, paternal investment necessarily coevolved with women's life history traits. The trade-offs associated with paternal investment are female-female competition over this investment and male choice (Trivers, 1972), along with the compromises needed to maintain a long-term

Sexual Selection and Human Life History

71

spousal relationship. As noted earlier and in Table IV, social competition covaries with a longer developmental period and a larger brain, at least in primates (e.g., Joffe, 1997). The longer developmental period presumably allows juveniles to practice and refine sociocompetitive competencies, that is, to improve their reproductive potential. If so, then female-female competition would have contributed to the long juvenile period found in humans and girls' developmental activities should involve, in part, a preparation for this competition in adulthood as well as preparation for family life as a parent and spouse, as described later. Across societies, male choice is related to physical indicators of female fertility as well as to personal and social indicators of fidelity and thus certainty of paternity (Buss, 1994; Buss, Larsen, Westen, 1996; Geary, 1998). Physical indicators of fertility include age, the hip-to-waist ratio, breast symmetry, and a youthful appearance (Cunningham, 1986; Kenrick & Keefe, 1992; Mr Soler, & Thomhill, 1995; Singh, 1993). The hip-to-waist ratio is a natural consequence of birthing children with large brains, but also influences male choice and may have been exaggerated as a result (Singh, 1993). Enlarged breasts in the absence of lactation is unusual in mammals and may have been shaped by male choice. Moreover, men's rating of women's physical attractiveness is related to breast symmetry, that is, similarity in the size of the two breasts. Breast symmetry, in tum, is a reliable indicator of women's fertility (Mc~ller et aL, 1995). A focus on youthful appearance follows from the age-related decline in women's fertility and appears to have influenced the evolution of certain facial characteristics in women. These characteristics are cross-species indicators of youth, including relatively large eyes and a small chin (Cunningham, 1986). In short, male choice may have been influenced by the life history pattem of women (e.g., menopause) and may have influenced the evolution of certain aspects of women's physical development. As noted in Table V, phenotypic plasticity in some aspects of women's life history is related to social conditions and to female hormones (Ellis, McFadyenKetchum, Dodge, Pettit, & Bates, 1999; Flinn & Low, 1986; Wilson & Daly, 1997). For example, girls with a warm relationship with their father and a father who is highly invested in the family experience menarche later than do gifts living in father-absent homes or with an emotionally distant father (Ellis et al., 1999). Relationship with father may be a proximate indicator of the stability and warmth of the girls' later spousal relationships as well as an indicator of mortality risks in adulthood (a deceased father cannot be invested or warm; Belsky, Steinberg, & Draper, 1991; Chisholm, 1993; Draper & Harpending, 1988). These cues in turn are presumed to influence women's reproductive pattem in adulthood, such that paternal warmth is associated with delayed maturation. Delayed maturation, in turn, should enable the acquisition of additional sociocompetitive competencies and greater reproductive potential. The reproductive potential may involve the acquisition of social and emotional traits that support high cooperation with a spouse and high investment in a small number of children (MacDonald, 1992).

72

David C. Geary

TABLE V Phenotypic Plasticity in Human Life History L Family functioning and wealth 1. Family functioning can influence the phenotypic expression of some life history traits in both men and women. As examples, psychosocial stressors during childhood are associated with smaller adult size and lower testosterone levels in adulthood for men (e.g., Flinn et al., 1996) and a warm relationship with father is associated with later maturation for women (Ellis et al., 1999). 2. Wealth of the family and kin group can influence nutritional and physical health during development and result in an earlier age of maturation, larger adult size in women and men, and lower infant and child mortality risks (Herman-Giddens, Wang, & Koch, 2001; Korpelainen, 2000; Stearns & Koella, 1986). II. Culture and the operational sex ratio 1. In ecologies with high adult mortality (often due to male-on-male violence) women and men have an earlier age of reproduction and have more children during the reproductive life span (Chisholm, 1993; Wilson & Daly, 1997). In ecologies with high infant and child mortality risks, women and men have more children during their reproductive life span (Korpelainen, 2000). 2. Socially imposed monogamy, as in Western culture, moderates the dynamics of intersexual choice and intrasexual competition. The intensity of male-male competition decreases and the standards of male choice increase, whereas the intensity of female-female competition increases but female choice does not appear to be as strongly affected (Geary, 1998). As an example, in these societies men's mate choices are influenced by women's material resources (e.g., dowry, income; Gaulin & Boster, 1990). As a result, the marriage and reproduction of women may be delayed while these resources are accrued by women and/or their families. 3. The intensity of competition for cultural resources and social status is predicted to result in corresponding delays in men's reproductive opportunities beyond the age of physical maturation, as younger men will typically be disadvantaged in comparison to older men, both in terms of female choice and male-male competition (e.g., Borgerhoff Mulder, 1988). One possible result is age-related difference in men's reproductive strategies, with younger men engaging in more opportunistic mating and older men investing in children (Draper & Harpending, 1988). 4. The operational sex ratio moderates the dynamics of intersexual choice and intrasexual competition. When the number of marriage-age men is lower than marriage-age women, men have more influence (male choice) over mating dynamics and the intensity of female-female competition increases. One result is that men invest more in mating effort and less in parenting (Geary, 2000; Guttentag & Secord, 1983). III. Sex hormones and endocrine functions 1. As with other species, hormonal and endocrine mechanisms (e.g., testosterone) are predicted to moderate and perhaps mediate many sex differences in life history traits and aspects of phenotypic plasticity (Sinervo & Svensson, 1998). Hormones associated with developing the physical and social competencies associated with male-male competition likely result in the trade-offs of higher mortality rates and a shorter maximum life span. Female hormones may facilitate immune responses in women, but with the trade-off of higher risk of autoimmune disorders (Wizemann & Pardue, 2001). 2. Hormonal fluctuations across the menstrual cycle appear to influence women's mate choice preferences, the likelihood of cuckolding her mate, and patterns of relationship jealousy; they may also influence indicators of fertility associated with male choice (Bellis & Baker, 1990; Gangestad & Thornhill, 1998; Geary, DeSoto, Hoard, Sheldon, & Cooper, 2001; Scutt & Manning, 1996).

Sexual Selection and Human Life History

73

Other evidence of phenotypic plasticity in women's life history comes from the work of Wilson and Daly (1997), who found that age of first reproduction, number of children borne per woman, mortality risks, and local resource availability are all interrelated in modern-day Chicago. Women who grow up in contexts with high risk of violent death (related to male-male competition over limited resources) begin reproducing at an earlier age and have more children during their reproductive life span than do women who grow up in low-risk, high-resource ecologies (see also Chisholm, 1993). Similarly, in preindustrial Europe, women living in contexts with high infant and child mortality risks had more children during their reproductive life span than did women living in contexts with lower mortality risks (Korpelainen, 2000). Wider social conditions also appear to influence women's reproductive pattern. In societies in which men's reproductive options are restricted by socially imposed monogamy, women's financial contributions to the family (e.g., a dowry) become a more prominent feature of male choice than in other societies (Flinn & Low, 1986; Geary, 1998). In these contexts, age of first marriage and reproduction may be delayed while women or her kin accrue the resources needed (e.g., for the dowry) to attract a high-status spouse (Gaulin & Boster, 1990). Hormones also influence women's reproductive and life history traits. For instance, hormonal fluctuations across the menstrual cycle are correlated with women's mate choice preferences, such that women may be more likely to cuckold their partner during the time of rising fertility risk, that is, around the time of ovulation (e.g., Bellis & Baker, 1990; Gangestad & Thornhill, 1998). Hormones may also contribute to the longer life span of women relative to men. Female hormones may contribute to enhanced immune responses and reduced risks for certain forms of premature death (e.g., heart disease; Wizemann & Pardue, 2001). All these findings are in keeping with the earlier described patterns of phenotypic plasticity in other species. For instance, Daly and Wilson's (1997) intriguing findings are consistent with the cross-species pattern of earlier maturation being related to high adult mortality risks and phenotypic plasticity in age of first reproduction (e.g. Roff, 1992; Reznick & Endler, 1982). However, all of the human studies are based on phenotypic correlations and thus the patterns might reflect phenotypic plasticity, genetic correlations (e.g., between age of menarche and paternal investment), or, most likely, some combination. 3. Sexual Selection and Men's Life History

With the exception of paternal investment, the pattern of men's life history is the same as that found for other mammalian species in which males compete physically for access to females or for control of the resources females need to reproduce (see Table V; Allman et al., 1998; Leigh, 1996; Wilson & Daly, 1985). Even paternal investment does not alter the general pattern, because men invest

74

David C. Geary

less in parenting and more in mating than do women. Relative to females, the life history traits that men share with other mammalian males include an average and maximum life span that is shorter than females' (Allman et al., 1998), a slower growth rate and longer growth spurt (Leigh, 1996; Tanner, 1990), higher mortality rates at all ages (Wizemann & Pardue, 2001), and greater behavioral risk taking and higher rates of competition-related violent death in adolescence and young adulthood (Daly & Wilson, 1988; Wilson & Daly, 1985). As with other species, all these life history traits are influenced by testosterone and other hormones and appear to involve a preparation for or engagement in male-male competition (e.g., Folstad & Karter, 1992; Dabbs & Dabbs, 2000; Wizemann & Pardue, 2001). Many of these same physical (e.g., height) and behavioral traits (e.g., risk taking) also influence female choice (Beck, Ward-Hull, & McClear, 1976; Kelly & Dunbar, 2001). The slow growth and extended developmental period result in larger men than women (Tanner, 1990, 1992). Among other physical and physiological sex differences that become exaggerated during this time are running speed, physical strength, and physical activity level, as well as throwing distance, velocity, and accuracy (Eaton & Enns, 1986; Kolakowski & Malina, 1974; Thomas & French, 1985). These and related sex differences are often attributed to the division of labor in general and men's hunting in particular (Kolakowski & Malina, 1974; Tiger, 1969). However, the finding that men are consistently better on tasks that involve the ability to track thrown objects and to evade or block these objects (Watson & Kimura, 1991) may be the evolutionary result of male-male competition that involved the use of projectile (e.g., rocks) and blunt force weapons (see Keeley, 1996, for examples), with tracking and blocking being evolved defensive competencies needed to avoid projectiles (Geary, 1998). As with women, individual differences in these physical and life history traits appear to be influenced by genetic and environmental factors (Gilger, Geary, & Eisele, 1991; Martorell, Rivera, Kaplowitz, & Pollitt, 1992). Due to slower growth, greater physical activity levels, and perhaps the immunosuppressive effects of testosterone, boys are more severely affected--physically and cognitively--by poor early conditions than are girls (Flinn, Quinlan, Decker, Turner, & England, 1996; Martorell et al., 1992). One result is that the sex difference in physical size is smaller in poorly nourished populations (Gaulin & Boster, 1992). The form of male-male competition clearly shows phenotypic plasticity, as noted in Table V. In contexts with high mortality risks, often due to male-on-male violence, men reproduce sooner and often show lower levels of parental investment than do men in other contexts (Geary, 2000; Wilson & Daly, 1997). When physical male-onmale violence is suppressed and the attainment of culturally important resources (e.g., money) requires prolonged education and training, men have lower mortality risks and a longer life span, but they reproduce later and often have fewer children (Geary, 1998; Ptrusse, 1993). To varying degrees, female choice is important in

Sexual Selection and Human Life History

75

all these contexts. In the latter contexts, for instance, women's preference for culturally successful men results in delayed marriage and reproduction for many men (Buss, 1996). As with other species, the proximate influences on men's (and women's) life history traits almost certainly include sex hormones and other endocrine functions as well as social context. The hormonal correlates of male-male competition are highly sensitive to social and contextual cues (Geary & Flinn, 2002). Loss in male-male competition often results in heightened elevation of stress hormones, reduced testosterone levels, and perhaps compromised health (Dabbs & Dabbs, 2000). For men, early childhood stressors, such as extended family conflict or lower levels of paternal investment, are associated with atypical stress responses later in life and smaller adult size (Flinn et al., 1996). These, in turn, can result in disadvantages with respect to both male-male competition and female choice. Still, many questions regarding men's life history and phenotypic plasticity in the expression of these traits remain unanswered. For instance, is the relation between adult mortality risks and early reproduction an expression of phenotypic plasticity, of genes that influence both early reproduction and tendency toward male-onmale violence, or of some combination? Do the immunosuppressive effects of stress hormones affect some men more than others, as appears to be the case with other species (Folstad & Karter, 1992)? Do these effects vary at different points in the life span? C. SEX DIFFERENCES IN DEVELOPMENTAL ACTIVITY

Unlike biologists and anthropologists, the research focus of developmental psychologists is largely on the developmental activity component of somatic effort shown in Figure 1. The details of developmental activity as related to somatic effort and the accumulation of reproductive potential are shown in Figure 6. In this view, parent-child relationships, the wider kin network in traditional societies, and

Developmental Activity Growth and Maintenance Parent-Offspring Relationship

Kin

Relationships

Self-Initiated Activities

Reproductive Potential Parent-Offspring Relationship

Peer Relations and Social Play

Solitary Play

Fig. 6. Components of developmental activity. Parent-offspring relations emerge from parents' reproductive efforts and the efforts of offspring to obtain additional parental resources, as related to the offspring's somatic effort and the accumulation of reproductive potential

76

David C. Geary

some self-initiated activities during childhood and adolescence, such as foraging in traditional societies, function to promote growth and maintenance during specific developmental periods (Bjorklund, 1997; Bjorklund & Pellegrini, 2002; Geary & Bjorklund, 2000). The function of other activities is, in theory, to accumulate reproductive potential, that is, to acquire the physical, social, behavioral, and cognitive competencies that enable successful reproduction in adulthood, whether the reproductive effort involves mating, parenting, or some combination. The function of parent-offspring relationships as related to social competition and reproductive effort (e.g., acquiring resources that facilitate competition with peers) in adulthood is described elsewhere (Geary & Flinn, 2001). The focus here is on the developmental activity component of peer relationships and various forms of play. As stated previously, from a life history perspective one function of these activities is to refine the social, behavioral, and cognitive competencies that covaried with survival and especially~given comparatively low adult mortality rates~reproductive outcomes during the species' evolutionary history (Geary, 1998; K~i~ & Jokela, 1998; Kaplan et al., 2000; Mayr, 1974). The basic skeletal structure of these competencies appears to be inherent, but fleshed out and adapted to local conditions as the juvenile engages in the associated activities, such as play hunting (Gelman, 1990; Gelman & Williams, 1998). The most fundamental of these competencies coalesce around the domains of folk psychology, folk biology, and folk physics (e.g., Atran, 1998; Leslie, 1987; Mandler, 1992), in keeping with the position that humans are fundamentally motivated to gain access to and control of the social (e.g., mates), biological (food), and physical (e.g., territory) resources that covary with survival and reproductive outcomes in the local social group and ecology (Geary, 1998; Geary & Huffman, 2002). The general prediction is that children will evince a pattern of self-initiated activities that results in the practice and refinement of social, behavioral, and cognitive competencies that covaried with survival and reproductive outcomes during human evolution. The specific predictions relating to the current discussion are: (1) sex differences will emerge in the form of intrasexual relationships and these will be related to patterns of male-male and female-female competition; (2) girls will engage in more play parenting and family-oriented play, on the basis of the sex difference in parental investment; and (3) boys will engage in more activities associated with ecological dominance, specifically activities involved in hunting and territory maintenance. Full review and discussion of research related to these predictions is beyond the scope of this chapter (see Geary, 1998, 2002), but overviews are provided in the following sections. 1. Peer Relationships and Social Play Basic patterns of children's peer relationships and social play are considered in the following subsections, including social segregation and separate looks at prominent features of the social activities of boys and gifts.

Sexual Selection and Human Life History

77

a. Social Segregation. If males and females differed in the competencies that covaried with survival and reproductive outcomes during human evolution, then children are predicted to segregate by sex and engage in activities that mirror this evolutionary history (Geary, 1998). In keeping with the first prediction, one of the most consistently found features of children's social behavior is the formation of same-sex play and social groups (Maccoby, 1998). Such groups are evident by the time children are 3-years-old and become increasingly frequent throughout childhood. For example, in a longitudinal study of children in the United States, 4- to 5-year-olds spent 3 h playing with same-sex peers for every 1 h they spent playing in mixed-sex groups; as 6- to 7-years-olds, the ratio of time spent in samesex versus mixed-sex groups was 11:1 (Maccoby & Jacklin, 1987). The same pattem has been found for children in Canada, England, Hungary, Kenya, and Mexico (Strayer & Santos, 1996; Turner & Gervai, 1995; Whiting & Edwards, 1988). The degree of segregation varies across contexts and is most common in situations in which children are free to form their own social groups (Maccoby, 1988; Strayer & Santos, 1996). The proximate mechanisms driving child-initiated segregation include the different play styles of girls and boys (described later) and differences in the strategies used to attempt to gain control of desired resources (e.g., toys) or to influence group activities. In situations where access to a desired object is limited, boys and girls use different social strategies (Charlesworth & Dzur, 1987). More often than not, boys gain access by playfully shoving and pushing other boys out of the way, whereas girls gain access by means of verbal persuasion (e.g., polite suggestions to share) and sometimes verbal command (e.g., "It's my turn now!"). Based on findings such as these, Maccoby (1988) argued that segregated social groups emerge primarily because children are generally unresponsive to the play and social-influence styles of the opposite sex. Boys, for instance, sometimes try to initiate rough-andtumble play with girls but most (not all) girls withdraw from these initiations, whereas most other boys readily join the fray (Pellegrini & Smith, 1998). Similarly, girls often attempt to influence the behavior of boys through verbal requests and suggestions but boys, unlike other girls, are generally unresponsive to these requests (Charlesworth & LaFrenier, 1983). Segregation may also be related to the formation of the social categories of "boy" and "girl" and a tendency to prefer individuals in the same category. However, same-sex segregation occurs before many children consistently label themselves and other children as a boy or a girl, indicating that social categorization is not likely to be a sufficient explanation for this phenomenon (Maccoby, 1988). The net result of segregation by sex is that boys and gifts spend much of their childhood in distinct peer cultures (Harris, 1995; Maccoby, 1988), and differences in the social styles of boys and girls congeal in the context of these cultures. b. Boys' Peer Relationships. Many features of boys' play, social relationships, and social motives support the position that one-on-one and coalitional male-male

78

David C. Geary

competition were (and still are) prominent social dynamics during human evolution (Geary, 1998; Geary & Flinn, 2002). As with other species exhibiting male-male competition, boys' activities are more consistently directed toward the achievement of hierarchical dominance than are girls' activities, a sex difference found across culture, age, and social context (e.g., Feingold, 1994; Maccoby, 1988; Whiting & Edwards, 1988). Rough-and-tumble play is one of the earliest social manifestations of physical one-on-one dominance, or at least the practice of this. In situations where activities are not monitored by adults and not otherwise restricted (e.g., a play area that is too small), groups of boys engage in various forms of roughand-tumble play--including playful physical assaults and wrestling--three to six times more frequently than do groups of same-age girls (DiPietro, 1981; Maccoby, 1988). In the United States the sex difference in playful physical assaults and other forms of rough-and-tumble play begin to emerge by about 3 years of age (Maccoby, 1988). The same general pattern is found in other industrial societies and in traditional societies in which it has been studied, although the magnitude of the sex difference varies from one culture to the next (Eibl-Eibesfeldt, 1989; Whiting & Edwards, 1973, 1988). Frequency of boys' rough-and-tumble play peaks between the ages of 8-10 years (Pellegrini & Smith, 1998). At this time, boys spend about 10% of their free time engaged in this form of play. As boys move from childhood to adolescence, the line between play and actual physical aggression begins to blur, and a relation between these activities (e.g., bullying) and social dominance emerges. In one study of 10- to 12-year-old boys, physical aggression and bullying first increased during the early part of the school year, but then decreased as the year progressed--presumably as boys developed stable dominance hierarchies (Pellegrini & Bartini, 2001; see also Savin-Williams, 1987). As physical aggression decreased, affiliative behaviors increased. Unlike younger boys where physical aggression is often associated with unpopularity and social rejection (Newcomb, Bukowski, & Pattee, 1993), these activities in adolescent boys are associated with social dominance, as defined by peers and teachers, and with a higher frequency of dating and higher rated attractiveness by girls (Pellegrini & Bartini, 2001). Affiliativebehaviors following aggressive within-group dominance encounters are common in primate species and function to maintain group cohesion (de Waal, 2000). For humans, the pattern is consistent with the prediction that, for boys, play functions, in part, to practice the formation of competition-related coalitions, that is, to maintain a large enough social group to effectively compete against other groups of boys (Geary, 1998; Geary & Flinn, 2002). The most common venue for the practice of these competencies, such as coordinating group activities, is competitive group-level games. As an example, Lever (1978) found that 10- and 1 I-year-old boys participated in group-level competitive activities, such as football, three times as frequently as did girls. In addition, boys' spontaneous social play involved larger groups, on average, than did girls' social play and involved greater

Sexual Selection and Human Life History

79

role differentiation within these groups. Similar differences have been reported by others (Sandberg & Meyer-Bahlburg, 1994; Sutton-Smith, Rosenberg, & Morgan, 1963). By late adolescence boys' competencies regarding the cooperation and social support needed to function effectively as a competitive coalition, as in the context of team sports, is very sophisticated (Savin-Williams, 1987). The sex difference in one-on-one and group-level competitive play is related, at least in part, to prenatal exposure to androgens (Collaer & Hines, 1995). For example, Berenbaum and Snyder (1995) reported that girls who were prenatally exposed to excess levels of androgens (i.e., congenital adrenal hyperplasia, CAH) engaged in more athletic competition than did their unaffected peers--about three out of four girls affected by CAH engaged in athletic competition more frequently than did the average unaffected girl. This difference, however, was not as large as the difference between unaffected boys and unaffected girls--more than nine out of ten unaffected boys reported engaging in athletic competition more frequently than the average unaffected girl. Hines and Kaufman (1994) found that girls affected by CAH engaged in more playful physical assaults, physical assaults on objects, wrestling, and rough-and-tumble play in general than did unaffected girls, but none of these differences were statistically significant. The lack of significance was possibly due to the testing arrangements used in this study. Here, most of the girls affected by CAH were observed as they played with one unaffected girl, a situation (two girls) that does not typically facilitate rough-and-tumble play. c. Girls' Peer Relationships. In comparison to boys, the social relationships that develop among dyads of girls are more consistently communal, manifesting greater empathy, more concern for the well-being of the other girl, and a greater emphasis on intimacy and social/emotional support (e.g., Maccoby, 1988; Whiting & Edwards, 1988). In addition, girls' social groups tend to be smaller, often including dyads or triads, and are characterized by a motivational disposition centered on cooperation and equality among group members, as contrasted with boys' focus on social dominance (Ahlgren & Johnson, 1979; Knight & Chao, 1989; Rose & Asher, 1999). During the preschool years, the focus of girls' social activities is often sociodramatic play with a family-oriented theme (Pitcher & Schultz, 1983). As they grow older, the focus is more explicitly on the development and maintenance of a small network of friends, with these relationships focusing on interpersonal dynamics (e.g., relationships with other girls or boyfriends) and providing social and emotional support, typically as related to interpersonal conflict (Belle, 1987; Savin-Williams, 1987; Taylor et al., 2000). As described earlier, girls are also competitive, but unlike male-male competition, female-female competition is less physical and functions largely to manipulate and disrupt social relationships through shunning, gossiping, spreading lies, and so on (Crick et al., 1997). As with rough-and-tumble play, relational

80

David C. Geary

aggression and the associated sex difference emerge by age 3 and continue through childhood, adolescence, and adulthood (Crick et al., 1999). Developmentally, relational aggression becomes increasingly sophisticated. A 3-year-old might try to control the behavior of another child by stating "If you don't play with me, I won't be your friend," whereas an older child will manipulate the social network, for example by spreading rumors (e.g., "She said that you... ") of the girl she is attempting to control. The social sophistication of this form of aggression increases into adolescence and becomes increasingly focused on the disruption of the romantic relationships of other girls (Crick & Rose, 2000). Studies conducted outside of the United States typically find the same developmental pattern and sex difference, although the relevant cross-cultural research is meager in comparison to research on rough-and-tumble play (Crick et al., 1999). The costs associated with being relationally aggressive include higher rates of social rejection and depression (although some relationally aggressive girls are well liked by some other girls; Crick et al., 1999). Relational aggression has benefits in disrupting the social networks of and creating distress in other girls and women--the presumed evolutionary function of this form of aggression. Geary (1998, 2002) argued that the function of women's social networks is to provide social support and stability as well as increased access to important resources, including men. These in turn are associated with better health for these women and, in some contexts, lower mortality and morbidity risks for their children (Flinn, 1999; Geary, 2000; Taylor et al., 2000). In this view, girls and women should highly value the reciprocal and intimate relationships that define and maintain their social networks and should react more strongly than boys and men to the disruption of these relationships; competition for dominance within boys' and men's in-groups result in males being more tolerant of conflicted relationships (de Waal, 1993; Geary & Flinn, 2002). Several analyses of the life events that trigger depressive symptoms in adolescent boys and girls support this position (Bond, Carlin, Thomas, Rubin, & Patton, 2001; Leadbeater, Blatt, & Quinlan, 1995). Both boys and girls experience symptoms of depression following personal failure, such as poor grades. However, a sex difference is found in reactivity to negative interpersonal events: Adolescent girls and women are much more likely to experience symptoms of depression following interpersonal conflict or loss of a significant relationship than are same-age boys and men. In addition, adolescent girls and women often experience symptoms of depression when negative life events affect individuals in their social networks, whereas boys and men typically do not. In fact, adolescent girls apparently are up to four times more likely than same-age boys to experience anxiety and depression as a result of disrupted interpersonal relationships, disruptions that are often the result of relational aggression (Bond et al., 2001). In sum, relational aggression is an effective method for disrupting the romantic and same-sex relationships of other girls and women and can result in significant

Sexual Selection and Human Life History

81

levels of distress, anxiety, and depression for the victims. The costs of relational aggression included the potential for social rejection, but the presumed benefits are increased control of and access to desired social relationships, including romantic relationships.

2. Play Parenting and Family-Oriented Play For most species of primate, play parenting (e.g., caring for siblings) is frequently observed in young females that have not yet had their first offspring, and it is often associated with higher survival rates of the firstborn, and sometimes later born, offspring (Nicolson, 1987). Across five primate species, firstborn survival rates were from two to more than four times higher for mothers with early experience with infant care--obtained through play parenting--than for mothers with no such experience (Pryce, 1993), suggesting that play parenting is indeed a form of practice that refines parenting competencies. Maternal care is also influenced by prenatal exposure to sex hormones and the hormonal changes that occur during pregnancy and the birthing process, such that a combination of early play parenting and hormonal influences contribute to the adequacy of female caregiving in many primate species (Lee & Bowman, 1995; Pryce, 1995). Humans are no exception. In addition to investing more in parenting than men (Geary, 2000), women engage in more play parenting and family-oriented play as children. The sex difference in play parenting is related, in part, to the fact that girls throughout the world are assigned child-care roles, especially for infants, much more frequently than are boys (Whiting & Edwards, 1988). Girls also seek out and engage in child-care, play parenting, and other domestic activities (e.g., playing house)--with younger children or child substitutes, such as dollsm much more frequently than do same-age boys (Pitcher & Schultz, 1983). During the early preschool years, these themes are commonly enacted during solitary play (e.g., playing house with a baby doll), but peers are incorporated into these themes as children become more socially experienced and competent. Beginning around 3-4 years of age, the social-symbolic play of boys tends to focus on issues of physical fighting and competition (e.g., "cowboys and Indians"), whereas that of girls is more commonly focused on family relationships (e.g., "mother and child"; Pitcher & Schultz, 1983). As in other domains, the magnitude of this sex difference varies across age and context. Prior to about age 6, both girls and boys are generally responsive to infants, but after this age, and continuing into adulthood, girls are more responsive to infants and younger children than are boys (Berman, 1986; Edwards & Whiting, 1993). The emergence of this sex difference is related to a significant drop in the frequency with which older boys attend to and interact with infants and younger children (Berman, 1986; Sandberg & Meyer-Bahlburg, 1994). This sex difference has persisted over nearly 40 years of significant change in the social roles and opportunities of women in the United States (Sutton-Smith et al., 1963), and it has

82

David C. Geary

been found across other industrial societies and in traditional societies in which it has been studied (Eibl-Eibesfeldt, 1989; Whiting & Edwards, 1988). The sex differences in interest in infants, children, and families, as well as in engagement in play parenting, are influenced by the prenatal hormonal environment and by hormonal changes occurring during puberty. Girls affected by CAH show less interest in infants and families and engage in play parenting, among other things, less frequently than their unaffected sisters (Berenbaum & Hines, 1992; Berenbaum & Snyder, 1995; Collaer & Hines, 1995; Leveroni & Berenbaum, 1998). Berenbaum and Hines (1992) compared 5- to 8-year-old girls affected with CAH with unaffected same-sex relatives and found that unaffected gifts played with dolls and kitchen supplies 2 1/2 times longer than did girls affected by CAH. These girls, in turn, played with boys' toys (e.g., toy cars) nearly 2 1/2 times longer than did unaffected girls. The same pattern was found in a follow-up study 3-4 years later (Berenbaum & Snyder, 1995). Furthermore, when allowed to choose a toy to take home after the assessment was complete, unaffected girls most frequently chose a set of markers or a doll to take home, whereas girls affected by CAH most frequently chose a transportation toy (e.g., toy car) or a ball. Unaffected girls also show an increased interest in children following menarche (Goldberg, Blumberg, & Kriger, 1982), whereas the interests of girls affected by CAH remain more malelike (Berenbaum, 1999). 3. Ecologically Related Play Patterns

If men are inherently motivated to attempt to achieve ecological dominance, then sex differences should be found in activities that support this goal. Included among these activities are tool use, hunting, and exploration and control of the wider ecology (e.g., as in control of natural resources). Across traditional societies, men are indeed more likely to use objects as tools (e.g., metal work, weapon making), to hunt, and to travel in unfamiliar territory as related to hunting and warfare (Murdock, 1949). As noted earlier, these activities enable coalitions of men to define and maintain--typically in conflict with other coalitions--a territory for their group and to extract physical (e.g., control of water supply) and biological (e.g., food, medicine) resources from this territory. Although women's gathering contributes to the latter as well, it typically occurs within the confines of the territory maintained by men. In any case, sex differences in play and other developmental activities that would provide the practice needed to refine these competencies are predicted. The associated research base is not as extensive as the base on sex differences in social-developmental activities, but extant research is consistent with the prediction. a. Tool Use. Skilled tool use in adulthood appears to be facilitated by objectoriented play during juvenility. Such play is uncommon in wild primates, except for a few tool-using species, including humans and chimpanzees (Byme, 1995;

Sexual Selection and Human Life History

83

Goodall, 1986). Object-oriented play is often solitary and involves the nonfunctional manipulation of objects, such as throwing them, banging them, and so forth. The function of this play appears to be to learn about the different ways in which various objects can be used, which in turn appears to facilitate later tool use and later problem-solving skills as related to tool use. Chimpanzees that lack objectoriented play during juvenility are, as adults, less successful in problem-solving with objects (Byrne, 1995). As with the chimpanzee, object-oriented play apparently helps children to learn about the physical properties of objects and the different ways in which these objects can be used and classified. For example, preschool children whose play was object oriented had higher scores on tests of spatial cognition (e.g., the ability to mentally represent and mentally manipulate geometric designs) and were better able to sort objects based on, for example, color and shape (Jennings, 1975). Boys and men are consistently found to be more object oriented than girls and women (Willingham & Cole, 1997). In addition, preschool boys learn to use tools more quickly and readily than preschool girls. Chen and Siegler (2000) found that 18-month-old boys were better than same-age gifts at inferring how objects could be used as tools; were more skilled at using objects as tools (e.g., to retrieve a desired toy); and, learned to use objects as tools more quickly, that is, with less practice and less need for adult demonstration. As previously mentioned, gifts affected with CAH show more male-typical object-oriented play than other girls, suggesting that exposure to prenatal androgens may influence this sex difference (Berenbaum & Hines, 1992).

b. Ecological Exploration and Manipulation. Beginning in middle childhood and increasing through adolescence, boys have larger play ranges than girls and explore and manipulate these ecologies (e.g., building things, such as forts) much more frequently than do girls (Matthews, 1992). These sex differences appear to contribute to the sex differences in certain spatial competencies, especially the ability to form a mental representation of the wider ecology, as contrasted with the ability to remember the location of specific objects in this ecology (see Silverman & Eals, 1992). The sex difference in the area of the play range appears to be related, in part, to greater parental restrictions on the ranges of girls than on those of boys. However, a sex difference in the size of the play range is found in the absence of any such restrictions, both in industrial societies and in those traditional societies in which it has been studied (Matthews, 1992; Munroe & Munroe, 1971). In studies of the exploratory play of children in suburban England, for instance, Matthews (1992) found that younger children--both boys and girlsmtended to play within close proximity of one or both of their parents (see also Whiting & Edwards, 1988). In contrast, 8- to 11-year-old children were more likely to play away from home, and the area of the unrestricted play range of boys covered from one and a half to nearly three times the area of the unrestricted play range of same-age gifts.

84

David C. Geary

Whiting and Edwards (1988) reported a similar sex difference for older children in three separate groups in Kenya as well as for children in Peru and Guatemala. Nonetheless, the age at which this sex difference emerges appears to vary with the ecology of the group. For the Ache, who live in dense, tropical rain forest, the size of the range of boys and girls does not typically diverge until adolescence (Hill & Hurtado, 1996). As with tool use, boys profit more from exploratory behavior than girls. Matthews' (1987) study of the relation between exposure to a novel environment and the pattern of sex differences in the ability to spatially represent this environment illustrates the point. Here, 8- to 1 I-year-old boys and girls were taken on a 1-h tour of an unfamiliar area in suburban England. In one condition, the children were given a map of the entire area and were then taken on the tour, with the guide pointing out various environmental features. In the secondmhigh memory demand---condition, children were given a map of half of the area and their tour was interrupted for 30 min at the halfway point, although the same environmental features were pointed out. At the end of the tour, the children were asked to draw a map of the entire area. The maps of boys and girls did not differ in the overall amount of information provided, but sex differences did emerge for other map features, especially with the high-memory-demand condition. Boys were better able than girls to mentally reconstruct the topography of an unfamiliar environment, retaining general orientation, clustering, and Euclidean (e.g., relative direction) relations among important environmental features. The evidence is not definitive, but studies of CAH suggest hormonal influences on these sex differences. Girls and women with CAH are consistently found to outperform their female relatives on tests of spatial ability and report engaging in more spatial-related behaviors while growing up (Kimura, 1999; Resnick, Berenbaum, Gottesman, & Bouchard, 1986). c. Play Hunting. With very few exceptions (e.g., Hewlett, 1992), hunting is almost exclusively a male activity in traditional societies (Murdock, 1949), and it is a skill that requires years of experience to master (Kaplan et al., 2000). Play hunting has not been directly studied by developmental psychologists, but several patterns support the predicted sex differences. Boys attend to potentially dangerous and wild animals more often than girls do, and in traditional societies engage in play hunting more often than girls. Boys' symbolic play more often involves wild animals than domestic ones (Eibl-Eibesfeldt, 1989). Similarly, the drawings of !Ko (central Kalahari) boys depicted domestic and wild animals about three times more frequently than did girls' drawings. In many traditional societies, a sex difference in the focus of boys and girls daily food-gathering activities emerges in late childhood (Kaplan et al., 2000). For example, until about 10 years of age, both Hadza (Tanzania) boys and girls forage. After this age, boys generally restrict their activities to hunting, despite the fact that their hunting returnsmin

Sexual Selection and Human Life History

85

terms of calories--are much lower than would be the case if they continued to forage (Blurton Jones, Hawkes, & O'Connell, 1997). The long-term benefits can be significant, however, as skilled hunters often have higher status (related to male-male competition) and more reproductive opportunities than other men, as mentioned earlier.

4. Phenotypic Plasticity If one function of the developmental period is to enable inherent neural, cognitive, behavioral, and affective systems to be adapted through developmental activities to local conditions, then plasticity in life history traits should covary with the length of the species' developmental period. If so, then humans should show a higher degree of phenotypic plasticity than any other species, but presumably within the constraints of norms of reaction. As described earlier, many human life history traits do in fact covary with local conditions but these analyses have been limited to physical traits, such as age of maturation (Stearns & Koella, 1986). If the sex differences in social behavior and play patterns are the result of human life history evolution and the influence of sexual selection and other evolutionary pressures, then all the behaviors previously described should evince some degree of phenotypic plasticity. Unfortunately, relevant research is meager, but available studies support the prediction (Low, 1989; MacDonald, 1992). To illustrate, the form and intensity of boys' rough-and-tumble play varies across cultures. In societies characterized by relatively high levels of physical male-male competition, the play fighting of boys tends to be rougher than the play fighting found in other societies. For instance, intergroup aggression is a pervasive feature of Yanomam6 society (Venezuela and Brazil; Chagnon, 1988) and young Yanomam6 boys often play fight with clubs or bows and arrows, practices that are typically discouraged in settings where physical male-male competition occurs infrequently. For the Sioux and Native American tribes that frequently engaged in intergroup hostilities, the activities of young boys were designed to encourage both oneon-one and coalition-based aggression and physical endurance (Hassrick, 1964; Loy & Hesketh, 1995). These activities were often sufficiently violent to draw blood, yet afterward the boys were friendly to each other. This pattern--intense but nonlethal in-group competition and mechanisms (e.g., positive affect during "horse play") to maintain the cohesion of the in-group in the service of intergroup competition----is a predicted feature of male-male competition in humans (Geary & Flinn, 2002). These and other games enable the social, behavioral, and affective systems that support the universal tendency of boys to engage in rough-and-tumble play and coalitional competition to be adapted to local conditions (see Geary, 1998). The mechanisms that support such phenotypic adaptations are not well understood but probably involve a combination of parental and peer influences as well as genotype by environment interactions (Low, 1989; MacDonald, 1992).

86

David C. Geary

5. Conclusion

The pattern of human developmental sex differences supports all three predictions described previously. First, many sex differences in peer relationships and play mirror adult forms of intrasexual competition, although the findings are more definitive for male-male than for female-female competition given the larger research base for the former (Maccoby, 1988; Crick et al., 1999). More specifically, the basic patterns are consistent with an evolutionary history of one-on-one and coalitional male-male competitions, which very likely involved the use of blunt force and projectile weapons (for further discussion, see Geary, 1998), as related to the achievement of social dominance and resource control. Female-female competition is much less physical but is also related to resource control, although the resources are largely social and relational. Second, consistent with the sex difference in parental effort, girls engage in play parenting and other forms of familyoriented play much more frequently than do boys (Pitcher & Schultz, 1983). Third, boys engage in ecologically related play and other related developmental activities, such as exploration, much more frequently than do girls (Matthews, 1992). Many of these differences are influenced by prenatal exposure to sex hormones and to hormonal changes associated with puberty. Proximate hormonal influences on the expression of sex differences are, in fact, the norm across species, and they are commonly associated with the expression of traits that have been influenced by sexual selection (Geary, 1998). Although not definitive, the results are also consistent with the view that one function of developmental activities is to practice and refine the competencies that covaried with survival and especially reproductive outcomes during human evolution. Reproductive outcomes are emphasized because of low adult mortality rates in traditional societies, such that much of the variance in evolutionary fitness is related to individual differences in mating success and parental success (i.e., keeping children alive) rather than survival per se (Blurton Jones et al., 1997; Hill & Hurtado, 1996; Irons, 1979). Nearly all of the developmental activities described earlier are consistent with survival and reproductive activities in traditional societies and presumably during human evolution. Equally important, the associated competencies (e.g., coalitional competition) become increasingly sophisticated and adultlike during the developmental period and presumably as a result of developmental activities (e.g., Crick et al. 1999; Pellegrini & Bartini, 2001). The majority of these developmental activities are highly social or are related to social issues, as in the relation between hunting competencies, social status, and men's reproductive options (Kaplan et al., 2000). The social nature of developmental activities combined with the more general covariations among length of the developmental period, brain size, and social complexity in primate species (e.g., Joffe, 1997), support the position that various forms of social competition (see Table III) have been driving forces in human evolution and have significantly influenced human life history (Alexander, 1989; Geary & Flinn, 2001).

Sexual Selection and Human Life History

87

The research reviewed here also supports the general prediction that the function of many developmental activities is the accumulation of reproductive potential and the specific hypothesis that sexual selection has been (and continues to be) an important influence on the evolution of sex differences in human life history traits and in developmental activity. In other words, the sex difference in the distribution of reproductive effort across mating and parenting is by definition sexual selection, and the associated sex differences in developmental activity enable the accumulation of competencies that will support sex differences in reproductive effort during adulthood.

VI. Conclusion Life history and sexual selection represent core theoretical principles in evolutionary biology and guide empirical research related to proximate influences and predicted evolutionary functions of developmental traits and patterns of social dynamics, respectively (Andersson, 1994; Darwin, 1871; Roff, 1992). The first goals of the current chapter were to introduce these principles to human developmental scientists and to illustrate their utility for predicting and explaining developmental patterns and sex differences in nonhuman species. The next and more important goal was to construct a theoretical framework for conceptualizing the potential relations between sexual selection in humans and human life history traits, including children's developmental activities. The construction of this framework was guided by the assumptions that social competition, including sexual selection, has been a driving force in human brain, cognitive, behavioral, and social evolution (Alexander, 1989; Geary & Flinn, 2001) and that developmental activities reflect an evolved motivational disposition to practice and thus refine the associated competencies. More precisely, the function of many developmental activities is presumed to be the accumulation of reproductive potential that is then expended in adulthood in the form of reproductive effort: mating, parenting or some combination (Alexander, 1987). As an example, consider that humans share with many other species the same basic life history traits that have been found to covary with male-male competition and an accompanying sex difference in parental effort. Included among these traits are sex differences in life span, maturational dynamics, adult size, and premature death due to male-on-male violence, among other traits (Allman et al., 1998; Leigh, 1996; Wilkins, 1996; Wilson & Daly, 1985). The pattern is clearly consistent with the position that male-male competition--likely exaggerated by female choice--has influenced the evolution of many features of human life history, including components of physical development, reproduction, and a few behavioral triats (i.e., male-on-male aggression; Roff, 1992). At the same time, standard life history analyses do not fully capture the potential influence of social-behavioral

88

David C. Geary

components of male-male competition and other features of sexual selection on life history evolution, although a few initial analyses have been conducted (e.g., Sawaguchi, 1997). Nor do standard life history analyses capture the importance of developmental activities as related to sexual selection and life history, with very few exceptions (e.g., Collis & Borgia, 1992). When social-behavioral and developmental activities are considered in relation to the importance of coalitional male-male competition and human ecological dominance in traditional societies and presumably during human evolution (Alexander, 1987, 1989; Geary & Flinn, 2001; Wrangham, 1999), many aspects of human life history and associated sex differences fall into place. Sex differences in rough-and-tumble play and participation in sports and other coalitional activities are well documented empirically (e.g., DiPietro, 1981; Lever, 1978; Maccoby, 1988), with the occasional consideration of these patterns as potentially related to male-male competition (Pellegrini & Smith, 1998). However, the consistency of these developmental sex differences with coalitional malemale competition and with one-on-one competition as related to the formation of within-coalition dominance hierarchies has not been fully appreciated. The position here is that these developmental activities are evolved components of human life history that reflect a motivational disposition for boys to engage in activities that prepare them for the forms of one-on-one and coalitional competition that defined male-male competition during human evolution (Geary, 1998). The social nature of male-male competition and associated developmental activities also support Alexander's (1987, 1989) hypothesis that social competition was a driving force during human evolution. Other features of boys' developmental activities, such as object play and play hunting, are in keeping with Kaplan and colleagues' hunting hypothesis-specifically, that hunting/foraging demands were important influences on the evolution of human life history and associated sex differences (Kaplan et al., 2000). The position here is that these activities are indeed important, and related to the evolution of human ecological dominance. At the same time, these activities (e.g., hunting) are aspects of a much broader suite of social, behavioral, and cognitive competencies related to social competition, within (e.g., mating choice) and between groups (Geary & Flinn, 2001). Stated differently, many of these developmental activities enable the accumulation of social, behavioral, and cognitive competencies needed for later reproductive efforts, at least for reproductive efforts in traditional societies and presumably during human evolution. The relation between life history and other components of sexual selection, specifically male choice and female-female competition, are not well understood. For humans, male choice and female-female competition are, nonetheless, integral aspects of sexual selection and, as with male-male competition and female choice, have likely influenced the evolution of human life history traits and developmental activity. As an example, pioneering research by Crick and colleagues on relational aggression provides solid empirical evidence that girls and women

Sexual Selection and Human Life History

89

do indeed compete with one another (Crick et al., 1997, 1999; Crick & Rose, 2000), findings readily interpreted in terms of sexual selection and female-female competition (Geary, 1998, 2002). As with boys' rough-and-tumble play and coalitional games, girls' relational aggression emerges during the preschool years and becomes increasingly sophisticated, socially and cognitively (e.g., in terms of theory of mind), with maturation and practice (Crick et al., 1999). The function of these social activities is to influence and attempt to control social behavior of other girls and women (sometimes boys and men) and with adolescence becomes increasingly focused on romantic relationships, that is, disrupting the romantic relationships of other girls and women. The latter is consistent with the prediction that female-female competition should, in part, be related to paternal investment, that is, developing a spousal relationship and thus access to the man's social and material resources (Geary, 2000; Trivers, 1972). As with boys, many other features of girls' self-initiated activities are interpretable from an evolutionary perspective, although not sexual selection per se. The most obvious of these are play parenting and family-oriented play, which, of course, follow from the sex difference in parental effort. In any case, developmental activities that focus on peer relationships and play parenting both enable the accumulation of competencies related to reproduction in adulthood, female-female competition and parenting, respectively. In sum, basic components of human development, such as length of the developmental period and life span, as well as the details of developmental activities and accompanying sex differences, are readily interpretable from the combined perspectives of life history and sexual selection. Proximate factors, such as gender categorization (Eagly, 1987), are potentially important influences on the expression of many sex differences. Indeed, phenotypic plasticity in life history traits and forms of sexual selection are predicted and have been illustrated in a few cases (Low, 1989; Steams & Koella, 1986). In other words, many of the proximate influences on sex differences might be studied in terms of the broader framework of phenotypic plasticity as related to life history and sexual selection. In any case, a fully informed, human developmental science must incorporate life history traits and the various components of sexual selection.

ACKNOWLEDGMENTS I thank David Bjorklund, Jennifer Byrd-Craven, Mark Flinn, Mary Hoard, and Amanda Rose for insightful comments on an earlier draft and Robert Sites for locating the illustrations used in Figure 2.

REFERENCES Ahlgren, A., & Johnson, D. W. (1979). Sex differences in cooperative and competitive attitudes from the 2nd to the 12th grades. Developmental Psychology, 15, 45-49.

90

David C. Geary

Alberts, S. C., & Altmann, J. (1995). Preparation and activation: Determinants of age at reproductive maturity in male baboon. Behavioral Ecology and Sociobiology, 36, 397-406. Alekseev, V., & Lampert, W. (2001, December 20/27). Maternal control of resting-egg production in Daphnia. Nature, (London), 414, 899-901. Alexander, R. D. (1987). The biology of moral systems. Hawthorne, NY: Aldine de Gruyter. Alexander, R. D. (1989). Evolution of the human psyche. In P. Mellars & C. Stringer (Eds.), The human revolution: Behavioural and biological perspectives on the origins of modem humans (pp. 455513). Princeton, NJ: Princeton University Press. Alexander, R. D. (1990). How did humans evolve ? Reflections on the uniquely unique species. Museum of Zoology (Special Publication No. 1). Ann Arbor: The University of Michigan. Allman, J., McLaughlin, T., & Hakeem, A. (1993). Brain weight and life span in primate species. Proceedings of the National Academy of Sciences USA, 90, 118-122. Allman, J., Rosin, A., Kumar, R., & Hasenstaub, A. (1998). Parenting and survival in anthropoid primates: Caretakers live longer. Proceedings of the National Academy of Sciences USA, 95, 68666869. Andersson, M. (1994). Sexual selection. Princeton, NJ: Princeton University Press. Archer, J. (1992). Ethology and human development. Savage, MD: Barnes & Noble Books. Atran, S. (1998). Folk biology and the anthropology of science: Cognitive universals and cultural particulars. Behavioral and Brain Sciences, 21, 547-609. Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of mind. Cambridge, MA: MIT Press/Bradford Books. Beck, S. P., Ward-Hull, C. I., & McClear, P. M. (1976). Variables related to women's somatic preferences of the male and female body. Journal of Personality and Social Psychology, 34, 1200-1210. Belle, D. (1987). Gender differences in the social moderators of stress. In R. C. Barnett, L. Biener, & G. K. Baruch (Eds.), Gender and stress (pp. 257-277). New York: The Free Press. Bellis, M. A., & Baker, R. R. (1990). Do females promote sperm competition? Data for humans.Animal Behaviour, 40, 997-999. Belsky, J., Steinberg, L., & Draper, P. (1991). Childhood experience, interpersonal development, and reproductive strategy: An evolutionary theory of socialization. Child Development, 62, 647670. Berenbaum, S. A. (1999). Effects of early androgens on sex-typed activities and interests in adolescents with congenital adrenal hyperplasia. Hormones and Behavior, 35, 102-110. Berenbaum, S. A., & Hines, M. (1992). Early androgens are related to childhood sex-typed toy preferences. Psychological Science, 3, 203-206. Berenbaum, S. A., & Snyder, E. (1995). Early hormonal influences on childhood sex-typed activity and playmate preferences: Implications for the development of sexual orientation. Developmental Psychology, 31, 31-42. Berman, P. W. (1986). Young children's responses to babies: Do they foreshadow differences between maternal and paternal styles? In A. Fogel and G. E Melson (Eds.), Origins ofnurturance: Developmental, biological and cultural perspectives on caregiving (pp. 25-51). Hillsdale NJ: Erlbaum. Bjorklund, D. E (1997). The role of immaturity in human development. Psychological Bulletin, 122, 153-169. Bjorklund, D. E, & Pellegrini, A. D. (2002). The origins ofhuman nature: Evolutionary developmental psychology. Washington, DC: American Psychological Association. Blurton Jones, N. G. (1986). Bushman birth spacing: A test for optimal interbirth intervals. Ethology and Sociobiology, 7, 91-105. Blurton Jones, N. G., Hawkes, K., & O'Connell, J. E (1997). Why do Hadza children forage? In N. L. Segal, G. E. Weisfeld, & C. C. Weisfeld (Eds.), Uniting psychology and biology: Integrative perspectives on human development (pp. 279-313). Washington, DC: American Psychological Association.

Sexual Selection and Human Life History

91

Bogin, B. (1999). Evolutionary perspective on human growth. Annual Review of Anthropology, 28, 109-153. Bond, L., Carlin, J. B., Thomas, L., Rubin, K., & Patton, G. (2001). Does bullying cause emotional problems? A prospective study of young teenagers. British Medical Journal, 323, 480-484. Borgerhoff Mulder, M. (1988). Reproductive success in three Kipsigis cohorts. In T. H. CluttonBrock (Ed.), Reproductive success: Studies of individual variation in contrasting breeding systems (pp. 419-435). Chicago, IL: University of Chicago Press. Borgerhoff Mulder, M. (1990). Kipsigis women's preferences for wealthy men: Evidence for female choice in mammals? Behavioral Ecology and Sociobiology, 27, 255-264. Borgia, G. (1985a). Bower destruction and sexual competition in the satin bower bird (Ptilonorhynchus violaceus). Behavioral Ecology and Sociobiology, 18, 91-100. Borgia, G. (1985b). Bower quality, number of decorations and mating success of male satin bower birds (Ptilonorhynchus violaceus): An experimental analysis. Animal Behaviour, 33, 266-271. Borgia, G., & Coleman, S. W. (2000). Co-option of male courtship signals from aggressive display in bowerbirds. Proceedings of the Royal Society of London, Ser. B, 267, 1735-1740. Brooks, R., & Kemp, D. J. (2001). Can older males deliver the good genes? Trends in Ecology & Evolution, 16, 308-313. Buss, D. M. (1989). Sex differences in human mate preferences: Evolutionary hypothesis tested in 37 cultures. Behavioral and Brain Sciences, 12, 1-49. Buss, D. M. (1994). The evolution of desire: Strategies of human mating. NewYork: Basic Books. Buss, D. M. (1996). The evolutionary psychology of human social strategies. In E. T. Higgins & A. E. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 3-38). New York: The Guilford Press. Buss, D. M., Larsen, R. J., & Westen, D. (1996). Sex differences in jealousy: Not gone, not forgotten, and not explained by altemative hypotheses. Psychological Science, 7, 373-377. Buss, D. M., & Schmitt, D. P. (1993). Sexual strategies theory: An evolutionary perspective on human mating. Psychological Review, 100, 204-232. Buss, D. M., & Shackelford, T. K. (1997). From vigilance to violence: Mate retention tactics in married couples. Journal of Personality and Social Psychology, 72, 346-361. Byers, J. A., & Walker, C. (1995). Refining the motor training hypothesis for the evolution of play. American Naturalist, 146, 25-40. Byrne, R. (1995). The thinking ape: Evolutionary origins of intelligence. New York: Oxford University Press. Campbell, A. (1999). Staying alive: Evolution, culture, and women's intrasexual aggression. Behavioral and Brain Science, 22, 203-252. Carranza, J. (1996). Sexual selection for male body mass and the evolution of litter size in mammals. American Naturalist, 148, 81-100. Chagnon, N. A. (1988, February 26). Life histories, blood revenge, and warfare in a tribal population. Science, 239, 985-992. Charlesworth, W. R., & Dzur, C. (1987). Gender comparisons of preschoolers' behavior and resource utilization in group problem-solving. Child Development, 58, 191-200. Charlesworth, W. R., & LaFrenier, P. (1983). Dominance, friendship utilization and resource utilization in preschool children's groups. Ethology and Sociobiology, 4, 175-186. Charnov, E. L. (1993). Life history invariants: Some explorations of symmetry in evolutionary ecology. New York: Oxford University Press. Chen, Z., & Siegler, R. S. (2000). Across the great divide: Bridging the gap between understanding toddlers' and older children's thinking. Monographs of the Society for Research in Child Development, 65, (No. 2, Serial no. 261). Chisholm, J. S. (1993). Death, hope, and sex: Life-history theory and the development of reproductive strategies. Current Anthropology, 34, 1-24.

92

David C. Geary

Clinton, W. L., & Le Boeuf, B. J. (1993). Sexual selection's effects on male life history and the pattern of male mortality. Ecology, 74, 1884-1892. Clutton-Brock, T. H. (1989). Mammalian mating systems. Proceedings of the Royal Society of London, Ser. B, 236, 339-372. Clutton-Brock, T. H. (1991). The evolution of parental care. Princeton, NJ: Princeton University Press. Clutton-Brock, T. H., Albon, S. D., & Guinness, E E. (1985, January 10). Parental investment and sex differences in juvenile mortality in birds and mammals. Nature (London), 313, 131-133. Clutton-Brock, T. H., Harvey, P. H., & Rudder, B. (1977, October 27). Sexual dimorphism, socionomic sex ratio and body weight in primates. Nature (London), 269, 797-800. Clutton-Brock, T. H., & Vincent, A. C. J. (1991, May 2). Sexual selection and the potential reproductive rates of males and females. Nature (London), 351, 58-60. Collaer, M. L., & Hines, M. (1995). Human behavioral sex differences: A role for gonadal hormones during early development? Psychological Bulletin, 118, 55-107. Collis, K., & Borgia, G. (1992). Age-related effects of testosterone, plumage, and experience on aggression and social dominance in juvenile male satin bowerbirds (Ptilonorhynchus violaceus). Auk, 109, 422-434. Conner, J. K. (2001). How strong is natural selection? Trends in Ecology & Evolution, 16, 215-217. Crick, N. R., Casas, J. E, & Mosher, M. (1997). Relational and overt aggression in preschool. Developmental Psychology, 33, 579-588. Crick, N. R., & Rose, A. J. (2000). Toward a gender-balanced approach to the study of social-emotional development: A look at relational aggression. In P. H. Miller & E. K. Scholnick (Eds.), Feminist approaches to developmental psychology (pp. 153-168). New York: Routledge. Crick, N. R., Werner, N. E., Casas, J. E, O'Brien, K. M., Nelson, D. A., Grotpeter, J. K., & Markon, K. (1999). Childhood aggression and gender: A new look at an old problem. In D. Bernstein (Ed.), Nebraska symposium on motivation (Vol. 45, pp. 75-141). Lincoln, NE: University of Nebraska Press. Cunningham, M. R. (1986). Measuring the physical in physical attractiveness: Quasi-experiments on the sociobiology of female beauty. Journal of Personality and Social Psychology, 50, 925-935. Dabbs, J. M., & Dabbs, M. G. (2000). Heroes, rogues, and lovers: Testosterone and behavior. New York: McGraw-Hill. Daly, M., & Wilson, M. (1988). Homicide. New York: Aldine de Gruyter. Darwin, C. (1859). The origin of species by means of natural selection. London: John Murray. Darwin, C. (1871). The descent of man, and selection in relation to sex. London: John Murray. Darwin, C., & Wallace, A. (1858). On the tendency of species to form varieties, and on the perpetuation of varieties and species by natural means of selection. Journal of the Linnean Society of London, Zoology, 3, 45-62. Dawkins, R. (1989). The selfish gene. (2nd ed.) New York: Oxford University Press. de Bruin, J. P., Bovenhuis, H., van Noord, P. A. H., Pearson, P. L., van Arendonk, J. A. M., te Velde, E. R., Kuurman, W. W., & Dorland, M. (2001). The role of genetic factors in age at natural menopause. Human Reproduction, 16, 2014-2018. de Gaudemar, B. (1998). Sexual selection and breeding patterns: Insights from salmonids (Salmonidae). Acta Biotheoretica, 46, 235-251. de Waal, E B. M. (1993). Sex differences in chimpanzee (and human) behavior: A matter of social values? In M. Hechter, L. Nadel, & R. E. Michod (Eds.), The origin of values (pp. 285-303). New York: Aldine de Gruyter. de Waal, F. B. M. (2000, July 28). Primates--A natural heritage of conflict resolution. Science, 289, 586-590. DiPietro, J. A. (1981). Rough and tumble play: A function of gender. Developmental Psychology, 17, 50-58.

Sexual Selection and Human Life History

93

Dixson, A. E, Bossi, T., & Wickings, E. J. (1993). Male dominance and genetically determined reproductive success in the mandrill (Mandrillus sphinx). Primates, 34, 525-532. Draper, P., & Harpending, H. (1988). A sociobiological perspective on the development of human reproductive strategies. In K. B. MacDonald (Ed.), Sociobiologicalperspectives on human development (pp. 340-372). New York: Springer-Verlag. Dunbar, R. I. M. (1993). Coevolution of neocortical size, group size and language in humans. Behavioral and Brain Sciences, 16, 681-735. Dunbar, R. I. M. (1995). The mating system of callitrichid primates: I. Conditions for the coevolution of pair bonding and twinning. Animal Behaviour, 50, 1057-1070. Eagly, A. H. (1987). Sex differences in social behavior: A social-role interpretation. Hillsdale, NJ: Erlbaum. Eaton, W. O., & Enns, L. R. (1986). Sex differences in human motor activity level. Psychological Bulletin, 100, 19-28. Edwards, C. P., & Whiting, B. B. (1993). "Mother, older sibling and me": The overlapping roles of caregivers and companions in the social world of two- to three-year-olds in Ngeca, Kenya. In K. MacDonald (Ed.) Parent-child play: Descriptions & implications (pp. 305-329). Albany, NY: State University of New York Press. Eibl-Eibesfeldt, I. (1989). Human ethology. New York: Aldine de Gruyter. Ellis, B. J., McFadyen-Ketchum, S., Dodge, K. A., Pettit, G. S., & Bates, J. E. (1999). Quality of early family relationships and individual differences in the timing of pubertal maturation in girls: A longitudinal test of an evolutionary model. Journal of Personality and Social Psychology, 77, 387-401. Ember, C. R. (1978). Myths about hunter-gatherers. Ethnology, 17, 439-448. Emlen, S. T. (1995). An evolutionary theory of family. Proceedings ofthe NationalAcademy of Sciences USA, 92, 8092-8099. Emlen, S. T., & Oring, L. W. (1977, July 15). Ecology, sexual selection, and the evolution of mating systems. Science, 197, 215-223. Endler, J. A. (1986). Natural selection in the wild. Princeton, NJ: Princeton University Press. Ergon, T., Lambin, X., & Stenseth, N. C. (2001, June 28). Life-history traits of voles in a fluctuating population respond to the immediate environment. Nature (London), 411, 1043-1045. Feingold, A. (1994). Gender differences in personality: A meta-analysis. Psychological Bulletin, 116, 429-456. Fenner, M. (1998). The phenology of growth and reproduction in plants. Perspectives in Plant Ecology, Evolution and Systematics, 1, 78-91. Flinn, M. V. (1999). Family environment, stress, and health during childhood. In C. Panter-Brick & C. Worthman (Eds.), Hormones, health, and behavior (pp. 105-138). Cambridge, MA: Cambridge University Press. Flinn, M. V., & Low, B. S. (1986). Resource distribution, social competition, and mating patterns in human societies. In D. I. Rubenstein & R. W. Wrangham (Eds.), Ecological aspects of social evolution: Birds and mammals (pp. 217-243). Princeton, NJ: Princeton University Press. Flinn, M. V., Quinlan, R. J., Decker, S. A., Turner, M. T., & England, B. G. (1996). Male-female differences in effects of parental absence on glucocorticoid stress response. Human Nature, 7, 125-162. Folstad, I., & Karter, A. J. (1992). Parasites, bright males, and the immunocompetence handicap. American Naturalist, 139, 603-622. Freedman, D. G. (1974). Human infancy: An evolutionary perspective. New York: Wiley. Gangestad, S. W., Bennett, K. L., & Thornhill, R. (2001). A latent variable model of developmental instability in relation to men's sexual behavior. Proceedings of the Royal Society of London, Ser. B, 268, 1677-1684.

94

David C. Geary

Gangestad, S. W., & Thornhill, R. (1998). Menstrual cycle variation in women's preferences for the scent of symmetrical men. Proceedings of the Royal Society of London, Ser. B, 265, 927-933. Gaulin, S. J. C., & Boster, J. S. (1990). Dowry as female competition. American Anthropologist, 92, 994-1005. Gaulin, S. J. C., & Boster, J. S. (1992). Human marriage systems and sexual dimorphism in stature. American Journal of Physical Anthropology, 89, 467-475. Gaulin, S. J. C., & Fitzgerald, R. W. (1986). Sex differences in spatial ability: An evolutionary hypothesis and test. American Naturalist 127, 74-88. Gaulin, S. J. C., & Fitzgerald, R. W. (1989). Sexual selection for spatial-learning ability. Animal Behaviour, 37, 322-331. Geary, D. C. (1998). Male, female: The evolution of human sex differences. Washington, DC: American Psychological Association. Geary, D. C. (1999). Evolution and developmental sex differences. Current Directions in Psychological Science, 8, 115-120. Geary, D. C. (2000). Evolution and proximate expression of human paternal investment. Psychological Bulletin, 126, 55-77. Geary, D. C. (2002). Sexual selection and sex differences in social cognition. In A. V. McGillicuddy-De Lisi & R. De Lisi (Eds.), Biology, society, and behavior: The development of sex differences in cognition (pp. 23-53). Greenwich, CT: Ablex/Greenwood. Geary, D. C., & Bjorklund, D. F. (2000). Evolutionary developmental psychology. Child Development, 71, 57-65. Geary, D. C., DeSoto, M. C., Hoard, M. K., Sheldon, M. S., & Cooper, L. (2001). Estrogens and relationship jealousy. Human Nature, 12, 299-320. Geary, D. C., & Flinn, M. V. (2001). Evolution of human parental behavior and the human family. Parenting: Science and Practice, 1, 5-61. Geary, D. C., & Flinn, M. V. (2002). Sex differences in behavioral and hormonal response to social threat: Commentary on Taylor et al. (2000). Psychological Review, 109. Geary, D. C., & Huffman, K. J. (2002). Brain and cognitive evolution: Forms of modularity and functions of mind. Psychological Bulletin, 128, 667-698. Gelman, R. (1990). First principles organize attention to and learning about relevant data: Number and animate-inanimate distinction as examples. Cognitive Science, 14, 79-106. Gelman, R., & Williams, E. M. (1998). Enabling constraints for cognitive development and learning: Domain-specificity and epigenesis. In D. Kuhl & R. S. Siegler (Eds.), Cognition, perception, and language Vol. 2 (pp. 575-630). W. Damon (Gen. Ed.), Handbook of child psychology (5th Ed.). New York: Wiley. Gilger, J. W., Geary, D. C., & Eisele, L. M. (1991). Reliability and validity of retrospective selfreports of the age of pubertal onset using twin, sibling, and college student data. Adolescence, 26, 41-53. Gilliard, E. T. (1969). Birds ofparadise and bower birds. London: Weidenfeld and Nicolson. Goldberg, S., Blumberg, S. L., & Kriger, A. (1982). Menarche and interest in infants: Biological and social influences. Child Development, 53, 1544-1550. Goodall, J. (1986). The chimpanzees of Gombe: Patterns of behavior Cambridge, MA: The Belknap Press. Gross, M. R. (1985, January 3). Disruptive selection for alternative life histories in salmon. Nature (London), 313, 47-48. Guttentag, M., & Secord, P. (1983). Too many women? Beverly Hills, CA: Sage. Hamilton, W. D. (1964). The genetical evolution of social behaviour. II. Journal of Theoretical Biology, 7, 17-52. Hamilton, W. D., & Zuk, M. (1982, October 22). Heritable true fitness and bright birds: A role for parasites? Science, 218, 384-387.

Sexual Selection and Human Life History

95

Harris, J. R. (1995). Where is the child's environment? A group socialization theory of development. Psychological Review, 102, 458-489. Harvey, P. H., & Clutton-Brock, T. H. (1985). Life history variation in primates. Evolution, 39, 559581. Hassrick, R. B. (1964). The Sioux: Life and customs of a warrior society. Norman, OK: University of Oklahoma Press. Hawkes, K., O'Connell, J. E, Blurton Jones, N. G., Alvarez, H., & Charnov, E. L. (1998). Grandmothering, menopause, and the evolution of human life histories. Proceedings of the National Academy of Sciences USA, 95, 1336-1339. Herman-Giddens, M. E., Wang, L., & Koch, G. (2001). Secondary sexual characteristics in boys: Estimates from the National Health and Nutrition Examination Survey III, 1988-1994. Archives of Pediatrics and Adolescent Medicine, 155, 1022-1028. Herskind, A. M., McGue, M., Holm, N. V., SCrensen, T. I. A., Harvald, B., & Vaupel, J. W. (1996). The heritability of human longevity: A population-based study of 2872 Danish twin pairs born 1870-1900. Human Genetics, 97, 319-323. Hewlett, B. S. (1992). Husband-wife reciprocity and the father-infant relationship among Aka pygmies. In B. S. Hewlett (Ed.), Father-child relations: Cultural and biosocial contexts (pp. 153-176). New York: Aldine de Gruyter. Hill, K., Boesch, C., Goodall, J., Pusey, A., Williams, J., & Wrangham, R. (2001). Mortality rates among wild chimpanzees. Journal of Human Evolution, 40, 437-450. Hill, K., & Hurtado, A. M. (1996). Ache life history: The ecology and demography of a foraging people. New York: Aldine de Gruyter. Hill, K., & Kaplan, H. (1999). Life history traits in humans: Theory and empirical studies. Annual Review of Anthropology, 28, 397-430. Hines, M., & Kaufrnan, E R. (1994). Androgens and the development of human sex-typical behavior: Rough-and-tumble play and sex of preferred playmates in children with congenital adrenal hyperplasia (CAH). Child Development, 65, 1042-1053. Hofer, M. A. (1987). Early social relationships: A psychobiologist's view. Child Development, 58, 633-647. Hrglund, J., & Alatalo, R. V. (1995). Leks. Princeton, NJ: Princeton University Press. Hrglund, J., & Sheldon, B. C. (1998). The cost of reproduction and sexual selection. Oikos, 83, 478483. Horowitz, D. L. (2001). The deadly ethnic riot. Berkeley, CA: University of California Press. Irons, W. (1979). Cultural and biological success. In N. A. Chagnon & W. Irons (Eds. ), Natural selection and social behavior (pp. 257-272). North Scituate, MA: Duxbury Press. Jennings, K. D. (1975). People versus object orientation, social behavior, and intellectual abilities in preschool children. Developmental Psychology, 11, 511-519. Joffe, T. H. (1997). Social pressures have selected for an extended juvenile period in primates. Journal of Human Evolution, 32, 593-605. Kaar, P., & Jokela, J. (1998). Natural selection on age-specific fertilities in human females: Comparison of individual-level fitness measures. Proceedings of the Royal Society of London, Ser. B, 265, 24152420. Kaplan, H., Hill, K., Lancaster, J., & Hurtado, A. M. (2000). A theory of human life history evolution: Diet, intelligence, and longevity. Evolutionary Anthropology, 9, 156-185. Keeley, L. H. (1996). War before civilization: The myth of the peaceful savage. New York: Oxford University Press. Kelly, S., & Dunbar, R. I. M. (2001). Who dares, wins: Heroism versus altruism in women's mate choice. Human Nature, 12, 89-105. Kenrick, D. T., & Keefe, R. C. (1992). Age preferences in mates reflect sex differences in human reproductive strategies. Behavioral and Brain Sciences, 15, 75-133.

96

David C. Geary

Kenrick, D. T., & Luce, C. L. (2000). An evolutionary life-history model of gender differences and similarities. In T. Eckes & H. M. Trautner (Eds.), The developmental social psychology of gender (pp. 35-63). Mahwah, NJ: Erlbaum. Kimura, D. (1999). Sex and cognition. Cambridge, MA: MIT/Bradford Books. Kingsolver, J. G., Hoekstra, H. E., Hoekstra, J. M., Berrigan, D., Vignieri, S. N., Hill, C. E., Hoang, A., Gibert, P., & Beerli, P. (2001 ). The strength of phenotypic selection in natural populations. American Naturalist, 157, 245-261. Kirk, K. M., Blomberg, S. P., Duffy, D. L., Heath, A. C., Owens, I. P. F., & Martin, N. G. (2001). Natural selection and quantitative genetics of life-history traits in Western women: A twin study. Evolution, 55, 423-435. Knight, G. P., & Chao, C.-C. (1989). Gender differences in the cooperative, competitive, and individualistic social values of children. Motivation and Emotion, 13, 125-141. Kokko, H. (1997). Evolutionarily stable strategies of age-dependent sexual advertisement. Behavioral Ecology and Sociobiology, 41, 99-107. Kolakowski, D., & Malina, R. M. (1974, October 4). Spatial ability, throwing accuracy and man's hunting heritage. Nature (London), 251, 410-412. Korpelainen, H. (2000). Fitness, reproduction and longevity among European aristocratic and rural Finnish families in the 1700s and 1800s. Proceedings of the Royal Society of London, Ser. B, 267, 1765-1770. Krebs, J. R., & Davies, N. B. (1993). An introduction to behavioural ecology (3rd ed.). Oxford: Blackwell Science. Lancaster, J. B., & Lancaster, C. S. (1987). The watershed: Change in parental-investment and familyformation strategies in the course of human evolution. In J. B. Lancaster, J. Altmann, A. S. Rossi, & L. R. Sherrod (Eds.), Parenting across the life span: Biosocial dimensions (pp. 187-205). New York: Aldine de Gruyter. Leadbeater, B. J., Blatt, S. J., & Quinlan, D. M. (1995). Gender-linked vulnerabilities to depressive symptoms, stress, and problem behaviors in adolescents. Journal of Research on Adolescence, 5, 1-29. Le Boeuf, B. J., & Reiter, J. (1988). Lifetime reproductive success in northern elephant seals. In T. H. Clutton-Brock (Ed.), Reproductive success: Studies of individual variation in contrasting breeding systems (pp. 344-362). Chicago, IL: University of Chicago Press. Lee, P. C., & Bowman, J. E. (1995). Influence of ecology and energetics on primate mothers and infants. In C. R. Pryce, R. D. Martin, & D. Skuse (Eds.), Motherhood in human and nonhuman primates: Biosocial determinants (pp. 47-58). Basel: Karger. Leigh, S. R. (1995). Socioecology and the ontogeny of sexual size dimorphism in anthropoid primates. American Journal of Physical Anthropology, 97, 339-356. Leigh, S. R. (1996). Evolution of human growth spurts. American Journal of Physical Anthropology, 101, 455-474. Leslie, A. M. (1987). Pretense and representation: The origins of "theory of mind." Psychological Review, 94, 412-426. Lever, J. (1978). Sex differences in the complexity of children' s play and games. American Sociological Review, 43, 471-483. Leveroni, C., & Berenbaum, S. A. (1998). Early androgen effects on interest in infants: Evidence from children with congenital adrenal hyperplasia. Developmental Neuropsychology, 14, 321340. Low, B. S. (1989). Cross-cultural patterns in the training of children: An evolutionary perspective. Journal of Comparative Psychology, 103, 311-319. Loy, J. W., & Hesketh, G. L. (1995). Competitive play on the plains: An analysis of games and warfare among Native American warrior societies, 1800-1850. In A. D. Pellegrini (Ed.), The future of

Sexual Selection and Human Life History

97

play theory: A multidisciplinary inquiry into the contributions of Brian Sutton-Smith (pp. 73-105). Albany, NY: State University of New York Press. Lummaa, V. (2001). Reproductive investment in pre-industrial humans: The consequences of offspring number, gender and survival. Proceedings of the Royal Society of London, Ser. B, 268, 1-7. Mac Arthur, R. H., & Wilson, E. O. (1967). The theory ofisland biogeography. Princeton, NJ: Princeton University Press. Maccoby, E. E. (1988). Gender as a social category. Developmental Psychology, 24, 755-765. Maccoby, E. E. (1998). The two sexes: Growing up apart, coming together. Cambridge, MA: Belknap Press. Maccoby, E. E., & Jacklin, C. N. (1987). Gender segregation in childhood. In H. W. Reese (Ed.), Advances in child development and behavior (Vol. 20, pp. 239-287). New York: Academic Press. MacDonald, K. (1992). Warmth as a developmental construct: An evolutionary analysis. Child Development, 63, 753-773. Mandler, J. M. (1992). How to build a baby: II., Conceptual primitives. Psychological Review, 99, 587-604. Manning, J. T. (1989). Age-advertisement and the evolution of the peacock's train. Journal of Evolutionary Biology, 2, 379-384. Martorell, R., Rivera, J., Kaplowitz, H., & Pollitt, E. (1992). Long-term consequences of growth retardation during early childhood. In M. Hern~indez and J. Argente (Eds.), Human growth: Basic and clinical aspects (pp. 143-149). Amsterdam: Elsevier Science. Publishers. Matthews, M. H. (1987). Sex differences in spatial competence: The ability of young children to map "primed" unfamiliar environments. Educational Psychology, 7, 77-90. Matthews, M. H. (1992). Making sense ofplace : Children's understanding of large-scale environments. Savage, MD: Barnes and Noble Books. Mayr, E. (1982). The growth of biological thought. Cambridge, MA: Belknap Press. Mayr, E. (1974). Behavior programs and evolutionary strategies. American Scientist, 62, 650-659. McElligott, A. G., & Hayden, T. J. (2000). Lifetime mating success, sexual selection and life history of fallow bucks (Dama dama). Behavioral Ecology and Sociobiology, 48, 203-210. McHenry, H. M. (1991). Sexual dimorphism in Australopithecus afarensis. Journal of Human Evolution, 20, 21-32. McLaren, I. A. (1966). Adaptive significance of large size and long life of the chaetognath Sagitta elegans in the arctic. Ecology, 47, 852-855. McNamara, J. M., & Houston, A. I. (1996, March 21). State-dependent life histories. Nature (London), 380, 215-221. Menken, J., Trussell, J., & Larsen, U. (1986, September 26). Age and infertility. Science, 233, 13891394. Miaud, C., Guyrtant, R., & Elmberg, J. (1999). Variations in life-history traits in the common frog Rana temporaria (Amphibia: Anura): A literature review and new data from the French alps. Journal of Zoology, 249, 61-73. MOiler, A. P. (1994). Sexual selection and the barn swallow. New York: Oxford University Press. MOiler, A. P., Soler, M., & Thornhill, R. (1995). Breast asymmetry, sexual selection, and human reproductive success. Ethology and Sociobiology, 16, 207-219. Morris, E O. (1891). A history of British birds. London: Nimmo. Mousseau, T. A., & Roff, D. A. (1987). Natural selection and the heritability of fitness components. Heredity, 59, 181-197. Munroe, R. L., & Munroe, R. H. (1971). Effect of environmental experience on spatial ability in an East African society. Journal of Social Psychology, 83, 15-22. Murdock, G. P. (1949). Atlas of world cultures. Pittsburgh, PA: University of Pittsburgh Press.

98

David C. Geary

Newcomb, A. E, Bukowski, W. M., & Pattee, L. (1993). Children's peer relations: A meta-analytic review of popular, rejected, neglected, controversial, and average sociometric status. Psychological Bulletin, 113, 99-128. Nicolson, N. A. (1987). Infants, mothers, and other females. In B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, & T. T. Struhsaker (Eds.), Primate societies (pp. 330-342). Chicago, IL: The University of Chicago Press. Nordling, D., Andersson, M., Zohari, S., & Gustafsson, L. (1998). Reproductive effort reduces specific immune response and parasite resistance. Proceedings of the Royal Society of London, Ser. B, 265, 1291-1298. Nowicki, S., Peters, S., & Podos, J. (1998). Song learning, early nutrition and sexual selection in song birds. American Zoologist, 38, 179-190. Oliver, M. B., & Hyde, J. S. (1993). Gender differences in sexuality: A meta-analysis. Psychological Bulletin, 114, 29-51. Parker, G. A., & Simmons, L. W. (1996). Parental investment and the control of selection: Predicting the direction of sexual competition. Proceedings of the Royal Society of London, Ser. B, 263, 315-321. Partridge, L., & Endler, J. A. (1987). Life history constraints on sexual selection. In J. W. Bradbury & M. B. Andersson (Eds.), Sexual selection: Testing the alternatives (pp. 265-277). New York: Wiley. Pasternak, B., Ember, C. R., & Ember, M. (1997). Sex, gender, and kinship: A cross-culturalperspective. Upper Saddle River, N J: Prentice-Hall. Pellegrini, A. D., & Bartini, M. (2001). Dominance in early adolescent boys: Affiliative and aggressive dimensions and possible functions. Merrill-Palmer Quarterly, 47, 142-163. Pellegrini, A. D., & Smith, P. K. (1998). Physical activity play: The nature and function of a neglected aspect of play. Child Development, 69, 577-598. Prrusse, D. (1993). Cultural and reproductive success in industrialized societies: Testing the relationship at the proximate and ultimate levels. Behavioral and Brain Sciences, 16, 267-322. Petrie, M. (1994, October 13). Improved growth and survival of offspring of peacocks with more elaborate trains. Nature (London), 371, 598-599. Petrie, M., Halliday, T., & Sanders, C. (1991). Peahens prefer peacocks with elaborate trains. Animal Behavior, 41, 323-331. Pinker, S. (1994). The language instinct. New York: William Morrow. Pitcher, E. G., & Schultz, L. H. (1983). Boys and girls at play: The development of sex roles. South Hadley, MA: Bergin and Garvey. Plavcan, J. M., & van Schaik, C. P. (1997a). Intrasexual competition and body weight dimorphism in anthropoid primates. American Journal of Physical Anthropology, 103, 37-68. Plavcan, J. M., & van Schaik, C. P. (1997b). Interpreting hominid behavior on the basis of sexual dimorphism. Journal of Human Evolution, 32, 345-374. Plomin, R., DeFries, J. C., McClearn, G. E., & McGuffin, P. (2001). Behavioral genetics (4th ed.). New York: Worth. Price, G. R. (1970, August 1). Selection and covariance. Nature (London), 227, 520-521. Pryce, C. R. (1993). The regulation of maternal behaviour in marmosets and tamarins. Behavioural Processes, 30, 201-224. Pryce, C. R. (1995). Determinants of motherhood in human and nonhuman primates: A biosocial model. In C. R. Pryce & R. D. Martin, & D. Skuse (Eds.), Motherhood in human and nonhuman primates: Biosocial determinants (pp. 1-15). Basel: Karger. Resnick, S. M., Berenbaum, S. A., Gottesman, I. I., & Bouchard, Jr., T. J. (1986). Early hormonal influences on congnitive functioning in congenital adrenal hyperplasia. Developmental Psychology, 22, 191-198. Reynolds, J. D. (1987). Mating system and nesting biology of the Red-necked Phalarope Phalaropus lobatus: What constrains polyandry? Isis, 129, 225-242.

Sexual Selection and Human Life History

99

Reynolds, J. D., & Szrkely, T. (1997). The evolution of parental care in shorebirds: Life histories, ecology, and sexual selection. Behavioral Ecology, 8, 126-134. Reznick, D. (1985). Costs of reproduction: An evaluation of the empirical evidence. Oikos, 44, 257-267. Reznick, D. N. (1992). Measuring the costs of reproduction. Trends in Ecology and Evolution, 7, 42-45. Reznick, D. N., & Bryga, H. (1987). Life-history evolution in guppies Poecilia reticulata 1. Phenotypic and genetic changes in an introduction experiment. Evolution, 41, 1370-1385. Reznick, D. N., & Bryga, H. A. (1996). Life-history evolution in guppies (Poecilia reticulata: Poeciliidae). V., Genetic basis of parallelism in life histories. American Naturalist, 147, 339-359. Reznick, D., & Endler, J. A. (1982). The impact of predation on life history evolution in Trinidadian guppies (Poecilia reticulata). Evolution, 36, 160-177. Reznick, D., Nunney, L., & Tessier, A. (2000). Big houses, big cars, superfleas and the costs of reproduction. Trends in Ecology & Evolution, 15, 421-425. Reznick, D. N., Shaw, E H., Rodd, F. H., & Shaw, R. G. (1997, March 28). Evaluation of the rate of evolution in natural populations of guppies (Poecilia reticulata). Science, 275, 1934-1937. Rilling, J. K., & Insel, T. R. (1999). The primate neocortex in comparative perspective using magnetic resonance imaging. Journal of Human Evolution, 37, 191-223. Rodd, E H., Reznick, D. N., & Skolowski, M. B. (1997). Phenotypic plasticity in life history traits of guppies: Responses to social environment. Ecology, 78, 419-433. Roff, D. A. (1992). The evolution of life histories: Theory and analysis. New York: Chapman & Hall. Rohwer, S., Fretwell, S. D., & Niles, D. M. (1980). Delayed maturation in passerine plumages and the deceptive acquisition of resources. American Naturalist, 115, 400-437. Rose, A. J., & Asher, S. R. (1999). Children's goals and strategies in response to conflicts within a friendship. Developmental Psychology, 35, 69-79. Saino, N., Bolzern, A. M., & MOiler, A. P. (1997). Immunocompetence, ornamentation, and viability of male barn swallows (Hirundo rustica). Proceedings of the National Academy of Sciences, USA, 94, 549-552. Saino, N., & Moiler, A. P. (1994). Secondary sexual characteristics, parasites and testosterone in the barn swallow, Hirundo rustica. Animal Behaviour, 48, 1325-1333. Saino, N., MOiler, A. P., & Bolzern, A. M. (1995). Testosterone effects on the immune system and parasite infestations in the barn swallow (Hirundo rustica): An experimental test of the immunocompetence hypothesis. Behavioral Ecology, 6, 397-404. Sandberg, D. E., & Meyer-Bahlburg, H. E L. (1994). Variability in middle childhood play behavior: Effects of gender, age, and family background. Archives of Sexual Behavior, 23, 645-663. Savin-Williams, R. C. (1987). Adolescence: An ethological perspective. New York: Springer-Verlag. Sawaguchi, T. (1997). Possible involvement of sexual selection in neocortical evolution of monkeys and apes. Folia Primatologica, 68, 95-99. Scutt, D., & Manning, J. T. (1996). Symmetry and ovulation in women. Human Reproduction, 11, 2477-2480. Seielstad, M. T., Minch, E., & Cavalli-Sforza, L. L. (1998). Genetic evidence for a higher female migration rate in humans. Nature Genetics, 20, 278-280. Selander, R. K. (1965). On mating systems and sexual selection. American Naturalist, 99, 129-141. Shackelford, T. K., & Larsen, R. J. (1997). Facial asymmetry as an indicator of psychological, emotional, and physiological distress. Journal of Personality and Social Psychology, 72, 456-466. Sheldon, B. C., Meril~i, J., Qvarnstr6m, A., Gustafsson, L., & Ellegren, H. (1997). Paternal genetic contribution to offspring condition predicted by size of male secondary sexual character. Proceedings of the Royal Society of London, Ser. B, 264, 297-302. Shine, R. (1978). Propagule size and parental care: The "safe harbor" hypothesis. Journal of Theoretical Biology, 75, 417-424. Shine, R. (1989). Alternative models for the evolution of offspring size. American Naturalist, 134, 311-317.

1O0

David C. Geary

Silk, J. B. (1993). The evolution of social conflict among female primates. In W. A. Mason and S. P. Mendoza (Eds.), Primate social conflict (pp. 49-83). Albany, NY: State University of New York Press. Silverman, I., & Eals, M. (1992). Sex differences in spatial abilities: Evolutionary theory and data. In J. H. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture (pp. 533-549). New York: Oxford University Press. Sinervo, B., & Svensson, E. (1998). Mechanistic and selective causes of life history trade-offs and plasticity. Oikos, 83, 432-442. Singh, D. (1993). Adaptive significance of female physical attractiveness: Role of waist-to-hip ratio. Journal of Personality and Social Psychology, 65, 293-307. Smith, E K. (1982). Does play matter? Functional and evolutionary aspects of animal and human play. Behavioral and Brain Sciences, 5, 139-184. Smuts, B. B. (1987). Gender, aggression, and influence. In B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, & T. T. Struhsaker (Eds.), Primate societies (pp. 400-412). Chicago, IL: The University of Chicago Press. Stamps, J. A. (1995). Using growth-based models to study behavioral factors affecting sexual size dimorphism. Herpetological Monographs, 9, 75-87. Stearns, S. C. (1992). The evolution of life histories. New York: Oxford University Press. Stearns, S. C., & Koella, J. C. (1986). The evolution of phenotypic plasticity in life-history traits: Predictions of reaction norms for age and size at maturity. Evolution, 40, 893-913. Strayer, E E, & Santos, A. J. (1996). Affiliative structures in preschool peer groups. Social Development, 5, 117-130. Sultan, S. E. (2000). Phenotypic plasticity for plant development, function and life history. Trends in Plant Science, 5, 537-542. Sutton-Smith, B., Rosenberg, B. G., & Morgan, Jr., E. E (1963). Development of sex differences in play choices during preadolescence. Child Development, 34, 119-126. Svensson, E., & Sheldon, B. C. (1998). The social context of life history evolution. Oikos, 83, 466477. Symons, D. (1979). The evolution of human sexuality. New York: Oxford University Press. Tanner, J. M. (1990). Foetus into man: Physical growth from conception to maturity. Cambridge, MA: Harvard University Press. Tanner, J. M. (1992). Human growth and development. In S. Jones, R. Martin, & D. Pilbeam (Eds.), The Cambridge encyclopedia of human evolution (pp. 98-105). New York: Cambridge University Press. Taylor, S. E., Klein, L. C., Lewis, B. P., Gruenewald, T. L., Gurung, R. A. R., & Updegraff, J. A. (2000). Biobehavioral responses to stress in females: Tend-and-befriend, not fight-or-flight. Psychological Review, 107, 411-429. Thomas, J. R., & French, K. E. (1985). Gender differences across age in motor performance: A metaanalysis. Psychological Bulletin, 98, 260-282. Tiger, L. (1969). Men in groups. New York: Random House. Trivers, R. L. (1972). Parental investment and sexual selection. In B. Campbell (Ed.), Sexual selection and the descent of man 1871-1971 (pp. 136-179). Chicago, IL: Aldine. Trivers, R. L. (1974). Parent--offspring conflict. American Zoologist, 14, 249-264. Turner, E J., & Gervai, J. (1995). A multidimensional study of gender typing in preschool children and their parents: Personality, attitudes, preferences, behavior, and cultural differences. Developmental Psychology, 31, 759-772. Vellenga, R. E. (1980). Molts of the satin bowerbird Ptilonorhynchus violaceus. Emu, 80, 49-54. Walter, A., & Dittami, J. (1997). Reproductive suppression in male alpine marmots. Animal Behaviour, 53, 53-66.

Sexual Selection and Human Life History

101

Waiters, J. R., & Seyfarth, R. M. (1987). Conflict and cooperation. In B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, & T. T. Struhsaker (Eds.), Primate societies (pp. 306-317). Chicago, IL: The University of Chicago Press. Watson, N. V., & Kimura, D. (1991). Nontrivial sex differences in throwing and intercepting: Relation to psychometrically-defined spatial functions. Personality and Individual Differences, 12, 375385. West, G. B., Brown, J. H., & Enquist, B. J. (2001, October 11). A general model for ontogenetic growth. Nature, 413, 628-631. Westendorp, R. G. J., & Kirkwood, T. B. L. (1998, December 31). Human longevity at the cost of reproductive success. Nature, 396, 743-746. Whiting, B. B., & Edwards, C. P. (1973). A cross-cultural analysis of sex differences in the behavior of children aged 3 through 11. Journal of Social Psychology, 91, 171-188. Whiting, B. B., & Edwards, C. P. (1988). Children of different worlds: The formation ofsocial behavior Cambridge, MA: Harvard University Press. Wiley, R. H. (1974). Evolution of social organization and life-history pattems among grouse. Quarterly Review of Biology, 49, 201-227. Wilkins, K. (1996, July). Causes of death: How the sexes differ. Canadian Social Trends, 11-17. (Catalogue No. 11-008) Williams, G. C. (1957). Pleiotropy, natural selection and the evolution of senescence. Evolution, 11, 398-411. Williams, G. C. (1966). Adaptation and natural selection: A critique of some current evolutionary thought. Princeton, N J: Princeton University Press. Williams, G. C. (1975). Sex and evolution. Princeton, NJ: Princeton University Press. Willingham, W. W., & Cole, N. S. (1997). Gender and fair assessment. Mahwah, NJ: Erlbaum. Wilson, M., & Daly, M. (1985). Competitiveness, risk taking, and violence: The young male syndrome. Ethology and Sociobiology, 6, 59-73. Wilson, M., & Daly, M. (1997). Life expectancy, economic inequality, homicide, and reproductive timing in Chicago neighbourhoods. British Medical Journal 314, 1271-1274. Winemiller, K. O. (1992). Life-history strategies and the effectiveness of sexual selection. Oikos, 63, 318-327. Wittenberger, J. E (1979). A model for delayed reproduction in iteroparous animals. American Naturalist, 114, 439-446. Wizemann, T. M., & Pardue, M.-L. (Eds.) (2001). Exploring the biological contributions to human health: Does sex matter? Washington, DC: National Academy Press. Wood, J. W. (1994). Dynamics of human reproduction: Biology, biometry, and demography. New York: Aldine de Gruyter. Wrangham, R. W. (1980). An ecological model of female-bonded primate groups. Behaviour, 75, 262-300. Wrangham, R. W. (1999). Evolution of coalitionary killing. Yearbook of Physical Anthropology, 42, 1-30. Zahavi, A. (1975). Mate selectionmA selection for a handicap. Journal of Theoretical Biology, 53, 205-214. Zuk, M., Thornhill, R., & Ligon, J. D. (1990). Parasites and mate choice in red jungle fowl. American Zoologist, 30, 235-244.

This Page Intentionally Left Blank

DEVELOPMENTS IN EARLY RECALL MEMORY: NORMATIVE TRENDS AND INDIVIDUAL DIFFERENCES

Patricia J. Bauer, Melissa M. Burch, and Erica E. Kleinknecht INSTITUTE OF CHILD DEVELOPMENT UNIVERSITY OF MINNESOTA MINNEAPOLIS, MINNESOTA 55455

I. INITIATING THE STUDY OF E A R L Y R E C A L L M E M O R Y A. ELICITED IMITATION AS A N O N V E R B A L M E A S U R E OF R E C A L L B. THE NEURAL SUBSTRATE OF EXPLICIT M E M O R Y II. C H A R A C T E R I Z I N G R E C A L L M E M O R Y IN THE FIRST TWO YEARS OF LIFE A. D E V E L O P M E N T A L CHANGES IN THE PREVALENCE AND ROBUSTNESS OF LONG-TERM R E C A L L B. E X P E R I M E N T A L MANIPULATIONS AFFECTING INFANTS' AND YOUNG CHILDREN'S R E C A L L III. INDIVIDUAL DIFFERENCES IN LONG-TERM RECALL: CHILDREN'S GENDER, CHILDREN'S L A N G U A G E PROFICIENCY, AND VARIABILITY IN INITIAL LEARNING A. EFFECTS OF CHILDREN'S GENDER B. EFFECTS OF CHILDREN'S PRODUCTIVE AND RECEPTIVE LANGUAGE C. EFFECTS OF INDIVIDUAL VARIABILITY IN INITIAL LEARNING D. S U M M A R Y OF THREE SOURCES OF VARIABILITY IN EARLY RECALL MEMORY IV. INDIVIDUAL DIFFERENCES IN LONG-TERM RECALL: CHILDREN'S T E M P E R A M E N T CHARACTERISTICS A. W H A T IS TEMPERAMENT? B. W H Y MIGHT TEMPERAMENT CHARACTERISTICS BE RELATED TO EARLY R E C A L L M E M O R Y ? C. EVIDENCE OF RELATIONS B E T W E E N TEMPERAMENT CHARACTERISTICS AND M E M O R Y IN INFANCY AND THE PRESCHOOL YEARS D. RELATIONS B E T W E E N T E M P E R A M E N T AND M E M O R Y IN THE 2ND YEAR: A G E - R E L A T E D DIFFERENCES E. RELATIONS B E T W E E N TEMPERAMENT AND MEMORY: DIFFERENCES WITHIN SUBJECTS F. S U M M A R Y

103 ADVANCESIN CHILDDEVELOPMENT AND BEHAVIOR,VOL.30

Copyright2002, ElsevierScience (USA). All rights reserved. 0065-2407/02 $35.00

104

Bauer et al.

V. CHILDREN'STEMPERAMENTAND MOTHERS' LANGUAGEAS INTERACTING SOURCES OF INDIVIDUALDIFFERENCES IN LONG-TERMRECALL A. RELATIONSBETWEENMATERNALLANGUAGEAND OLDER CHILDREN'S MEMORYNARRATIVES B. MATERNALLANGUAGEVARIABILITYWITH YOUNGERCHILDREN IN THE CONTEXTOF ELICITED IMITATION C. RELATIONSBETWEENMOTHERS' LANGUAGE,CHILDREN'S RECALL PERFORMANCE, AND CHILDREN'S TEMPERAMENT CHARACTERISTICS D. SUMMARY VI. CONCLUSIONSAND IMPLICATIONS REFERENCES

The end of the 20th century was witness to nothing short of a sea change in perspective on the mnemonic abilities of infants and very young children. Early in the cognitive era, children younger than 18-24 months were thought to lack the capacity to represent information not available to the senses (i.e., to mentally r e p r e s e n t it; Piaget, 1952). As a consequence, infants were thought to live in a "here and now" world that included physically present entities but which had no future and no past. With the advent of techniques for probing the minds of infants via the eyes of infants (e.g., visual paired comparison: Fantz, 1956), the assumption of an ahistorical infancy began to be questioned. By the early to middle 1970s, it was apparent that infants habituated (Friedman, 1972), could be conditioned (Rovee & Fagen, 1976), and could recognize some types of stimuli (e.g., faces) even after a delay (Fagan, 1973). These demonstrations made clear that even in the first months of life, infants have significant mnemonic competence. Nevertheless, conspicuously absent from the literature was compelling evidence of the capacity for recall and, in particular, for long-term recall. In contrast to recognition, recall involves "... accessing (bringing to awareness) a cognitive structure pertaining to a past experience not currently available to perception" (Mandler, 1984, p. 79). Recall is the process in which we engage when we retrieve from memory what we had for dinner last evening or where we spent the most recent summer vacation. Perhaps especially against the backdrop of demonstrations of other types of memory, the dearth of evidence of recall by infants and young children led to suggestions that this particular mnemonic ability was late to develop. The suggestion that infants and young children lack an ability that most adults take for granted seemingly was supported by at least two other sources: research on adults' memories of childhood and cognitive neuroscience. Moreover, the

Developments in Early Recall

105

assumptions were hard to examine for methodological reasons. Beginning in the mid-1980s, however, the suggestion that long-term recall ability is late to develop was challenged and found wanting. In this chapter, we discuss methodological innovations that enabled examination of recall abilities in pre- and early verbal children. We then review the empirical literature illustrating developmental changes in this important cognitive ability during the first 2 years of life, as well as neurodevelopmental changes that may underlie them. We first focus on normative trends of development in the first 2 years and then on individual differences that influence the normative trends. Specifically, we examine possible influences on early recall memory of children's gender and their levels of proficiency with language as well as children's temperament characteristics. Whereas there appear to be few associations between gender and recall and language and recall, we present evidence that temperament characteristics have direct relations with developing recall memory skills. Children's temperament characteristics also work to influence the verbal support that children's mothers provide them in mnemonic contexts, which in turn affects children's recall performance. Finally, we discuss the implications of the findings for our understanding of early developmental change as well as of later developing episodic and autobiographical memory competence.

I. Initiating the Study of Early Recall Memory One source of the assumption that infants and very young children are unable to recall the past was research on adults' memories of childhood. Since Sigmund Freud first labeled it in 1905, the phenomenon of infantile amnesia has been one of the great curiosities in the field of memory. Infantile or childhood amnesia refers to the relative paucity among adults of verbally accessible memories from the first years of life (e.g., Pillemer & White, 1989) and fewer memories from ages 3 89 years than expected based on forgetting alone (e.g., Rubin, 1982). In his interviews with adult patients, Freud noticed that few had memories from their early years. The memories they did have were sketchy and incomplete. Freud termed this phenomenon infantile amnesiac"the amnesia that veils our earliest youth from us and makes us strangers to it" (1916/1966, p. 326). In the years since Freud identified infantile amnesia, many investigators have studied adults' memories of their childhoods. This work has yielded the robust finding that among adults, in Western culture, the average age of earliest verbalizable memory is 3 5 years (Dudycha & Dudycha, 1941; see West & Bauer, 1999, for a review of contemporary research; see, for example, Mullen, 1994, for different estimates of earliest memories among non-Western adults). Over the course of his career, Freud developed two hypotheses regarding the source of infantile amnesia. The most widely known is his hypothesis that early

106

B a u e r et al.

memories exist but are repressed (Freud, 1916/1966). His other suggestion was that early memories are mere fragments of experiences that are not organized in a coherent, adultlike fashion (Freud, 1905/1953). Piaget's (1952) characterization of the first 18-24 months of life as a period devoid of a means to represent absent objects (and events) was a natural ally for this suggestion. Also consistent were early conclusions (based largely on research with rats; see C. A. Nelson, 1995, 1997, for discussion) that the neural substrate supporting recall was functionally late to develop. Even as tenets attributed to Piagetian theory were being challenged (e.g., Gelman & Baillargeon, 1983), and as data suggesting earlier functional maturity of the neural substrate of recall were accumulating (e.g., Schacter & Moscovitch, 1984), the suggestion that infants were unable to recall previous experiences prevailed. Indeed, in the early 1980s, there were virtually no data on long-term recall memory in children younger than age 3 years. From the work of Katherine Nelson and her colleagues (e.g., Nelson & Gruendel, 1981, 1986), we knew that children 3 years and older had impressive long-term recall abilities (e.g., Fivush, 1984; Hudson, 1986; Nelson & Gruendel, 1981). By that age, children evidence well-organized representations of familiar events, such as going to fast-food restaurants. Diary reports and naturalistic observations also revealed evidence of recall memory in preschool-age children (e.g., Ratner, 1980). However, recall of the past by children younger than 3 years had not been examined. The major reason for the absence of research on children younger than 3 years was that conceptually and methodologically, the ability to recall was associated with the ability to provide a verbal report. Because children 3 years and older could use language to talk about the past, they clearly could recall it. However, 1- to 2-year-old children are not particularly adept at language. Even by late in the 2nd and into the 3rd year, when children have made substantial gains in language, they have difficulty using their language to talk about past experiences. Not until children are 3 years of age and older do they become reliable partners in conversations about past events (e.g., Fivush, Gray, & Fromhoff, 1987). This reality, and that the average age of adults' first memories is 3 71 , reinforced the notion that it was not until the 4th year of life that children develop the ability to recall. Such was the backdrop against which research on recall of events by infants and very young children began. The first step in this research was to establish a means of testing recall nonverbally. Imitation-based procedures fit the bill. Instead of being required to tell what they remember, in imitation procedures, children are afforded the opportunity to show what they remember. Specifically, elicited and deferred imitation involve using objects to produce a unique action or sequence of actions, and then allowing the infant or child to imitate the actions either immediately, or after a delay, or both. Examples of stimuli used in imitation-based studies in our laboratory are provided in Table I. Imitation after exposure to a model was viewed by Piaget (1952) as one of the hallmarks of the capacity for representation.

Developments in Early Recall

107

TABLE I Examples of Stimuli Used in Imitation-Based Studies

Example of two-step sequence Sequence: Make Big Bird turn on the light. Materials: Small plastic car containing "Big Bird" character; clear plastic L-shaped base; plastic plunger extended from one end of the horizontal section of the base; small light bulb on end of horizontal section opposite plunger. Steps: (1) Putting the car in the base (i.e., inserting it down the vertical section of the base), and (2) pushing in the plunger, thereby causing the car to roll to the other end and illuminate the light bulb. Example of three-step sequence Sequence: Make a rattle. Materials: Two rectangular nesting cups; wooden block. Steps: (1) Putting the block into one nesting cup, (2) inverting one nesting cup into the other, and (3) shaking the cups to make them rattle. Example offour-step sequence Sequence: Make a gong. Materials: Base resembling the support for a swing set; small cup attached to one post of the swing-set-shaped base; bar resting in the cup; metal plate with a lip; small plastic mallet. Steps: (1) Lifting the bar from the cup, (2) putting the bar across the posts (to form a crosspiece), (3) hanging the plate from the bar, and (4) hitting the plate with the mallet, thereby causing it to ring.

A. ELICITED IMITATION AS A NONVERBAL MEASURE OF RECALL T h e i m i t a t i o n - b a s e d p r o c e d u r e s u s e d in our laboratory involve several steps. First, before s e q u e n c e s of actions are m o d e l e d , the associated objects or props are given to the infant or child for a baseline period, during w h i c h s p o n t a n e o u s p r o d u c t i o n of the target actions and s e q u e n c e s is assessed. After the baseline, e x p e r i m e n t e r s label the s e q u e n c e to be p r o d u c e d and then narrate their actions as they use the props to enact the sequence. In situations in w h i c h i m m e d i a t e recall is assessed, after m o d e l i n g , the e x p e r i m e n t e r returns the props to the infant or child and e n c o u r a g e s imitation. 1 M e m o r y is inferred if imitative levels of p r o d u c t i o n of the target actions and s e q u e n c e s e x c e e d baseline levels. To test retention over time, the infant or child returns to the l a b o r a t o r y after a delay at w h i c h point the props are p r o v i d e d but the s e q u e n c e s are not r e m o d e l e d . In s o m e studies, verbal r e m i n d e r s 1 In some applications of the imitation task, the opportunity to imitate is deferred for a period of hours (e.g., Meltzoff, 1988b) to months (e.g., Meltzoff, 1995). That is, children are not permitted to imitate prior to imposition of a delay. Although elicited-imitation protocols that permit imitation prior to the delay and those that do not (i.e., deferred-imitation protocols) might by thought to test different underlying abilities (Piaget, 1952), apparently they do not (see Bauer, Wenner, Dropik, & Wewerka, 2000, for extended discussion). Because the term "elicited imitation" characterized protocols both that require deferred imitation and that do not, unless the distinction between protocol types is relevant, for the balance of this chapter we will use the more generic term to refer to the task.

108

Bauer et al.

that the child could, for example, "Use this stuff to make a rattle," are given along with the props; in other studies, only the props serve to remind the children of the events. Higher levels of performance after the delay, relative to baseline, are taken to indicate long-term memory. From each phase of testing we determine both the number of different individual steps or actions of the event produced and, as a measure of organization of the event representation, the number of pairs of steps or actions produced in the target order. There are excellent reasons to believe that elicited imitation serves as a nonverbal analog to verbal report. One very telling observation in this regard is that once children have the linguistic capacity to do so, they talk about the events that they experienced in the context of imitation (Bauer, Kroupina, Schwade, Dropik, & Wewerka, 1998; Bauer, Wenner, & Kroupina, 2002; Bauer & Wewerka, 1995, 1997). That children subsequently are able to talk about events experienced in the elicited-imitation paradigm is strong evidence that the representational format in which the memories are encoded is compatible with language if not itself symbolic (see Mandler, 1998, for further discussion of representational formats). Ideally, the logical argument that elicited imitation is a nonverbal analog to a verbal report would be complemented by evidence showing that conditions that undermine verbal recall ability also produce deficits in imitative performance. To examine this possibility, McDonough, Mandler, McKee, and Squire (1995) tested deferred imitation by adults with amnesia associated with damage to the medial temporal lobe structures implicated in recall memory in particular and declarative, or explicit, memory more generally. 2 Relative to normal adults who readily learned and remembered the multistep sequences, patients with amnesia did poorly. Indeed, they performed no better than control participants who had never seen the sequences demonstrated. This finding strongly suggests that imitation procedures tap explicit memory, which gives rise to the capacity to recall (for further development of the argument that elicited imitation taps explicit mnemonic processes, see, for example, Bauer, 1996, 1997, 2002b; Carver & Bauer, 2001; Mandler, 1990; Meltzoff, 1990). The findings of McDonough et al. ( 1 9 9 5 ) make clear that damage to the medial temporal lobe structures implicated in amnesia impairs adults' recall memory as measured by the elicited-imitation task. A logical expectation based on this 2 Declarative, or explicit, memory involvesthe capacity for conscious recognition or recall of names, places, dates, events, and so on (e.g., Squire, 1982, 1986). It is the type of memory that most of us refer to when we talk about "memory for" or "remembering" a past event. Declarative, or explicit, memory is characterized as fast (e.g., supporting one-trial learning), fallible (e.g., memory traces degrade, retrieval failures occur), and flexible (i.e., not tied to a specific modality or context). In contrast, nondeclarative, or implicit, memory represents a variety of nonconscious abilities, including the capacity for learning habits and skills, some forms of classical conditioning, and priming (i.e., facilitated processing of a stimulus as a result of previous exposure to it). It is characterized as slow (i.e., with the exception of priming, it results from gradual or incremental learning), reliable, and inflexible (Squire, Knowlton, & Musen, 1993). For ease of exposition, throughout this chapter, we use the term explicit to refer to memory characterized as either declarative or explicit.

Developments in Early Recall

109

finding is that the developmental status of the neural circuitry also would impact performance. In other words, neurological development should place a lower limit on the capacity for recall memory as evidenced by performance on the elicitedimitation task. In the next section we provide a brief overview of the neural substrate implicated in explicit memory in general and recall memory in particular and of our emerging, yet still incomplete, understanding of the course of development of the circuitry. B. THE NEURALSUBSTRATEOF EXPLICITMEMORY In adult humans, recall memory, and in particular long-term recall, is thought to be dependent on particular neural structures and their connections. The literature suggests that the formation, maintenance, and subsequent retrieval of memories over the long term depends on a multicomponent network involving temporal (including the hippocampus, entorhinal, and perirhinal cortices) and cortical (including prefrontal cortex and limbic/temporal association areas) structures (e.g., Bachevalier & Mishkin, 1994; Murray & Mishkin, 1998). The medial temporal lobes are involved in initial encoding and consolidation of memory traces, the cortical association areas are the presumed long-term storage sites, and the prefrontal cortex is implicated in retrieval of memories from long-term stores (e.g., Bachevalier & Mishkin, 1994; Eichenbaum & Cohen, 2001; Squire, Knowlton, & Musen, 1993). One of the major functions of the hippocampus is to bind together distributed sites in the neocortex that together represent an event (e.g., Kandel & Squire, 2000; Moscovitch, 1992; Zola-Morgan & Squire, 1990). Specifically, inputs from multiple neocortical association areas are thought to converge on parahippocampal structures (e.g., entorhinal cortex) where, without further processing, they are maintained temporarily and then only as isolated elements. The processing that creates relations among the elements is carried out by the hippocampus (Eichenbaum & Cohen, 2001). The organization and consolidation of a memory trace involves neurochemical and neuroanatomical changes that occur over the course of hours to months and eventually render the memory representation independent of the hippocampus (Squire, 1992). Consistent with this suggestion, patients with damage to medial temporal structures have impairments in storing new information but not in retrieving information stored long ago (e.g., Squire, 1992; Squire & Zola-Morgan, 1991; Zola-Morgan, Squire, & Amaral, 1986). Impaired storage is particularly pronounced for episodic features of new events, such as the specific time or place of the experience (e.g., Squire, 1986; Vargha-Khadem et aL, 1997). Behavioral (e.g., Jetter, Poser, Freeman, & Markowitsch, 1986; Markowitsch, 1995) and neuroimaging (e.g., Cabeza et al., 1997; Kapur et al., 1995) evidence implicates the prefrontal cortex in retrieval of memories from long-term stores. The structure seems to play a particularly prominent role in retrieval of episodic features, such as temporal information (e.g., Janowsky, Shimamura, & Squire,

110

B a u e r et al.

1989; Lepage & Richer, 1996; Squire, 1982, 1986) and the specific identity of previously encountered elements (e.g., words lists; Dywan & Jacoby, 1990; Jacoby, 1991). Autobiographical or personal memory may be especially dependent on the prefrontal cortex (e.g., Barnett, Newman, Richardson, Thompson, & Upton, 2000). Moreover, recall of episodic information requires not only the prefrontal, association, and temporal structures themselves, but also intact connections between them (Yasuno et al., 1999). A logical implication of this analysis is that neurological development will limit the capacity for long-term recall. That is, the capacity to recall experiences after a delay will emerge only once the neural structures in the explicit memory network, as well as the connections between them, reach a minimum level of functional maturity. In primates, most of the hippocampus matures early (Serres, 2001). However, several structures are later to develop, including the dentate gyrus of the hippocampus (a critical link in the circuit that connects parahippocampal structures to the CA3 and CA1 regions of the hippocampus), the frontal cortex (implicated in retrieval from long-term stores), and temporal-cortical connections. The full network begins to coalesce in the second half of the 1st year of life (C. A. Nelson, 1995, 1997; Schacter & Moscovitch, 1984) and it continues to develop for months thereafter (Carver & Bauer, 2001; Nelson & Webb, 2002; Serres, 2001). Consistent with these suggestions, in the next section we review evidence that long-term recall memory is newly or recently emergent at 9 months of age and that it increases in prevalence and robustness over the course of the 2nd year of life.

II. Characterizing Recall M e m o r y in the First Two Years of Life A. DEVELOPMENTALCHANGESIN THE PREVALENCEAND ROBUSTNESS OF LONG-TERMRECALL

1. Age-Related Changes in the Prevalence of Long-Term Recall Some of the evidence that long-term recall memory processes are newly (or recently) emergent at 9 months, and that they become increasingly prevalent over the 2nd year of life, comes from a study by Carver and Bauer (1999). We tested 9-month-olds' recall of two-step sequences over a 1-month delay. We used multistep sequences (rather than single actions: e.g., Meltzoff, 1988a,b) because reproduction of multistep sequences in the correct temporal order provides especially compelling evidence of recall, as opposed to recognition (e.g., Bauer, 2002b). Consider that to reproduce an ordered sequence the infant or child cannot rely on recognition: Once modeling or demonstration of the sequence is complete, information about the order in which the sequence unfolded is not perceptually available. To reproduce an ordered sequence then, the infant or child must encode order information during demonstration of the event, and later retrieve

Developments in Early Recall

111

the information from a representation of the event, in the absence of ongoing perceptual support (see Bauer, 1996, 1997, 2002b, for further discussion). In this requirement, the task is analogous to verbal recall (Mandler, 1990). Because reproduction of multistep sequences in the correct temporal order provides the strongest evidence of recall, in discussing normative changes in long-term recall memory over the first 2 years of life, we focus on the extent to which infants and children adhere to the target order in their reproductions of event sequences. In the Carver and Bauer (1999) study, as a group the 9-month-olds recalled the individual actions of the events. That is, after the 1-month delay, the infants produced a larger number of the individual actions of the events to which they had been exposed, relative to new, control events. However, only 45% of the infants evidenced temporally ordered recall of the sequences. We since have replicated this pattern in two independent samples. In studies by Bauer, Wiebe, Carver, Waters, and Nelson (2002) and Bauer, Wiebe, Waters, and Bangston (2001), 46 and 43% of 9-month-olds evidenced ordered recall, respectively. Thus, at 9 months, individuals differ considerably in long-term ordered recall. Nevertheless, the behavior of 9-month-olds differs from that of 6-month-olds, only a small proportion (25%) of whom evidence ordered recall over a delay as brief as 24 h (Barr, Dowden, & Hayne, 1996). Together these findings suggest that long-term ordered recall is newly (or recently) emergent at around 9 months of age. Over the 2nd year of life, long-term recall ability becomes increasingly prevalent. In a large-scale study of the parameters of remembering and forgetting in the 2nd year of life (Bauer, Wenner, Dropik, & Wewerka, 2000), we enrolled children at the ages of 13, 16, or 20 months and tested them for recall of multistep sequences after delays of 1, 3, 6, 9, or 12 months (delay was between-subjects). Table II shows the percentages of children of each age who, at delayed testing, evidenced ordered recall (i.e., higher performance on previously experienced as compared with new control sequences). An asterisk indicates that the number of children with the pattern is greater than the number that would be expected by chance. In contrast to 9-month-olds, only roughly half of whom evidence ordered recall after 1 month (Bauer et al., 2001; Bauer, Wiebe, et al., 2002; Carver & Bauer, 1999), after 1 month, high, and roughly comparable, percentages of 13-, 16-, and 20-month-olds evidenced ordered recall. As the delay interval increased, fewer children maintained information about the order of the events; the younger the children were at the time of experience of the events, the steeper was the forgetting function. These data thus indicate age-related increases in the prevalence of longterm ordered recall. In summary, 9 months seems to represent a developmental point at which a newly (or recently) emergent behavior is seen in a subset of the population. Only 4 months later, at 13 months, the behavior is readily observed. Over the course of the 2nd year, the temporal extension of the ability increases, such that by 20 months, almost 70% of children show temporally ordered recall after as many as 12 months.

112

B a u e r et al.

T A B L E II Percentages of Children S h o w i n g Evidence of Ordered Recall Over Retention Intervals of 1 to 12 Months a Delay interval (in months) A g e at experience

1

3

6

9

12

13 months

78*

67

39

44

39

16 months

94*

94*

72*

50

61

20 months

100"

100"

83*

78*

67*

a Results from Bauer, Wenner, Dropik, and Wewerka (2000). Note: A total of 360 children participated, 180 of whom were 16-month-olds. All of the 13-month-olds and half of the 16-month-olds were tested on sequences three steps in length; all of the 20-month-olds and half of the 16-month-olds were tested on sequences four steps in length. Differences in sequence length accommodated age-related changes in the lengths of sequences that children can accurately imitate (see Bauer, 1995, 1996, 1997, for reviews). At each of three sessions, spaced 1 week apart, the children were exposed to the same six event sequences. Three of the sequences they never were permitted to imitate; three of the sequences they were permitted to imitate one time, at the end of the third exposure session. The children returned for delayed-recall testing after delay intervals of either 1, 3, 6, 9, or 12 months. At the delayed-recall session, the children were tested for recall of the six sequences to which they previously had been exposed as well as on three new sequences, as a within-subjects control. Since performance did not differ as a function of whether imitation was permitted, the data are collapsed across this manipulation. The specific values are for 16-month-olds tested on four-step sequences; the pattern applies to both groups of 16-month-olds (see Bauer et al., 2000, for details). Asterisks indicate that the number of children with higher levels of performance on previously experienced rather than on new, control sequences was reliably greater than chance. Because determination of chance levels is affected both by the number of observations and by the number of tied observations, identical values will not necessarily yield identical outcomes (e.g., 13-month-old 3-month delay and 20-month-old 12-month delay).

2. Age-Related Changes in the Robustness of Long-Term Recall In addition to evidence of age-related increases in the prevalence of long-term recall across the first 2 years of life, our research has revealed changes in its robustness. At 9 months of age, when the capacity is newly or recently emergent, long-term recall is fragile, apparently depending on multiple experiences of events. If infants receive fewer than three exposures to events prior to imposition of the delay, a maximum of only 21% demonstrate ordered recall over 1 month (Bauer et al., 2001). In contrast, by 13 months, only a single experience is necessary for ordered recall (Bauer & Hertsgaard, 1993). By 13 months of age, children no longer require multiple experiences to maintain temporally ordered event representations over a delay. Nevertheless, the robustness of long-term recall memory changes over the 2nd year of life. In the Bauer et al. (2000) study, across the delay intervals of 1, 3, 6, 9, and 12 months, older children recalled more than younger children. Age effects were especially apparent in recall of temporal order and under conditions of greater cognitive demand. Specifically, in regression analyses, age contributed more unique variance in children's ordered

Developments in Early Recall

113

recall than in recall of the individual actions of the sequences. Age contributed more variance at the longer retention intervals of 6, 9, and 12 months, relative to the shorter retention intervals of 1 and 3 months. In addition, age effects were especially apparent when children's recall was supported by the event-related props alone. Age differences diminished when verbal reminders were provided (i.e., after a period during which recall was prompted by event-related props alone, the experimenter provided a verbal reminder in the form of the label used to introduce the sequence at each exposure session). 3. Summary Together, changes in the prevalence and robustness of long-term recall late in the 1st year and throughout the 2nd year of life suggest consolidation of mnemonic function in this time frame. Whereas at 9 months individual differences in recall over 1 month are the rule, by early in the 2nd year they are the exception. Over the 2nd year of life, the capacity to recall over yet longer delays consolidates such that by 20 months, a large proportion of children demonstrate ordered recall after as many as 12 months have passed.

B. EXPERIMENTALMANIPULATIONSAFFECTINGINFANTS'AND YOUNGCHILDREN'SRECALL Although there are regular, normative changes in long-term recall memory in the first 2 years of life, a child's age is not the sole determinant of whether or for how long she or he will remember. Moreover, within an age group, there are individual differences in how much children remember. In this section we consider some of the experimentally controlled determinants of long-term recall memory. In section III we discuss individual differences in performance. A number of factors have been experimentally manipulated and found to influence children's recall over both the short and the long term. Because the influence of these factors has been reviewed in detail elsewhere (e.g., Bauer, 2002a,b; Bauer et al., 2000), we summarize them briefly here, highlighting: (a) the nature of temporal connections in events, (b) the number and timing of experiences of events, (c) active participation in events, and (d) verbal reminders of to-be-remembered events. First, children's ordered recall of event sequences is facilitated by enabling relations. Enabling relations are said to exist when, to achieve a particular endstate or goal, the actions in the sequence must occur in a particular temporal order. For example, to Make Big Bird turn on the light as described in Table I, the car must be in the base before the plunger is pushed. If the steps are performed in alternate order (i.e., push in the plunger and then put the car in the base), then the light bulb would not illuminate. Children's ordered recall of sequences constrained by enabling relations is better than their ordered recall of sequences lacking enabling

114

B a u e r et al.

relations (which are ordered arbitrarily). The effect is apparent both at immediate testing (e.g., Bauer, 1992; Bauer & Thai, 1990) and after short delays (e.g., Barr & Hayne, 1996; Bauer & Hertsgaard, 1993; Mandler & McDonough, 1995). As delay length stretches from weeks to months, although children still remember the events, the facilitating effects of enabling relations diminish (Bauer et al., 2000). Whereas even the youngest children tested accurately recall sequences constrained by enabling relations (e.g., Bauer & Mandler, 1992; Carver & Bauer, 1999; Mandler & McDonough, 1995), not until the second half of the 2nd year of life do children reliably reproduce arbitrarily ordered sequences in the correct temporal order (Bauer, Hertsgaard, Dropik, & Daly, 1998; Wenner & Bauer, 1999). (See Bauer, 1992, 1995, and Bauer & Travis, 1993, for discussions of the means by which enabling relations in events may influence ordered recall.) Second, the number and timing of experiences of events also influence children's recall. Repeated experience aids memory both in terms of the amount of information remembered, and in terms of the length of time over which events are recalled (e.g., Bauer, Hertsgaard, & Wewerka, 1995; Bauer et al., 2001; Fivush & Hamond, 1989). For children as young as 13 months, repeated experience is not necessary for recall over the short term (Bauer & Hertsgaard, 1993; Bauer et al., 1995). Nevertheless, particularly over the longer term, repeated experience facilitates recall. For example, in the Bauer et al. (1995) study, for events experienced only once, performance after a 1-month delay was substantially lower than performance after a 1-week delay. In contrast, events experienced three times before imposition of a 1-month delay were well recalled. Notably, recall after 1 month of events experienced three times was comparable to that after 1 week of events experienced only once (see Fivush & Hamond, 1989, for similar effects with 24and 29-month-olds). The timing of exposures to events also affects the efficacy of repeated experience. Hudson and Sheffield (1998) found that the effects of reexperiencing events after 8 weeks were more pronounced than reexperiencing the events after 15 min or after 2 weeks. Events need not be reexperienced in their entirety for recall to be facilitated. Simply showing children a subset of events (i.e., three of an original six experienced) facilitates memory for the entire set, presumably through spreading activation (Sheffield & Hudson, 1994). Third, active participation in events is not necessary for later recall (e.g., Barr & Hayne, 1996; Bauer et al., 2000; Meltzoff, 1995). Nevertheless, active participation in the form of imitating to-be-remembered events is associated with better recall, at least over short delays (e.g., Bauer et al., 1995). As is the case for effects of the temporal structure of event sequences, over longer delays, effects of active participation dissipate. For example, in the Meltzoff (1995) study, recall after 2 and 4 months was comparable for 14- and 16-month-old children who were and were not allowed to imitate prior to imposition of the delays. Similarly, in the Bauer et al. (2000) study, regardless of the length of the delay (i.e., 1 to 12 months), recall was comparable for events that 13- to 20-month-old children had and had not been allowed to

Developments in Early Recall

115

imitate prior to imposition of the delays. Thus, active participation in the form of imitation enhances recall over the short term but not necessarily over the long term. Fourth, in children preschool age and older, cues or reminders of previously experienced events facilitate memory retrieval after a delay (e.g., Hudson & Fivush, 1991). Verbal reminding also aids memory retrieval over delay intervals of many months in children as young as 13 months at the time of experience of novel events (Bauer et al., 2000; see also Bauer et al., 1995). Indeed, as noted earlier, verbal reminding has the effect of reducing age-related differences in the amount of information that young children recall after a delay (Bauer et aL, 2000). Critically, the effects of verbal reminding cannot be attributed to "suggestion" of plausible event sequences: Children do not generate the target actions and sequences of events for which they have been provided a verbal label but to which they have not been exposed (Bauer et al., 1995, 2000). That memories can be triggered by verbal reminders is particularly important to recall after long periods of time: Over significant delays, regardless of age, little that is not reminded is retrieved (e.g., Hudson & Fivush, 1991).

III. Individual Differences in Long-Term Recall: Children's Gender, Children's Language Proficiency, and Variability in Initial Learning Knowledge of the factors that affect long-term recall permits prediction of group trends in mnemonic performance. For example, we expect high levels of recall when children are tested under "optimal" conditions such as when to-beremembered events are constrained by enabling relations; when children have well-timed, multiple experiences of events; when children are permitted active participation at encoding; and when verbal reminders are provided at the time of retrieval. Conversely, we expect less accurate recall of sequences lacking enabling relations, of which children receive only a single observational experience, and of which they are given limited reminders. Whereas knowledge of what factors affect long-term recall permits substantial leverage in predicting group trends, it constitutes only one category in the catalog of the determinants of infants' and young children's recall memory. An equally important category of explanation is of individual differences in performance. Early in development, when long-term recall ability is newly emergent, individual differences are apparent in whether or not children show evidence of memory. That is, roughly 50% of 9-month-olds demonstrate ordered recall whereas roughly 50% do not (e.g., Bauer et al., 2001). As recall ability stabilizes and becomes more prevalent, individual differences take the form of variability in the a m o u n t remembered. For example, in the Bauer et al. (2000) study, under "optimal" conditions (i.e., verbally cued memory for enabling sequences that children had experienced three times and imitated one time prior to imposition of a 1-month delay),

116

Bauer et al.

even among children in whom long-term recall ability is prevalent and robust (20-month-olds), some children recalled all four possible target actions whereas others recalled as few as one. Similarly, some children evidenced perfect temporally ordered recall whereas others revealed no evidence of ordered recall. What accounts for such wide variability in performance? Given that we cannot explain it by appeal to experimentally controlled factors (i.e., all of the children in this cell of the design were subjected to the same encoding and test conditions), we must look to characteristics that vary across individuals. Among preschool age and older children, individual differences in memory are associated with factors such as strategy use, metamemory, content knowledge, speed of processing, and overall intelligence (see Schneider & Bjorklund, 1998, for a review). However, research on individual differences in early recall has not concerned the factors familiar in research with older children. In some cases, the reason is obvious. Very young children are not very strategic in their memory performance (although see DeLoache, Cassidy, & Brown, 1985) nor are they skilled at describing their own or others' cognitive processes. For these reasons, examinations of variability in strategy use and metamemory as sources of variability in very young children's recall are unlikely to be productive. Differential knowledge of content may well influence young children's recall in the same manner as it influences older children's. This possible source of individual differences has not been pursued, however, primarily because we do not have adequate tools for evaluating semantic memory content or organization in pre- and early verbal children (although see Bauer, 1993, for gender-differential recall of masculineand feminine-typed event sequences by 25-month-old boys, possibly as a function of greater knowledge of same-sex behaviors). Finally, although factors such as speed of processing and overall intelligence have been addressed in the literature on recognition memory in infants (e.g., Fagan, 1984; Rose, Feldman, & Wallace, 1992), they have not been examined as possible sources of variability in young children's recall memory. In the literature on early recall memory, two possible sources of differences in performance have been studied: one is a potential source of group differences, namely, children's gender, and the other is a potential source of individual differences, namely, children's language competence. We review the literatures concerning each of these sources in turn. We then discuss a third possible source of influence, namely, individual variability in initial learning of events. A. EFFECTSOF CHILDREN'SGENDER In the case of gender, the expectation of possible systematic variability in longterm recall memory performance stems from suggestions of faster maturation of girls relative to boys throughout infancy (e.g., Hutt, 1978; although see Reinisch, Rosenblum, Rubin, & Schulsinger, 1991). Although differential maturity provides a reasonable basis for expecting gender differences, few relations have been found.

Developments in Early Recall

117

For example, in 16 of 34 experiments published from our laboratory since 1987, tests for possible effects of gender were conducted. In 12 of the 16 cases, levels of performance by girls and boys did not differ. In four experiments, effects of gender were detected. Three of the four cases involved 9-month-olds, an age at which long-term recall ability is recently or newly emergent. In two cases, the effect of gender favored boys (Bauer et al., 2001; Experiments 1 and 2); in one case, the effect favored girls (Carver & Bauer, 1999). In the final experiment (Bauer, Hertsgaard, et al., 1998), 28-month-old children were challenged to reproduce lengthy, arbitrarily ordered event sequences (a late-developing skill; e.g.,Wenner & Bauer, 1999). In this case, the advantage was for girls relative to boys. Others using imitation procedures with children in this age range have had similar actuarial experience. A sample of 31 published experiments from the laboratories of Harlene Hayne and Andrew Meltzoff, both of whom employ imitation paradigms with children aged 2 years and younger, yielded six cases in which tests for possible gender effects were conducted. All yielded null effects. The remaining 25 experiments did not report tests for possible effects of children's gender. One reason that gender has not been examined more systematically is that studies of early recall memory typically include small samples of children. Across the 65 experiments examined for treatment of possible effects of gender (i.e., 34 from our laboratory and 31 from the laboratories of Hayne and Meltzoff), sample sizes ranged from 8 to 32 participants per cell of the research designs; the modal sample size was 12 participants. To divide such small samples in half to examine gender differences is to invite nulluor worse, spuriousmfindings. However, with larger samples, possible gender-related differences can be examined, perhaps profitably. Fortunately, the Bauer et al. (2000) study provides an opportunity to address the issue. The sample included 360 children, roughly half girls and half boys. In the Bauer et al. (2000) study, 185 girls and 175 boys were tested for longterm recall after delays of 1, 3, 6, 9, or 12 months. Examination of gender-related patterns of performance in the large sample thus affords a relatively definitive test of gender as a potential source of variability in young children's recall memory performance. We conducted separate analyses for the 20- and 16-month-olds, tested on four-step event sequences, and for the 13- and 16-month-olds, tested on three-step event sequences. As noted earlier (see section II.A.2), children's long-term recall was tested under two conditions. First, for each sequence, the children were provided with the props for the event. After a child-controlled period of manipulation of the props, the experimenter provided a verbal reminder in the form of the title given the sequence at the time of its introduction (e.g., "You can use this stuff to make a rattle"). When the children were prompted only by the event-related props, some gender differences emerged. For the children tested on four-step sequences, in the 9-month delay condition, girls produced fewer of the individual target actions of the events, relative to the boys (M = 1.15 and 1.73, respectively). Lower levels of production of the individual target actions of the events translated into lower levels

118

Bauer et al.

of ordered recall by the girls, relative to the boys (M = 0.25 and 0.50, respectively). Among the children tested on four-step event sequences there were no reliable gender effects in the 1-, 3-, 6-, or 12-month delay conditions. Moreover, when the children were supported both by the event-related props and by verbal reminders of the events, girls and boys did not differ, even in the 9-month delay condition. For the children tested on three-step sequences, prior to provision of the verbal reminders, girls in the 6-month delay condition performed at higher levels, relative to boys in the same condition (M = 1.41 and 1.07, respectively). However, the effect was only apparent on the events that the children were permitted to imitate. On events that they had only watched, girls and boys produced comparable numbers of individual target actions (M-- 1.11 and 1.02, respectively). The only other gender effect was in production of ordered pairs of actions. Although the performance of the 13-month-olds was not affected by gender, among the 16-month-olds, girls produced fewer correctly ordered pairs of actions, relative to boys (M = 0.62 and 0.86, respectively). Again, as was the case for the children tested on four-step event sequences, when verbal reminders were provided, no gender-related differences in performance were observed. In summary, although one cannot prove the null hypothesis, the relative paucity of reports of gender differences in the literature on early recall memory, coupled with the lack of meaningful effects in a large-scale study, implies that gender is not a major source of variability in early recall memory performance. There are, however, two exceptions to this general statement. First, gender effects seem to be apparent at times of "transition." They were apparent at the "dawning" of the ability to engage in long-term temporally ordered recall (Bauer et al., 2001; Carver & Bauer, 1999) and at the dawning of the ability to accurately reproduce lengthy, arbitrarily ordered sequences (Bauer, Hertsgaard, et al., 1998). Second, in a large-scale study, gender effects were apparent under conditions of high cognitive demand (i.e., recall after a long delay) and lesser contextual support (i.e., in the absence of verbal reminders; Bauer et al., 2000). B. EFFECTSOF CHILDREN'SPRODUCTIVEAND RECEPTIVELANGUAGE Despite the fact that elicited- and deferred-imitation tasks are nonverbal, children's own language facility might influence performance in several ways. First, in many studies (including all published studies from our laboratory), demonstration of to-be-remembered events is accompanied by verbal narration. Children's abilities to comprehend the language spoken to them is thus one potential source of variability. Second, even when narration is not provided (e.g., Meltzoff, 1985, 1988a, 1988b, 1995), language development is a source of variability because children with greater language skills potentially could verbally encode the events that they observe, thus providing themselves with additional retrieval cues. Indeed, the

Developments in Early Recall

119

ability to "augment" nonverbal representations with language plays a role in the later verbal accessibility of memories likely encoded without the benefit of language (Bauer, Kroupina, et al., 1998; Bauer, Wenner, et al., 2002; Bauer & Wewerka, 1995, 1997). Nevertheless, systematic variability attributable to differences in children's language proficiency has been examined only in very specific, limited contexts. For example, in the Bauer, Hertsgaard, et al. (1998) study, we found that total productive vocabulary and early grammatical development were correlated with 28-month-old children's performance on lengthy (five-step) arbitrarily ordered sequences. The language variables were not correlated with performance on lengthy sequences constrained by enabling relations nor with performance on shorter (three-step) sequences. To our knowledge, language development has not been examined for potentially more general relations with memory performance. The Bauer et al. (2000) study affords the opportunity to examine the possibility that variability in language development accounts for systematic variance in longterm recall. Specifically, to match the children assigned to the different delay conditions (1-12 months), we used children's scores on the MacArthur Communicative Development Inventory for Toddlers (20-month-olds) and the MacArthur Communicative Development Inventory for Infants (13- and 16-month-olds; Fenson et al., 1994). Only vocabulary production data were available for the 20-month-olds, but data on both productive and receptive vocabulary were available for the 13- and 16-month-olds. We had completed MacArthur inventories for 93% of the children (336 of the 360 children in the sample). Children's reported productive vocabulary scores ranged from 0 to 651 words (across the 13- to 20-month age range); their reported receptive vocabulary scores ranged from 11 to 393 words (across the 13- to 16-month age range). The sample thus featured ample power and ample variability to permit detection of systematic relations between children's language and children's recall memory performance. To examine relations between vocabulary and recall across delay conditions (and thus take advantage of the large samples available), we converted children's mnemonic performance into z-scores. We then calculated correlations between children's recall and their reported vocabularies. For reported productive vocabulary, the correlations between language and memory performance ranged from -.02 to .15. None of the correlations was statistically reliable. For reported receptive vocabulary, the correlations ranged from -.09 to .23. The correlation of .23 between reported receptive vocabulary and delayed-recall of individual actions by 16-month-olds tested on four-step sequences approached significance (p < .07). None of the other correlations approached significance (p > .40). Thus, as measured in this large sample of children, neither reported productive nor receptive vocabulary was reliably related to recall memory performance. In summary, despite conceptual reasons for expecting relations between children's language facility and recall memory performance, virtually no relations

120

Bauer et al.

obtained, even in a large sample with ample variability. The picture changes only slightly as children get older. For example, Gordon et al. ( 1 9 9 3 ) found that the language abilities of 5-year-olds were related to the amount of detail they reported for physical examinations at the doctor's office. However, in a comparable study, including both 3- and 5-year-old children, Greenhoot, Ornstein, Gordon, and Baker-Ward (1999) found no relations between language and either verbal report or physical enactment of the examination. Thus, although language ability might logically be related to recall memory in young children, compelling relations have not been observed. c. EFFECTSOF INDIVIDUALVARIABILITYIN INITIALLEARNING Individual differences in initial learning or encoding of to-be-remembered events are another possible source of variability in long-term recall. Retrieval of information about a previously experienced event depends on what was originally encoded about the event. Indeed, this fact presents a challenge to valid research on developmental differences in the length of time over which memory persists (Howe, 2000): Because older children learn at a faster rate than younger children, older children likely have more detailed and elaborated memory representations, relative to younger children. In the present context, we focus attention not on age-related differences in initial learning, but on individual differences therein as a possible source of individual differences in recall. Factors as fundamental as the speed of processing (e.g., Kail & Salthouse, 1994) will affect how much event-related information is encoded, as well as how effectively and efficiently information is consolidated for long-term storage and subsequent retrieval (see Bauer, Cheatham, Cary, & Van Abbema, 2002; Bauer, Van Abbema, & de Haan, 1999, for discussion). Until more is known about the processes of consolidation of to-be-remembered information for long-term storage, and possible individual differences therein, a potentially large source of variance in early long-term recall will remain unexplored. Moreover, identification of individual differences in initial learning as a source of variance in long-term recall only begs the question of the source of individual differences in initial learning. D. SUMMARYOF THREE SOURCESOF VARIABILITYIN EARLY RECALL MEMORY The study of variability in early recall memory does not have a long history. Two of the "first guess" possibilities for systematic variability in performance, namely, gender and language, have been rather disappointing as potential explanatory sources. Individual studies rarely include tests of possible effects associated with these factors. When possible effects are tested, they are not often found. Even in a large-scale study with sufficient power to detect variance associated with children's gender and vocabulary, effects were few. In the case of children's gender,

Developments in Early Recall

121

although some isolated effects were detected, they disappeared under conditions of greater contextual support. In the case of children's language, no systematic relations were observed. Whereas a third potential source of individual differences has been identified, namely, variability in initial learning of events, its effects rarely have been acknowledged explicitly. Moreover, detection of significant variance associated with differences in initial learning would only change the focus of the question from the source (or sources) of individual differences in long-term recall to the source (or sources) of individual differences in initial mastery. Regardless of the specific focus of the question, significant variance remains to be explained. In the Bauer et al. (2000) study, we were able to quantify the amount. We investigated the variance accounted for by a combination of nine factors, including the child-related variables of age at exposure to the test events (13, 16, or 20 months), gender, and productive vocabulary (ranging from 0 to 651 words); features of the research design, including the number of steps in the tobe-remembered event sequences (three or four steps), the experimenter by whom the child was tested (two different experimenters conducted the sessions; a given child was tested by the same experimenter at each session), and the length of the delay between exposure and test (1, 3, 6, 9, or 12 months); and measures of children's performance in the imitation context, including baseline levels of performance, levels of initial learning of the events (as measured by immediate imitation of half of the sequences), and levels of performance on the new control events. Together, the nine variables accounted for 32% of the variance in children's recall of the individual target actions of the sequences; they accounted for 26% of the variance in children's recall of temporal order. Clearly, substantial variance remains to be explained. Accordingly, we look beyond children's gender, language, and initial learning to other characteristics of the developing child that in research with both younger infants and older children have been shown to influence memory performance, namely, temperament characteristics.

IV. Individual Differences in Long-Term Recall: Children's Temperament Characteristics A. WHATIS TEMPERAMENT? Temperament or behavioral style refers to constitutionally based patterns of responding to environmental stimuli (e.g., Gunnar, 1990; Rothbart, Ahadi, & Hershey, 1994; Rothbart & Bates, 1998). Many theories and operationalizations of the spectrum of behavioral patterns exist, yet most stem from the nine dimensions originally outlined by Thomas and Chess (1977). Moreover, three psychobiological dimensions underlie each theory or scale of temperament: positive affectivity or approach (e.g., response to an attractive toy), negative affectivity and inhibition (e.g., response to fear-provoking stimuli), and attention (i.e., orienting, regulation

122

B a u e r et al.

and control; for more detailed discussions, see Goldsmith, 1996; Gunnar & Nelson, 1994; and Rothbart, Derryberry, & Posner, 1994). B. WHY MIGHT TEMPERAMENT CHARACTERISTICS BE RELATED TO EARLY RECALL MEMORY?

One major motivation for examination of possible relations between early recall memory and temperament characteristics stems from behavioral observations. For example, encoding of a stimulus should be facilitated by attention to it. This fact is implicitly recognized in studies of infant memory. In research using visual recognition paradigms, researchers use "flashy" orienting stimuli (e.g., blinking lights) to ensure adequate attention prior to the display of to-be-encoded stimuli; as trials progress, testers tap on the screen in attempts to reengage infants' attention. In the context of research using elicited and deferred imitation, researchers demonstrate individual actions and sequences two or more times in succession to ensure that infants and young children, whose attention may "wander" during demonstration, be given adequate encoding opportunities. Given the significance of encoding to subsequent recall, young children's performance might well differ as a function of temperament characteristics such as regulation and control of attention. A second major motivation for examining relations between temperament and memory is that in the case of the temperament aspect of attention, in particular, they rely on a shared neural substrate (see C. A. Nelson & Dukette, 1998, for discussion). In both cases, the systems seem to consolidate and stabilize over the course of the latter part of the 1st year and throughout the 2nd year of life. The neural substrate implicated in recall memory was reviewed earlier (see section I.B). Behavioral expression of positive affectivity, negative affectivity, and attention is associated with activity in the limbic system, specifically the amygdala and hippocampus, and the parietal and frontal regions (Derryberry & Reed, 1996; Derryberry & Rothbart, 1997; Rothbart & Posner, 2001). This dimension of temperament becomes reliably functional by 6 months, when infants begin to smile, laugh, and approach rewarding stimuli (Derryberry & Rothbart, 1997; Rothbart & Bates, 1998). By 9 months, infants begin to evidence fearful behaviors linked with negative affectivity and inhibition. Such behaviors are supported by hippocampal circuits that respond to biologically prepared fear stimuli, punishment, and nonreward signals. Patterns of negative affectivity and inhibition stabilize over the course of the 2nd year, presumably in concert with developments in the associated neural circuitry (Derryberry & Reed, 1996; Derryberry & Rothbart, 1997). The third psychobiological dimension of temperament--attention--has been described as relying on two relatively independent neural systems (Posner, Petersen, Fox, & Raichle, 1988), the more anterior of which overlaps substantially with the network thought to subserve recall memory (C. A. Nelson & Dukette, 1998). The first system, the posterior attention network, is implicated in orienting, which

Developments in Early Recall

123

involves the disengagement and shifting of attention from one stimulus to another. The neural structures involved in this network include the pulvinar, parietal cortex, and superior colliculus (Posner et al., 1988). Consistent with evidence of relatively early metabolic activity in these structures (Chugani, 1994), posterior attention network function is evident by 2 to 4 months of age. The second system, the anterior attention network, is implicated in sustained attention. The anterior network involves interconnections between the limbic system (especially the hippocampus) and the frontal cortex. Presumably as a function of the later development of some of the structures involved in this network (e.g., dentate gyrus of the hippocampus, prefrontal cortex), evidence of its function is not apparent until late in the 1st year and early in the 2nd year of life (e.g., Vecera, Rothbart, & Posner, 1991). Moreover, the ability to sustain and control attention changes considerably throughout the course of the 2nd year and beyond (Derryberry & Rothbart, 1997; Rothbart & Posner, 2001). c. EVIDENCEOF RELATIONSBETWEENTEMPERAMENT CHARACTERISTICSAND MEMORYIN INFANCYAND THE PRESCHOOLYEARS Although the logic of behavioral links between memory and aspects of temperament is compelling, there are few empirical tests of possible relations. Wachs, Morrow, and Slabach (1990) examined relations between parent-reported temperament (assessed via the Revised Infant Temperament Questionnaire: Carey & McDevitt, 1978), home environment (assessed via the Purdue Home Stimulation Inventory), and recognition memory (assessed via novelty preference) in 3-month-olds. Initial analyses revealed relations between home environment and memory performance. However, the relations seemingly were mediated by infants' temperament characteristics: When temperament scores were partialled out of the correlations, half of the observed relations between home environment and recognition memory fell below significance. Relations between infant temperament and visual recognition memory (measured via event-related potential, ERPs) also were reported by Gunnar and Nelson (1994). They found that 12-month-old infants whose parents reported them to express more positive affect showed more ERP activity during the visual recognition task. This finding suggests that the biological components of memory and positive affect are associated. Aspects of temperament also have been found to relate to verbal mnemonic expression by 3- and 5-year-old children. Specifically, Gordon et al. (1993) found that the parent-reported temperament characteristic of approach (assessed via the Temperament Assessment Battery for Children, TABC; Martin, 1988) predicted 3-year-olds' correct responses to open-ended questions, whereas greater negative emotionality predicted overall correct recall for 5-year-olds. In a separate study, Greenhoot et al. (1999) examined the impact of parent-reported temperament

124

Bauer et al.

(assessed via the TABC) on memory suggestibility in 3-year-olds. Low persistence was associated with adoption of a "yes" bias during the suggestibility portion of the study, and high manageability was associated with more memory errors. Finally, Geddie, Fradin, and Beer (2000) found that for 4- and 7-year-olds, adaptability was associated with better recall as well as with correct denial of misleading information. Two features of the existing literature on relations between memory and temperament are noteworthy. One is its very sparseness. The second is that there do not appear to be any examinations of relations between memory and temperament in children between 12 months and 3 years. This is an unfortunate omission, given the magnitude of changes that occur during this time period. Despite the absence of research, we can generate some predictions about specific relations in this space of time. For example, if memory is assessed in a novel laboratory context, children who characteristically experience a greater degree of fear or inhibition might be less likely to engage in the task and thus not perform as well. Similarly, children who are highly active might find it difficult to remain on task and as a result also not perform as well. In contrast, children who are high in positive affect and approach, or children who are able to regulate and sustain their attention, or both, might more readily engage in the task, and, as a result, evidence better memory. Given that the temperament dimensions of positive and negative affectivity stabilize before the dimension of attention, we might further predict that emotionality would be related to memory performance at a younger age (such as that observed in Gunnar & Nelson, 1994), whereas attention would be predictive only later in development. D. RELATIONSBETWEENTEMPERAMENTAND MEMORYIN THE SECOND YEAR: AGE-RELATEDDIFFERENCES The Bauer e t al. (2000) study provides an opportunity to address possible relations between temperament and memory in children 13-20 months of age at the time of experience of the to-be-remembered events. Although the study was not designed to examine relations between temperament and memory, in the spirit of a pilot study, we obtained reports of temperament from some of the children's parents. We used the Toddler Behavior Assessment Questionnaire (TBAQ; Goldsmith, 1996), a parent-report instrument consisting of 111 items referring to a variety of scenarios describing situations in which young children might be engaged. An example question is, "When your child wanted to eat something sweet before dinner was finished but did not get it, how often did s/he protest by crying loudly?" Parents were asked to indicate how often their children had such a reaction within the last month (scales range from 1, "never," to 7, "always," with the opportunity to indicate that the situation does not apply). The 111 items on the questionnaire cluster into five independent subscales: Activity Level, Anger Proneness, Interest and Persistence, Pleasure, and Social Fearfulness.

Developments in Early Recall

125

Completed instruments are available for twenty-eight 20-month-olds and twenty-five 13-month-olds. We calculated correlations between children's observed long-term recall and their scores on each of the five subscales of the TBAQ (Kleinknecht, Rademacher, & Bauer, 2002). No significant correlations were observed between recall and scores on the Anger Proneness and Social Fearfulness subscales. However, as shown in Table III, relations between parent-reported temperament characteristics on the other three subscales and recall performance were observed; different patterns of relations obtained for the 13- and 20-month-olds. Among the 13-month-olds, children's long-term recall was positively correlated with the Pleasure subscale of the TBAQ. The relation between the Pleasure subscale and long-term ordered recall was statistically significant, whereas that with production of the individual actions of the sequences was a trend (p < . 10). In other words, higher levels of recall, and in particular ordered recall, were observed for children whose parents indicated that, for example, (a) when meeting other children in the park or playground, their children willingly joined the other children; (b) when seeing a familiar adult, their children expressed joy and babbled and talked; and (c) when making a new discovery, their children smiled and seemed pleased. Importantly, children's scores on the Pleasure subscale were not related to their performance on the new control sequences. Thus the positive relations between Pleasure and performance after the delay are indicative of an effect on recall, as opposed to a more general influence on performance on the imitation task.

TABLE III Relations Between Parent-Reported Temperament Characteristics and 13- and 20-Month-Old Children's Delayed Recall TBAQ Subscale

Age group/event type 13-month-olds (df= 23) Old sequences New sequences 20-month-olds (df= 26) Old sequences New sequences

Dependent measure

Pleasure

Activity level

Recall of actions Ordered recall Recall of actions Ordered recall

.34 .45* -.12 - . 11

.35 .25 .29 .19

Recall of actions Ordered recall Recall of actions Ordered recall

-.12 .07 .02 .08

-.42* -.34 -.33 -.08

Interest and persistence

- . 12 .05 .05 -.06 .48** .46"* -.07 .10

Note: *, p < .05; **, p < .01; df, degrees of freedom. Correlations are based on children's performances after the verbal reminder. Source: Kleinknecht, Rademacher, and Bauer (2002).

126

B a u e r et al.

Among the 20-month-olds, long-term recall was negatively correlated with the Activity Level subscale and positively correlated with the Interest and Persistence subscale of the TBAQ. In the case of Activity Level, the correlation was statistically significant on the variable of production of individual actions and a trend for production of pairs of actions in the target order (p < . 10). In the case of Interest and Persistence, the correlations were reliable for both dependent measures. Specifically, lower levels of recall were observed for children whose parents said that, for example, their children (a) run through the house; (b) climb on furniture; and (c) like to play games that involve running, banging, and dumping out toys. In contrast, more robust recall was observed among children whose parents said that, for example, (a) when playing alone, their children remain interested in toys for extended periods; (b) when looking at picture books alone, their children stay interested and do not get bored quickly; and (c) when playing with a detailed or complicated toy, their children explore the toy thoroughly. In the case of relations with the Activity Level subscale, similar patterns were observed on both old and new control sequences. In contrast, in the case of the Interest and Persistence subscale, the relations were specific to performance on the old sequences. Thus, children whose parents reported them to have characteristically high levels of activity did less well on the imitation task more generally, whereas children whose parents reported them to have characteristically high levels of interest and persistence showed higher levels of recall memory. That the patterns of relations were different at 13 and 20 months is noteworthy. At the younger age, high levels of recall, and in particular ordered recall, were observed among infants whose parents reported them to find pleasure in new experiences. Perhaps these infants "embrace" the novelty of the situation and the events, and their resulting focus of attention on the model afforded them better recall. A similar pattern was observed in the Bauer, Wiebe, et al. (2002) study, in which infants 9 months of age were tested for recall after a 1-month delay. As noted earlier (see section II.A.1), approximately half of the infants demonstrated ordered recall over the delay whereas half did not. To study possible correlates of this pattern of performance, we examined the infant version of the TBAQ, the Infant Behavior Questionnaire (IBQ). The group of infants who demonstrated ordered recall after 1 month differed from the group that did not on only one of the subscales, namely, Smiling and Laughter. The Smiling and Laughter subscale is the infant analog of the toddler Pleasure subscale. The consistent pattern of relations at 9 and 13 months indicates that for children who are just "breaking into" the recall memory system, being a "happy camper" is a benefit. Whereas pleasure in novel situations facilitates memory for young infants, by 20 months deriving pleasure from new experiences apparently no longer is sufficient to ensure high levels of recall. Instead, by 20 months, in order to remember, children need to be able to focus and regulate their attention to the specific properties of the materials and their manner of combination. Characteristically high

Developments in Early Recall

127

levels of activity are not conducive to this "contemplative" attitude and thus were observed to be negatively related to successful mnemonic (and problem solving, as in the case of new events) performance (see Kleinknecht et al., 2002, for discussion). This pattern--positive relations with the Pleasure subscale at 13 months but positive relations with the Interest and Persistence subscale at 20 months--suggests that with development, Pleasure "merges into" Interest and Persistence. In other words, the pattern leads to the expectation that infants who score high on Pleasure at 13 months will at 20 months have high scores on Interest and Persistence. Because the Kleinknecht et al. (2002) study is based on cross-sectional data, we cannot use it to address this possibility. However, preliminary analyses of data from a longitudinal sample on whom we have temperament data at 13 and 24 months are consistent with this suggestion. Specifically, in this sample, high Pleasure scores at 13 months are correlated with high Interest and Persistence scores at 24 months (r = .44, p < .002). In contrast, scores on the Interest and Persistence subscale at 13 months are not significantly correlated with scores on the Pleasure subscale at 24 months (r = .17, ns). It is perhaps more than coincidence that over the period during which long-term recall ability seems to consolidate and become reliable, we see that the ability to sustain interest and focus attention becomes important for acquiring and remembering new material. As reviewed earlier (see section IV.B), the anterior attention network implicated in controlled attention (Rothbart & Posner, 2001) overlaps the temporal-cortical network implicated in long-term recall (see, for example, Carver & Bauer, 2001, for a review). In both cases, the functions are dependent on feedforward and feed-backward connections between the frontal lobes and other brain regions such as the hippocampus. Indeed, developments in the control of attention probably have direct implications for developments in mnemonic function, and vice versa. In summary, early in the development of typical infants and children, being able to engage with new objects and people and finding joy in the process is associated with better memory. By 20 months, pleasure in new experiences is not enough. Rather, being able to regulate activity level, to focus and control attention, and thereby to maintain interest facilitates performance. E. RELATIONSBETWEENTEMPERAMENTAND MEMORY: DIFFERENCESWITHINSUBJECTS The Kleinknecht et al. (2002) study provides insight into relations between recall performance and temperament throughout the 2nd year of life. Over this space of time, being able to sustain and focus attention and to regulate activity level apparently becomes increasingly important to successful mnemonic performance. The results of work by Burch and Bauer (2002) suggest that these characteristics also shift in significance within subjects, as task demands change. The major

128

Bauer et al.

purpose of this study was to examine effects of a different source of individual differences on young children's recall, namely, variation in maternal language during initial learning and delayed-recall testing. The motivation for and findings of the study as they relate to that purpose are discussed later (see section V.C). At this point we outline the design of the study and describe relations between parent-reported temperament characteristics and recall memory as a function of task demands. 1. Study Design Typically in the elicited- or deferred-imitation paradigm, to-be-remembered events are modeled by an experimenter and children are encouraged to imitate. However, in the Burch and Bauer (2002) study, we were interested in examining possible effects of variability in maternal language on young children's recall. To do so, we had mothers, rather than experimenters, test the children. Sixteen mothers and their 24-month-old children participated. Mothers demonstrated the test sequences for their children and then elicited their children's imitation (i.e., immediate recall) of the sequences. One week later, the dyads returned to the laboratory and mothers tested their children's delayed recall. To examine effects of cognitive demand on maternal language and children's recall, we presented the dyads with sequences of three different lengths: four-step, five-step, and six-step. Based on previous related research, we expected four-step sequences to be relatively easy for the 24-month-olds (Bauer et al., 2000) and five-step sequences to be an optimal level of challenge (Bauer & Travis, 1993). We expected six-step sequences to represent a level of difficulty within or perhaps even beyond the 24-month-olds' "zone of proximal development," because not until 30 months do children typically perform at above-chance levels on six-step sequences (Bauer, Dow, Bittinger, & Wenner, 1998; Bauer & Fivush, 1992). In addition to data on children's immediate and 1-week delayed-recall performance, we collected TBAQs for the children. 2. Relations Between Child Temperament Characteristics and Recall On the relatively "easy for age" four-step sequences, higher scores on the Activity Level subscale of the TBAQ were related to higher levels of both immediate and delayed recall of the individual actions of the sequences ( r = .58 and .57, respectively; unless noted, alpha levels are p < .05). However, activity was not clearly related to memory, per se, because in the baseline phase, prior to modeling, children rated higher on Activity Level tended to produce more of the individual actions of the sequences spontaneously (r = .42, p = . 10) and the relation between Activity Level and spontaneous production of ordered pairs was reliable (r = .63). Thus, a characteristically high level of activity was perhaps beneficial to engagement in the imitation task and this carded over to recall performance when task demands were low. Consistent with this suggestion, on the five-step sequences (considered to represent an optimal level of challenge), Activity Level was related

Developments in Early Recall

129

to immediate recall of the individual actions of the events (r = .50). The relation was not observed for ordered recall or for recall after the delay, both of which impose greater demands. Moreover, Activity Level was not related to performance on the most challenging six-step sequences. That positive relations with the Activity Level subscale were observed only when task demands were low accounts for the different patterns of relations between children's scores on this subscale and their performance in the present sample relative to the Kleinknecht et al. (2002) study, in which the relations were negative. In that study, the cognitive demand was significantly greater, in that 20-monthold children's recall of four-step sequences was tested after a minimum delay of 1 month. In the Burch and Bauer (2002) study, children 4 months older were tested for recall immediately and after a 1-week delay. Across studies, the findings suggest that a characteristically high level of activity may be beneficial to performance when task demands are low but may interfere with performance when greater demands are imposed. In the Kleinknecht et al. (2002) study, we observed that high scores on the Interest and Persistence subscale were associated with high levels of long-term recall among 20-month-olds. A similar pattern was observed by Burch and Bauer (2002) on the most challenging sequences. That is, on the six-step sequences, higher scores on the Interest and Persistence subscale were associated with higher levels of delayed recall of the temporal order of the sequences (r = .50). Scores on the Interest and Persistence subscale were unrelated to performance on the fourand five-step sequences. Thus, greater control over attention was associated with strong performance on the most challenging sequences. In addition to the changing pattern of relations with the Activity Level and Interest and Persistence subscales as a function of task difficulty, there also was a pattern of changing relations between the Pleasure subscale and recall performance. Whereas on the four-step sequences there were no relations with Pleasure, on the five-step sequences, children rated high on Pleasure remembered fewer of the individual actions of the sequences over the delay ( r = - . 5 5 ) ; they also tended to remember fewer ordered pairs of actions ( r = - . 4 2 , p--.11). On the six-step sequences, the correlation between high ratings on the Pleasure subscale and delayed recall of the temporal order of the sequences was reliable (r = -.50). This pattern of negative correlation between scores on the Pleasure subscale and recall by 24-month-olds of the more challenging sequences is in contrast to the pattern observed for 9- and 13-month-olds, for whom high scores on the Pleasure subscale are positively related to recall performance. Based only on the currently available data, there is not a ready interpretation for this difference in direction of relation. However, we offer an interpretation when we consider relations between maternal language behavior and children's memory performance and between maternal language behavior and children's temperament characteristics (see section V.C).

130

B a u e r et al.

E SUMMARY

Although there are conceptual reasons to expect that children's temperament might be related to memory performance, there have been few studies of possible relations. In the context of the Bauer et al. (2000) study, we examined relations between parent-reported temperament characteristics and recall memory in 13- and 20-month-olds. Among the 13-month-olds, delayed recall was positively related to children's scores on the Pleasure subscale of the TBAQ. In contrast, among the 20-month-olds, delayed recall was positively related to children's scores on the Interest and Persistence subscale and negatively related to scores on the Activity Level subscale. Whether these relations will increase the proportion of variance in children's long-term recall for which we are able to account is yet to be seen (we were able to account for a maximum of 32% of the variance in Bauer et al., 2000): With the small samples that are available, we are not able to test multifactor regression models. In addition, the patterns await tests for replication using parent-report instruments, as well as tests based on direct observations of children's behavior in situations considered to be diagnostic of temperament (e.g., using the Laboratory Temperament Assessment Battery: Goldsmith & Rothbart, 1996). Finally, in a separate sample, relations between children's mnemonic performance and parental reports of children's temperament varied as a function of the level of challenge of the to-be-recalled material. Most significantly, whereas scores on the Activity Level subscale were related to performance under less challenging conditions (i.e., simpler sequences, immediate recall, memory for individual target actions as opposed to ordered recall), scores on the Interest and Persistence subscale were related to performance under the most challenging conditions (i.e., six-step sequences). These findings highlight the importance of evaluation of relations across multiple contexts that make different demands.

V. Children's Temperament and Mothers' Language as Interacting Sources of Individual Differences in Long-Term Recall The results of the Kleinknecht et al. (2002) and Burch and Bauer (2002) studies indicate that children's characteristic reactions and behaviors are a systematic source of variance in their recall memory performance. In the Kleinknecht et al. study, the effects were observed in a carefully controlled experimental context: Children were tested by professional staff trained to administer test materials following a constrained standard protocol. In contrast, in the Burch and Bauer study, the context was less controlled because children were tested by their mothers-none of whom had any prior experience with the elicited-imitation paradigm. It is noteworthy that the patterns of relations between children's temperament and their recall memory performances converged under such different conditions.

Developments in Early Recall

131

The less-well-controlled context of the Burch and Bauer study affords more than converging evidence for the results of Kleinknecht et al., however. Maternal administration of the test sequences provides the opportunity to examine relations between mother's behavior and children's mnemonic performance. In the literature on memory development in preschool-age and older children, maternal differences in verbal behavior in mnemonic contexts have predicted children's performance. In the section to follow we review this literature briefly. We then describe variability in maternal verbal behavior with 24-month-old children in the context of the elicited-imitation paradigm. Finally, we examine relations between maternal language, children's recall performance, and children's temperament characteristics. A. RELATIONSBETWEENMATERNALLANGUAGEAND OLDER CHILDREN'SMEMORYNARRATIVES The literature on preschool-age and older children reveals relations between variability in maternal language and children's autobiographical or personal memory narratives. In brief, researchers have observed two "styles" of conversation. Mothers who frequently engage in conversations about the past, provide rich descriptive information about previous experiences, and invite their children to "join in" on the construction of stories about the past are said to use an elaborative style. In contrast, mothers who provide fewer details about past experiences and instead pose specific questions to their children (e.g., "What was the name of the restaurant where we had breakfast?") are said to use a repetitive or low elaborative style. Maternal verbal stylistic differences have implications for children's autobiographical memory reports. Specifically, children of mothers whose language more closely approximates the elaborative style report more about past events than children of mothers whose language more closely resembles the repetitive style (e.g., Fivush & Fromhoff, 1988; Hudson, 1990; Tessler & Nelson, 1994). Relations between matemal language style and children's memory narratives are observed concurrently and over time. For example, Reese, Haden, and Fivush (1993) found that maternal use of a more elaborative style when children were 40 and 46 months of age facilitated children's independent narrative accounts at 58 and 70 months of age. B. MATERNALLANGUAGEVARIABILITYWITHYOUNGERCHILDREN IN THE CONTEXTOF ELICITEDIMITATION Whereas most of the work on relations between maternal language style and children's memory narratives has been conducted with older children, there has been some research with pre- and early verbal children. For example, Hudson (1990) reported effects of differential degrees of maternal verbal elaboration on 24- to 30-month-olds' participation in memory conversations (see also Farrant & Reese, 2000). In the Burch and Bauer (2002) study, we sought to expand this small

Bauer et al.

132

literature into the context of the elicited-imitation paradigm, thereby forging a link between the literatures on verbal and nonverbal recall. We asked mothers to present to their 24-month-old children four-, five-, and six-step sequences and then to elicit their children's recall of the sequences immediately and after a 1-week delay. To prepare them for their task, mothers were exposed to a silent videotape on which the sequences were demonstrated. Beyond that they were to "talk naturally"-mothers were given no instructions regarding how to verbally communicate as they demonstrated and then tested their children's memories for the sequences. To examine variability in maternal verbal style, we coded each of the mothers' utterances into one of the seven mutually exclusive and exhaustive categories described in Table IV. The first five categories, namely, elaborations, repetitions, affirmations, negations, and off-task, are those used in studies of maternal style variability in older children (Reese et al., 1993). The last two categories, deflections and regulations, were added to capture interactions that occurred as a result of the context of the prop-supported elicited-imitation task. Because very few negations were observed, they will not be discussed; although regulations of children's behavior were not infrequent, there were few correlations with this category and for this reason, it will not be discussed; and because off-task utterances are not related to the task, they will not be discussed (see Burch & Bauer, 2002, for additional details). Mothers exhibited variability in their language both when teaching the sequences and when testing their 24-month-old children's memories. As illustrated in Table V, the number of utterances in each category varied considerably, as did the total TABLE IV Categories of Maternal Utterances Used with 24-Month-Old Children in the Context of Elicited Imitation

Elaborations: Utterances that served to introduce an object or event or that provided additional information related to an object or event previously introduced. For example, "look at the farm stuff," and "where does the horse go on the farm?"

Repetitions: Utterances that were a verbatim or gist repetition of a prior utterance. For example, asking "what do we do on the farm?" after having just said "do we do something on the farm?"

Affirmations: Utterances that served to affirm a child's behavior or previous utterance. For example, "good job," or "that's fight."

Negations: Utterances that served to deny a child's behavior or previous utterance. For example, "is that what we do next?"

Off-task: Utterances not directed to the task, including such things as conversation with the professional staff member directing the sessions and comments about activities outside the laboratory. Deflections: Utterances that served to hold a place in the conversation (without providing any new information) or that directed the turn to the child. For example, "that's a farm (elaboration), isn't it (deflection)," and "look at this (regulation)---wow! (deflection)." Regulations: Utterances the purpose of which seemingly was to direct the child's behavior or attention to the task. For example, "sit down," and "watch me."

Developments in Early Recall

133

TABLE V Maternal Language Variability with 24-Month-Old Children in the Context of Elicited Imitationa Step length Four- step

Five-step

Six- step

Category of maternal utterance

M

(SD)

M

(SD)

M

(SD)

Elaborations

36.73

(10.86)

38.78

(9.32)

60.90

(22.42)

Repetitions

4.48

(3.76)

5.43

(3.97)

8.30

(5.19)

Affirmations

27.58

(14.54)

23.20

(12.80)

34.89

(18.57)

Deflections Total a M,

15.06

(5.05)

14.40

(7.04)

125.73

(36.90)

124.29

(26.19)

17.71

(8.03)

168.95 (41.70)

Mean; SD, standard deviation.

number of category tokens produced. The number of utterances also varied across sequences step lengths, such that across categories, mothers produced substantially more total category tokens on the six-step sequences relative to the four- and fivestep sequences, which did not differ from one another. Nevertheless, mothers' levels of production of the different utterance types were correlated across levels of difficulty of the events. For example, the number of elaborations used as mothers elicited their children's immediate recall of the four-step sequences was correlated with the number of elaborations used to elicit immediate recall of the five- and sixstep sequences (r = .80 and .58, respectively); the number of elaborations used to elicit recall of the five- and six-step sequences also was correlated (r = .85). In other words, although mothers varied in the number of tokens produced across levels of difficulty of the task, they nevertheless exhibited stability across sequence lengths. Thus, even though in Burch and Bauer (2002) we used a nonverbal task with 24month-olds, we observed the same type of maternal language variability that has been reported in the literature on verbal narrative production with preschool-age and older children. C. RELATIONS BETWEEN MOTHERS' LANGUAGE, CHILDREN'S RECALL PERFORMANCE, AND CHILDREN'S TEMPERAMENT CHARACTERISTICS

Having observed variability in measures related to maternal style, we explored whether it was related to children's immediate or 1-week delayed recall performance, or to children's temperament characteristics, or to both. As will become apparent, the patterns of relations differed as a function of the level of difficulty of the to-be-remembered events. Because relations on the five-step sequences were a mixture of the patterns observed on the simpler and more difficult sequence lengths,

134

Bauer et al.

for convenience, we present the relations for the four- and six-step sequence lengths only. We focus our discussion on the TBAQ subscales of Interest and Persistence, Pleasure, and Activity Level because, as observed in the Kleinknecht et al. (2002) study, they were the most predictive. Details on relations in the context of the fivestep sequences, and on the small number of relations involving the other subscales of the TBAQ, are available in Burch and Bauer (2002).

1. Four-Step Sequences On the relatively "easy for age" four-step sequences, maternal verbal behavior during the spontaneous, child-controlled baseline period was related to the total number of target actions that the children produced both immediately and after the delay. 3 Specifically, maternal use of elaborations ( r = .72), affirmations of children's behaviors (r = .62), and overall talkativeness (as measured by the total number of category tokens produced; r = .73) as children manipulated the eventrelated objects spontaneously all were correlated with the total number of target actions children produced in immediate recall (all p < .05). The same variables were correlated with the total number of target actions produced at delayed recall as well (r = .51, .85, and .79, respectively). Maternal verbal behavior was also related to children's engagement during immediate recall of the four-step sequences. Mothers who produced relatively more verbal elaborations during the immediate-recall period had children who produced both a larger number of total target actions and a larger number of different target actions (both r = .50). Production of a large number of total target actions by the children also was related to maternal repetition of utterances and to maternal affirmations (r -- .55 and .69, respectively). Maternal affirmations during the immediate-recall period also had a cross-lag relation to children's performance after the delay: Mothers who affirmed their children during immediate recall had children who were more engaged in the task at delayed recall, as measured by the total number of target actions produced (r = .58). The children also did their part to keep the system going: Children who produced a larger number of target actions during immediate recall had mothers who produced both a large number of affirmations and more total category tokens during delayed recall (r = .67 and .55, respectively). Finally, maternal affirmations during the delayed-recall period were related to production of individual target actions by the children (r = .64). 3 In the research discussed thus far, we have considered the number of different target actions that children produce as a measure of the exhaustiveness of recall and the number of pairs of actions produced in the target order as a measure of organization of the event representation. Early in our research on effects of maternal language on children's recall, we noted another dimension of difference, namely, variability in the sheer number of target actions (including repetitions) performed by the children on each trial. As discussed by Bauer (1993), we consider the total number of target actions produced, including repetitions, as a general indication of the children's engagement with the props and sequences.

Developments in Early Recall

135

Two patterns in the data from the four-step sequences are especially noteworthy. First, on these easy sequences, maternal behavior was not in response to the children's memory performance, per se: Systematic variability in maternal language was apparent in the baseline phase, before any memory behavior was observed. Second, maternal verbal behavior was related to children's levels of engagement in the task (i.e., to the total number of individual target actions produced) but not to the exhaustiveness of children's recall or to its organization. That is, there was only a single correlation between maternal verbal behavior and the diversity of target actions that the children produced; there were no correlations with ordered reproduction of the sequences. These patterns suggest that on event sequences that are well within a child's developmental level, the mother-child interaction is not related to memory but rather to the dyad's established patterns of interaction in the context of new materials or experiences. Both mothers and children behavemthe one verbally and the other nonverbally; mothers who are more verbally engaged have children who are more nonverbally engaged. Examination of the pattern of relations between maternal verbal behavior and children's temperament characteristics revealed that mothers were especially verbally engaged with children rated as having high levels of Interest and Persistence. Specifically, in the context of the simplest sequences, mothers who rated their children high on the Interest and Persistence subscale of the TBAQ produced more elaborations (r = .58), more repetitions (r = .55), and more total category tokens ( r - - . 6 6 ) during the baseline phase. At immediate recall, they produced more repetitions (r = .52), more affirmations (r = .62), more elaborations (r = .49), and more total category tokens (r = .53). At delayed recall, they produced more affirmations (r = .63). Thus, children who were perceived as typically showing evidence of interest and persistence received in the elicited-imitation context more verbal scaffolding during baseline and immediate recall and, to a lesser extent, during delayed recall. Higher levels of maternal verbal scaffolding were in turn related to higher levels of engagement in the task. In the context of the four-step sequences, relations between maternal verbal behavior and children's temperament characteristics were confined to the Interest and Persistence subscale; maternal verbal behavior was unrelated to either the Pleasure or Activity Level subscales of the TBAQ. Consideration of the relations observed among maternal language, children's recall performance, and their temperament characteristics, suggests two "pathways" to high levels of recall in the context of a relatively easy task. The first route is the direct one taken by children whose parents rated them as having high levels of activity. As reviewed earlier (see section IV.E.2), children rated high on Activity Level performed well, both immediately and after the delay. Because variability in children's reported levels of activity was not related to variability in maternal language, we interpret the relation between Activity Level and recall as direct: Under low levels of challenge, a characteristically high level of activity may have beneficial effects on performance.

136

Bauer et al.

The second route to a high level of performance is the indirect one afforded to children rated high in Interest and Persistence. Children whose parents perceived them as being high in Interest and Persistence were treated to high levels of maternal language during both baseline and recall. In turn, the children were highly engaged in the task, both immediately and after the delay. The association between levels of Interest and Persistence and children's performance is best characterized as indirect because scores on the Interest and Persistence subscale were not related to recall performance. Rather, the relations were between the temperament subscale and maternal behavior, and between maternal behavior and children's performance. We believe that this was observed because in the case of a relatively low level of challenge, such as that posed by four-step sequences, higher levels of selfgenerated and self-regulated attention are not necessary to ensure engagement in the task. Nevertheless, even in an "easy" context, a characteristically high level of attention is beneficial because of the high level of supportive maternal language that it engenders.

2. Six-Step Sequences In contrast to the four-step sequences, on the very difficult six-step sequences, maternal verbal behavior in the baseline phase was largely unrelated to children's behavior in the immediate recall period. The only relation observed was between total category tokens during baseline and children's engagement at immediate recall, as measured by the total number of target actions produced: r = .49, p = .05. The only other relations during the first session were between maternal verbal behavior and children's temperament characteristics. During baseline, mothers of children with low scores on the Pleasure subscale produced more elaborations (r = -.52); during modeling, they produced more deflections (r = -.50). Similarly, as they modeled the six-step sequences, mothers of children with lower scores on the Activity Level subscale produced more deflections (r = -.50) and more total category tokens (r = -.56). One interpretation of these patterns is that mothers whose children do not typically spontaneously show high levels of pleasure and activity saw greater need to engage their children in the more difficult sequences. The "ploy" was effective: Children whose mothers frequently deflected the turn to them as they modeled the event sequences during the first session produced a larger number of different target actions during delayed recall (r = .55). Whereas there were few concurrent relations between mothers' and children's behavior during the first session, there were a number of concurrent relations during the second session. Specifically, mothers who during delayed recall produced more elaborations, more repetitions, and more affirmations had children who were more engaged in the difficult task, as measured by the total number of target actions produced (r = .59, .54, and .80, respectively). Maternal verbal behaviors were delivered without regard for children's temperament characteristics. The only relation between children's temperament and maternal behavior at the second

Developments in Early Recall

137

session was that mothers of children highly rated on the Interest and Persistence subscale provided more affirmations during delayed recall ( r = .55). The sparse pattern of correlation with temperament characteristics on the more challenging sequences relative to the simpler four-step sequences suggests that as the level of challenge increased, mothers responded more to the "situational" factor of their children's needs for more scaffolding and support and less to the "dispositional" factor of whether, under typical circumstances, their children show a high level of interest and persistence. Indeed, as noted earlier (see section V.B), across the sample, mothers provided more total category tokens on the more difficult six-step sequences, relative to the easier four- and five-step sequences. Nevertheless, as reviewed earlier (see section IV.E.2), children's scores on the Interest and Persistence subscale were related to levels of delayed recall on the six-step sequences. Thus, against a backdrop of uniformly high levels of maternal verbal scaffolding (i.e., provided regardless of children's temperament), children with characteristically higher levels of self-regulation of attention and interest had an advantage over their less self-regulated peers, perhaps because they were better able to capitalize on the scaffolding provided by their mothers. Finally, as noted earlier (see section IV.E.2), on the six-step sequences, children with high scores on the Pleasure subscale had lower levels of delayed recall. Why might a behavioral style characteristic that seemingly benefited recall at 9 and 13 months of age (Bauer, Wiebe, et al., 2002, and Kleinknecht et al., 2002, respectively) seemingly impair it at 24 months? We suggest that mothers of children rated high on Pleasure were not providing them with high levels of verbal scaffolding. Indeed, maternal language was more frequently directed toward children low on Pleasure than to children high on Pleasure. Children with characteristically high levels of engagement in novel situations may have been perceived as not needing assistance with getting or staying involved in the task and thus did not receive it. Although we do not mean to imply that mothers were consciously aware of the relation, we note that given the history of their children, mothers had reason to believe this to be true: At younger ages, high levels of Pleasure are associated with high levels of performance. In effect then, in the context of a difficult task, children with higher levels of positive affectivity were doubly disadvantaged: (1) their temperament characteristic of Pleasure did not afford them a direct benefit (the situation called instead for high Interest and Persistence), yet (2) their mothers did not seem to know that they might need assistance! D. SUMMARY

Variability in the ways in which parents talk with their preschool-age and older children has pronounced effects on the children's concurrent and subsequent autobiographical memory narratives (e.g., Reese et aL, 1993). With few exceptions (e.g., Farrant & Reese, 2000; Hudson, 1990), the impact of these differences in

138

B a u e r et al.

maternal style on younger children, and on other memory tasks (e.g., elicited imitation), has not been examined. In the Burch and Bauer (2002) study, we found that mothers of 24-month-old children differed substantially from one another in their verbal behavior as they engaged their children in event sequences of different levels of challenge; variability in maternal language was related to variability in children's performance. On the simplest sequences, mother--child interaction was not related to memory, per se, as much as it was to the dyad's established patterns of interaction in the context of new materials or experiences: Mothers who were more verbally engaged had children who were more nonverbally engaged. On the most difficult sequences, mothers seemed to vary their verbal behavior in response to the greater level of challenge imposed on their children. Suggestively, maternal verbal behavior also was related to children's temperament characteristics. Again the pattern differed as a function of level of difficulty of the task. On the simplest, four-step, sequences, mothers seemed to behave in accord with their perceptions of their children's abilities to regulate their own attention and interest. Children rated as higher on Interest and Persistence were treated to more verbal tokens than children rated as lower on this subscale. Maternal verbal "interventions" seemingly were not directed toward moderation of children's pleasure or active involvement in the task. Conversely, on the more challenging, six-step, sequences, there were few relations with the Interest and Persistence subscale and negative relations with the Pleasure and Activity Level subscales. One interpretation of this pattern is that mothers were responding to the greater demands of the more challenging sequences and providing more verbal tokens, without regard for their children's typical levels of interest. Similarly, on the more challenging sequences, they may have perceived the need for greater verbal scaffolding for children with characteristically lower levels of Pleasure and Activity Level. Although the Burch and Bauer (2002) results need to be replicated, they provide the first evidence of relations between maternal verbal behavior and 2-year-old children's temperament characteristics and thus represent an important step in understanding the interactions between characteristics of the child and characteristics of her or his parent.

VI. Conclusions and Implications The end of the 20th century marked the beginning of the study of developments in recall during the first 2 years of life. Since the initiation of inquiry, we have made great strides in describing mnemonic behavior and in identifying its determinants. For example, we have established that the capacity for long-term ordered recall is newly or recently emergent at about 9 months of age (Bauer et al., 2001; Carver & Bauer, 1999, 2001) and that it becomes both prevalent and more robust over the course of the 2nd year of life (Bauer et al., 2000). Moreover, contrary to what we might have predicted in the face of tenacious assumptions that early memory

Developments in Early Recall

139

abilities were qualitatively different from later abilities (e.g., Piaget, 1952), without exception, factors that affect recall in older children and even in adults exert similar influences on recall memory in the first years of life (see, e.g., Bauer, 2002a, for a review). Because recall memory performance is multiply determined, we doubt the value in attempting to produce "growth chart" type functions of the lengths of time over which children of different ages will remember. Nevertheless, when we specify the conditions under which children will be tested (e.g., after a single experience or after multiple experiences, with or without verbal reminders), we can make sound predictions regarding the mean levels of performance of children of different ages. We have heretofore been less mindful of the variation about the mean, however. Limited attention on variability in early recall memory has been directed toward children's gender and their language proficiency. Individual studies, typically based on small samples, have revealed little evidence of individual differences associated with these potential sources of variance. The picture was not altered substantially when we analyzed for these sources of variability in the sample of 360 children who participated in the Bauer et al. (2000) study. Discussion of a third potential source of individual differences in long-term recall, namely, variability in initial learning, also proved disappointing, not because it accounts for little variance (on the contrary: see Bauer, Cheatham, et al., 2002) but because detection of the variance associated with differences in initial learning only shifts the focus of attention from the source of individual differences in long-term recall to the source of variability in initial learning. Children's gender and their vocabulary are plausible sources of variability in recall memory, but they were not perhaps the most "inspired" initial foci. In contrast, we suggest, on both behavioral and neurobiological grounds, examinations of relations between characteristics of children's temperaments and recall performance. Behaviorally, for example, a temperamentally inhibited child might, in a novel laboratory context, be less likely to engage in the tasks at hand. Conversely, children who are high in positive affect and approach, or children who regulate their own attention and interest, or both, might more readily engage in the task and as a result, evidence better memory. In addition, shared substrate serves as a basis for hypotheses concerning relations between children's temperament characteristics and their recall memory performance: The neural circuitry that supports recall memory overlaps with that implicated in regulation and control of emotion and attention. Although others have recognized this potentially rich soil (e.g., Greenhoot et aL, 1999; Gunnar & Nelson, 1994; Wachs et al., 1990), to our knowledge we are the first to work it in the age period during which both long-term recall ability and certain aspects of temperament (e.g., control of attention) consolidate and stabilize. In Figure 1 we provide a schematic representation of the observed relations. In the context of controlled laboratory tasks administered by professional staff who faithfully adhered to precise protocols, we found associations between temperament characteristics and memory. Specifically, 9-month-olds who were rated

140

Bauer et al.

Temperament

characteristics

I

Haternal "~ verbal "~1 behavior

Neurodevelopment .,

Basic mnemonic ability

First Year

~. Second Year

Fig. 1. Schematic representation of relations between neurodevelopmental factors, temperament characteristics, and basic mnemonic abilities in infancy, and maternal verbal behavior at the end of the 2nd year of life. Note that not all observed or hypothesized relations are represented.

by their parents as having higher levels of positive affectivity evidenced higher levels of recall, relative to their same-age peers (Bauer, Wiebe, et al., 2002). Consistent with speculation that one source of the association between temperament and memory in this age period is shared neural substrate are findings of relations between measures of temperament and electrophysiological indices of brain activity (Gunnar & Nelson, 1994) and measures of recall memory and electrophysiological indices of recognition memory (Bauer, Wiebe, et al., 2002; Carver, Bauer, & Nelson, 2000). Thus, late in the 1st year of life, "cognitive" (memory) and "social" (temperament) measures are related to one another. We speculate that the relations are mediated by overlap in the neural substrates implicated in the respective domains. Throughout the 2nd year, there is a paucity of direct evidence of relations between changes in cognitive and social behavior and changes in brain. The lack of evidence is due in large part to 1- to 2-year-olds' intolerance of imaging procedures such as functional magnetic resonance or even electrophysiological recording. Nevertheless, as reviewed earlier, there is compelling logical evidence of relations between developments in brain and developments in both memory and temperament. There also is evidence of relations between temperament and memory. In the Kleinknecht et al. (2002) study, we reported that at 13 months of age, as they had been at 9 months, high levels of recall were related to dimensions of positive affectivity. At 20 months, high levels of recall were related to dimensions of attentional control. In the context of a laboratory task administered by 24-month-old

Developments in Early Recall

141

children's mothers who were free to create their own protocols, high levels of recall were related to dimensions of activity under less challenging conditions and to dimensions of attentional control under more challenging conditions (Burch & Bauer, 2002). Together, these patterns could be taken to indicate that early in development, before infants and young children have much in the way of selfregulatory capacity, their characteristic reactions and affective responses influence their encoding and subsequent memory. As children develop the capacity to direct and control their own attentional resources, and thus to control their own actions and reactions, that capacity becomes a prominent determinant of mnemonic behavior, especially under conditions of cognitive challenge. The Burch and Bauer (2002) study, in which children's mothers served as their "experimenters," revealed that the dimensions of temperament that influence mnemonic performance differ as a function of task difficulty. It also provided insight on interactions between one source of individual differences in children's recall memory, namely, temperament, and another, namely, maternal verbal behavior. We found that, as it is in preschool-age and older children (e.g., Reese et al., 1993; Tessler & Nelson, 1994), variability in maternal verbal behavior was related to variability in 24-month-old children's recall. Specifically, mothers who used more elaborative language had children who were more engaged in the memory task. Maternal verbal behavior also was related to children's temperament characteristics. When memory demands were well within 24-month-old children's range of competence (i.e., on four-step sequences), mothers of children rated as high on the Interest and Persistence subscale used more elaborative language. As memory demands increased (i.e., on six-step event sequences), relations no longer were apparent. This suggests that temperament characteristics not only are direct sources of variance in children's recall, but also that they influence other factors that themselves contribute to variability in performance. Whether relations between measures of temperament and matemal verbal behavior in a mnemonic context would obtain in children earlier in the 2nd year of life is not known at this time. Perhaps because variations in children's temperament are more obvious or more ubiquitous than variations in mnemonic behavior, the mothers of the 24-monthold children in the Burch and Bauer (2002) study seemed to respond more to the "social" characteristic of their children's temperaments than to the "cognitive" characteristic of their children's mnemonic performance. That is, at the lower levels of challenge, maternal verbal behavior varied as a function of children's characteristic levels of attentional control. In contrast, regardless of the level of challenge, there were few relations between children's mnemonic performance in one phase of the protocol and maternal verbal behavior in the next phase of the protocol. This is consistent with reports from the autobiographical narrative literature with older children. In conversations with their preschool-age children, mothers exhibiting a more elaborative style behave "elaboratively" regardless of whether their children are contributing to the conversation (e.g., Fivush & Fromhoff, 1988). Indeed, one

142

Bauer et al.

of the "hallmarks" of an elaborative style is that elaborations are offered, even of the most paltry contributions. What are the developmental implications of the web of relations between neurodevelopment, cognitive, and social influences on early memory? In the age of speculation that early mnemonic abilities were limited and qualitatively different from later abilities (e.g., Piaget, 1952), the relations would not necessarily have held interest beyond the bounds of infancy and very early childhood. However, in the age of recognition of essential continuities in mnemonic processes across wide developmental spans (e.g., Bauer, 2002a; Howe & Courage, 1993; Meltzoff, 1995), they compel consideration of how early patterns of association might relate to later autobiographical or personal narrative competence as well as later episodic memory and narrative ability more broadly. Indeed, we have grounds to speculate that the observed relations between characteristics of 2-year-old children and maternal verbal behavior in a mnemonic task have implications for later memory performance. In Figure 2 we provide a schematic illustration of just some of the relations that already have been observed. In section V.A, we mentioned existing evidence of relations between maternal verbal behavior and children's autobiographical memory narratives. Both concurrently and over time, maternal elaboration in joint mother-child conversational contexts is related to children's autobiographical memory contributions (e.g., Reese et al., 1993). Preliminary analyses of ongoing research in our laboratory suggest that maternal elaboration in the context of joint reminiscing also has implications for episodic memory more broadly as well as for general narrative conpetence. For example, mothers who use a more elaborative style in the context of reminiscing with their 6- to 9-year-old children have children who recall more items in a sort-recall task (E J. Bauer, M. M. Burch, M. Larkina, & A. Tian, work in progress). Maternal elaboration also is related to 6- to 9-year-old children's independent non-mnemonic narrative competence: Mothers who use a more "elaborative" questioning style have children who produce higher quality narratives in a story production task (Wenner, Lynch, Wilson, Bramhall, & Galbraith, 2002). Based on findings such as these and those reported in sections IV and V, we seem to be on relatively firm ground in speculating that, early in life, children who characteristically evidence the temperament characteristic of Interest and Persistence will be treated to a more elaborative maternal style. In turn, over the course of development, a more elaborative maternal style will be related to greater autobiographical narrative and episodic memory abilities as well as general narrative competence. We make two qualifications to this discussion. First, in this report we have focused on children's temperament characteristics as a source of systematic variability in mnemonic behavior. Temperament characteristics likely are not the only individual difference that eventually will prove to be related to the behavior of socializing agents such as parents and thus, directly or indirectly, to memory. As

143

Developments in Early Recall

Children's language

Autobiographical memory

!

Temperament characteristics

Maternal verbal "J behavior behavi,,,

I Episodic memory

Neurodevelopment Narrative competence

Basic mnemonic ability

Infancy

~-

Early Childhood

Fig. 2. Schematic representation of relations between neurodevelopmental factors, temperament characteristics, and basic mnemonic abilities in infancy; and maternal verbal behavior at the end of the 2nd year of life; as well as relations between children's language competence in infancy and maternal verbal behavior at the end of the 2nd year; and maternal verbal behavior at the end of the 2nd year and subsequent developments in autobiographical memory, episodic memory, and general narrative competence in early childhood. Note that not all observed or hypothesized relations are represented.

suggested by inspection of Figure 2, another likely candidate source of influence is children's language. In section III.B, we discussed the lack of direct relations between expressive language (at 13, 16, and 20 months) and receptive language (at 13 and 16 months) and nonverbal expression of memory. However, just as there are indirect links between memory and maternal verbal behavior, via temperament, so might there be indirect links between memory and language, via maternal verbal behavior. Although not discussed in the current chapter, in the Burch and Bauer (2002) study, we found that children with higher reported productive vocabularies had mothers who were more elaborative in the context of presentation of the four-, five-, and six-step sequences. As we observed with temperament, the relations were strongest on the shortest event sequences, and as sequence length and thus challenge increased, the relations diminished. Again, we see that when material is within children's levels of competence, mothers seemingly respond to

144

Bauer et al.

their children's "dispositions"; as material becomes more challenging, mothers seemingly respond to the situation, and provide more verbal support (regardless of their children's dispositions). It is important to note that even if, on more challenging material, maternal behavior does not differ as a function of children's characteristics, these characteristics still matter. In the Burch and Bauer (2002) study, we found correlations that suggested that children with high levels of Interest and Persistence were able to take advantage of the increases in verbal support that all of the mothers provided on the more challenging sequences, and they were able to parlay it into higher performance on the most difficult tasks. That is, there were relations between Interest and Persistence and maternal elaborations as well as relations between maternal elaborations and children's mnemonic performance. We may speculate that a similar pattern would be observed with children's language. That is, children with higher levels of language competence may be better positioned to take advantage of the increased verbal scaffolding provided by mothers on more difficult tasks (i.e., they would be able to understand more of what their mothers were saying). The second qualification that we make is that Figure 2 does not provide an exhaustive depiction of existing or hypothesized relations. Rather, Figure 2 represents the major links discussed in this report (largely on the left side of the figure) and some of their possible implications for later development (largely on the fight side of the figure); consistent with their prominent roles in this report, temperament characteristics and maternal verbal behavior are emphasized. Links representing (a) continuity in mnemonic function across the time frame and (b) between-domain relations in early childhood are just some of the elements missing from the figure. That there is continuity in mnemonic ability across the period of transition from infancy to early childhood, and even beyond, is suggested by data from long-term longitudinal research in our laboratory. The first link in the chain is that basic mnemonic abilities late in infancy, as measured in the elicited-imitation paradigm, are related to verbal expression of episodic memory at age 3 years (Bauer, Kroupina, et al., 1998). Second, at age 3 years, children's verbal expressions of episodic memory are related to their verbal contributions in the context of mother--child autobiographical reminiscing (Bauer & Burch, 2002). Third, children's contributions in the context of joint autobiographical reminiscing at age 3 years are related to independent autobiographical narratives at ages 6 to 9 years (Van Abbema, 2002). There also are relations between types of memory in early childhood. For example, preschoolers' autobiographical memory narrative production skills are related to independent measures of episodic memory task performance (Kleinknecht & Beike, 2002). It will be left to a future report to provide a more fully elaborated picture of the web of relations between and among mnemonic and extramnemonic contributors to adultlike memory performance across a variety of domains.

Developments in Early Recall

145

Finally, throughout this chapter we have focused on normative trends and individual differences in typically developing infants and young children. We close by suggesting that in addition to affecting memory in typically developing children, individual differences in characteristics such as temperament and perhaps language competence might be related to the course of memory development in children from special populations. Specifically, they may function as "protective" factors in populations of children at risk for memory deficits (Bauer, 2001). At the close of the 20th century, most of the extant research on cognitive outcomes in at-risk populations concerned rather global measures of cognitive function, such as the Denver Developmental Scales and the Bayley Scales of Infant Development. To address the need for more fine-grained assessments, researchers began to use specific tasks, such as deferred imitation, to determine whether in special populations, deficits in mnemonic function are apparent. In one study of memory in a population potentially at risk, we used immediate and 10-min deferred imitation to examine mnemonic function in infants who had been born prematurely but who were otherwise healthy. Performance after the 10-min delay was correlated with gestational age at birth (i.e., infants born prematurely performed poorly on the task, relative to full-term infants), suggesting that deferred imitation may be especially sensitive to subtle differences in cognitive function associated with premature birth (de Haan, Bauer, Georgieff, & Nelson, 2000). Similarly, in Kroupina, Parker, Bruce, Gunnar, and Bauer (2000), we found that children who as infants had been adopted from international orphanages showed deficits on the 10-min deferred-imitation task, relative to matched homereared infants. Currently in progress are studies employing the 10-min deferredmemory task with infants born to diabetic mothers who, as a result, were prenatally iron deficient (C. A. Nelson, M. K. Georgieff, & P. J. Bauer, work in progress) and with infants who have been physically abused or neglected (D. Cicchetti, S. Toth, & P. J. Bauer, work in progress). In each of these cases, there is reason to believe that development of hippocampal structures may have been undermined, with resuiting deficits in explicit memory. If, in these samples, relations between infants' and children's temperament and maternal verbal behavior, or between language development and maternal verbal behavior, or both, obtain, then they might work to "protect" infants and young children from the worst of these deficits. In conclusion, in the last 2 decades of the 20th century, we documented general developmental trends in early recall memory. We discovered systematic interactions among factors that influence recall competence and performance in the first years of life, including task difficulty, children's temperament characteristics, and maternal language. Major tasks for the first decades of the 21 st century will be to elaborate these relations and to systematically explore other potential mediators and moderators of early recall memory development. This course will ensure continued progress in understanding normative developmental trends as well as the individual variability that heretofore has been largely hidden behind the mean.

146

Bauer et al.

ACKNOWLEDGMENTS Support for the research reported in this chapter was provided by grants from the National Institutes of Health (HD-28425) to Patricia Bauer. Additional support for preparation of the manuscript was provided for Erica Kleinknecht through an Institutional National Research Service Award provided by the National Institutes of Mental Health (MH-15755) to the Institute of Child Development at the University of Minnesota. We also thank the many collaborators who have helped to make the research possible. Those most directly associated with the research discussed are Stephanie Bangston, Leslie Carver, Patricia Dropik, Christine Leehey, Jennifer Rademacher, Emily Stark, Jennie Waters, Jennifer Wenner, and Sandi Wewerka. We also thank Hill Goldsmith for consultation and the children and parents who graciously volunteered their time and energy to the efforts reported herein.

REFERENCES Bachevalier, J., & Mishkin, M. (1994). Effects of selective neonatal temporal lobe lesions on visual recognition memory in rhesus monkeys. The Journal of Neuroscience, 14, 2128-2139. Barnett, M. P., Newman, H. W., Richardson, J. T. E., Thompson, P., & Upton, D. (2000). The constituent structure of autobiographical memory: Autobiographical fluency in people with chronic epilepsy. Memory, 8, 413-424. Barr, R., Dowden, A., & Hayne, H. (1996). Developmental changes in deferred imitation by 6- to 24-month-old infants. Infant Behavior and Development, 19, 159-170. Barr, R., & Hayne, H. (1996). The effect of event structure on imitation in infancy: Practice makes perfect? Infant Behavior and Development, 19, 253-257. Bauer, P. J. (1992). Holding it all together: How enabling relations facilitate young children's event recall. Cognitive Development, 7, 1-28. Bauer, P. J. (1993). Memory for gender-consistent and gender-inconsistent event sequences by 25-month-old children. Child Development, 64, 285-297. Bauer, P. J. (1995). Recalling past events: From infancy to early childhood. Annals of Child Development, 11, 25-71. Bauer, P. J. (1996). What do infants recall of their lives? Memory for specific events by 1- to 2-year-olds. American Psychologist, 51, 29-41. Bauer, P. J. (1997). Development of memory in early childhood. In N. Cowan (Ed.), The development of memory in childhood (pp. 83-111). Hove East Sussex: Psychology Press. Bauer, P. J. (2001, June). Advances in early memory development: Much to write home about. Address presented to the American Psychological Society, Toronto, Canada. Bauer, P. J. (2002a). Early memory development. In U. Goswami (Ed.), Handbook of cognitive development. Oxford: Blackwell. In press. Bauer, P. J. (2002b). New developments in the study of infant memory. In D. M. Teti (Ed.), Handbook of research methods in developmental psychology. Oxford: Blackwell, In press. Bauer, P. J., & Burch, M. M. (2002). Developments in early memory: Multiple mediators of foundational processes. In J. Lucariello, J. A. Hudson, R. Fivush, and P. J. Bauer (Eds.), Development of the mediated mind: Culture and cognitive development--Essays in honor of Katherine Nelson. Mahwah, NJ: Erlbaum. In press. Bauer, P. J., Cheatham, C. L., Cary, M. S., & Van Abbema, D. L. (2002). Short-term forgetting: Charting its course and its implications for long-term remembering. In S. P. Shohov (Ed.), Advances in psychology research, Vol. 9 (pp. 53-74). Huntington, NY: Nova Science. Bauer, P. J., Dow, G. A., Bittinger, K. A., & Wenner, J. A. (1998). Accepting and exempting the unexpected: 30-Month-olds' generalization of event knowledge. Cognitive Development, 13, 421452.

Developments in Early Recall

147

Bauer, P. J., & Fivush, R. (1992). Constructing event representations: Building on a foundation of variation and enabling relations. Cognitive Development, 7, 381-401. Bauer, P. J., & Hertsgaard, L. A. (1993). Increasing steps in recall of events: Factors facilitating immediate and long-term memory in 13.5- and 16.5-month-old children. Child Development, 64, 1204-1223. Bauer, E J., Hertsgaard, L. A., Dropik, E, & Daly, B. E (1998). When even arbitrary order becomes important: Developments in reliable temporal sequencing of arbitrarily ordered events. Memory, 6, 165-198. Bauer, E J., Hertsgaard, L. A., & Wewerka, S. S. (1995). Effects of experience and reminding on longterm recall in infancy: Remembering not to forget. Journal of Experimental Child Psychology, 59, 260-298. Bauer, E J., Kroupina, M. G., Schwade, J. A., Dropik, E, & Wewerka, S. S. (1998). If memory serves, will language? Later verbal accessibility of early memories. Development and Psychopathology, 10, 655-679. Bauer, E J., & Mandler, J. M. (1992). Putting the horse before the cart: The use of temporal order in recall of events by one-year-old children. Developmental Psychology, 28, 441-452. Bauer, P. J., & Thal, D. J. (1990). Scripts or scraps: Reconsidering the development of sequential understanding. Journal of Experimental Child Psychology, 50, 287-304. Bauer, E J., & Travis, L. L. (1993). The fabric of an event: Different sources of temporal invariance differentially affect 24-month-olds' recall. Cognitive Development, 8, 319-341. Bauer, E J., Van Abbema, D. L., & de Haan, M. (1999). In for the short haul: Immediate and shortterm remembering and forgetting by 20-month-old children. Infant Behavior and Development, 22, 321-343. Bauer, E J., Wenner, J. A., Dropik, E L., & Wewerka, S. S. (2000). Parameters of remembering and forgetting in the transition from infancy to early childhood. Monographs of the Society for Research in Child Development, 65(4, Serial No. 263). Bauer, E J., Wenner, J. A., & Kroupina, M. G. (2002). Making the past present: Verbal reports of preverbal memories. Journal of Cognition and Development, 3, 21-47. Bauer, E J., & Wewerka, S. S. (1995). One- to two-year-olds' recall of events: The more expressed, the more impressed. Journal of Experimental Child Psychology, 59, 475-496. Bauer, E J., & Wewerka, S. S. (1997). Saying is revealing: Verbal expression of event memory in the transition from infancy to early childhood. In E van den Brock, P. J. Bauer, & T. Bourg (Eds.),

Developmental spans in event comprehension and representation: Bridging fictional and actual events (pp. 139-168). Mahwah, NJ: Erlbaum. Bauer, E J., Wiebe, S. A., Carver, L. J., Waters, J. M., & Nelson, C. A. (2002). Developments in longterm explicit memory late in the first year of life: Behavioral and electro physiological indices. Manuscript in review. Bauer, E J., Wiebe, S. A., Waters, J. M., & Bangston, S. K. (2001). Reexposure breeds recall: Effects of experience on 9-month-olds' ordered recall. Journal of Experimental Child Psychology, 80, 174-200. Burch, M. M., & Bauer, E J. (2002). Relations between maternal style, children's recall memory, and temperamental characteristics. Manuscript in preparation. Cabeza, R., Kapur, S., Craik, E I. M., McIntosh, A. R., Houle, S., & Tulving, E. (1997). Functional neuroanatomy of recall and recognition: A PET study of episodic memory. Journal of Cognitive Neuroscience, 9, 254-265. Carey, W. B., & McDevitt, S. C. (1978). Stability and change in individual temperament diagnoses from infancy to early childhood. Journal of the American Academy of Child Psychiatry, 17, 331-337. Carver, L. J., & Bauer, P. J. (1999). When the event is more than the sum of its parts: Nine-month-olds' long-term ordered recall. Memory, 7, 147-174. Carver, L. J., & Bauer, P. J. (2001). The dawning of a past: The emergence of long-term explicit memory in infancy. Journal of Experimental Psychology: General 130, 726-745.

148

Bauer et al.

Carver, L. J., Bauer, E J., & Nelson, C. A. (2000). Associations between infant brain activity and recall memory. Developmental Science, 3, 234-246. Chugani, H. T. (1994). Development of regional brain glucose metabolism in relation to behavior and Plasticity. In G. Dawson & K. W. Fisher (Eds.), Human behavior and the developing brain (pp. 153-175). New York: Guilford. de Haan, M., Bauer, P. J., Georgieff, M. K., & Nelson, C. A. (2000). Explicit memory in low-risk toddlers born between 27-42 weeks of gestation. Developmental Medicine and Child Neurology, 42, 304-312. DeLoache, J. S., Cassidy, D. J., & Brown, A. L. (1985). Precursors of mnemonic strategies in very young children's memory. Child Development, 56, 125-137. Derryberry, D., & Reed, M. A. (1996). Regulatory processes and the development of cognitive representations. Development and Psychopathology, 8, 215-234. Derryberry, D., & Rothbart, M. K. (1997). Reactive and effortful processes in the organization of temperament. Development and Psychopathology, 9, 633-652. Dudycha, G. J., & Dudycha, M. M. (1941). Childhood memories: A review of the literature. Psychological Bulletin, 38, 668-682. Dywan, J., & Jacoby, L. L. (1990). Effects of aging on source monitoring: Differences in susceptibility to false fame. Psychology and Aging, 5, 379-387. Eichenbaum, H., & Cohen, N. J. (2001). From conditioning to conscious recollection: Memory systems of the brain. New York: Oxford University Press. Fagan, J. E (1973). Infants' delayed recognition memory and forgetting. Journal of Experimental Child Psychology, 16, 424-450. Fagan, J. F. (1984). The intelligent infant: Theoretical implications. Intelligence, 8, 1-9. Fantz, R. L. (1956). A method for studying early visual development. Perceptual and Motor Skills, 6, 13-15. Farrant, K., & Reese, E. (2000). Maternal style and children's participation in reminiscing: Stepping stones in children's autobiographical memory development. Journal of Cognition and Development, 1, 193-225. Fenson, L., Dale, P. S., Renznick, J. S., Bates, E., Thai, D. J., & Pethick, S. J. (1994). Variability in early communicative development. Monographs of the Society for Research in Child Development, 59(5, Serial Number 242). Fivush, R. (1984). Learning about school: The development of kindergartners' school scripts. Child Development, 55, 1697-1709. Fivush, R., & Fromhoff, F. (1988). Style and structure in mother-child conversations about the past. Discourse Processes, 11, 337-355. Fivush, R., Gray, J. T., & Frornhoff, E A. (1987). Two-year-olds talk about the past. Cognitive Development, 2, 393-409. Fivush, R., & Hamond, N. R. (1989). Time and again: Effects of repetition and retention interval on 2-year-olds' event recall. Journal of Experimental Child Psychology, 47, 259-273. Freud, S. (1905/1953). Three essays on the theory of sexuality. In J. Strachey (Ed.), The standard edition of the complete psychological works of Sigmund Freud (Vol. 7, pp. 135-243). London: Hogarth Press. Freud, S. (1916/1966). Introductory lectures on psychoanalysis. Translated and edited by J. Strachey. New York: Norton. (Original work published 1916-1917). Friedman, S. (1972). Habituation and recovery of visual response in the alert human newborn. Journal of Experimental Child Psychology, 13, 339-349. Geddie, L., Fradin, S., & Beer, J. (2000). Child characteristics which impact accuracy of recall and suggestibility in preschoolers: Is age the best predictor? ChiM Abuse and Neglect, 24, 223-235. Gelman, R., & Baillargeon, R. (1983). A review of some Piagetian concepts. In J. H. Flavell & E. M. Markman (Eds.), Cognitive development (pp. 167-230). New York: Wiley.

Developments in Early Recall

149

Goldsmith, H. H. (1996). Studying temperament via construction of the toddler behavior assessment questionnaire. Child Development, 67, 218-235. Goldsmith, H. H., & Rothbart, M. K. (1996). The Laboratory Temperament Assessment Battery (LABTAB). Department of Psychology, University of Wisconsin, Madison, WI. Gordon, B. N., Ornstein, P. A., Nida, R. E., Follmer, A., Crenshaw, M. C., & Albert, G. (1993). Does the use of dolls facilitate children's memory of visits to the doctor? Applied Cognitive Psychology, 7, 459-474. Greenhoot, A. F., Ornstein, P. A., Gordon, B. N., & Baker-Ward, L. (1999). Acting out details of a pediatric check-up: The impact of interview condition and behavioral style on children's memory reports. Child Development, 70, 363-380. Gunnar, M. R. (1990). The psychobiology of infant temperament. In J. Colombo, & J. Fagan (Eds.), Individual differences in infancy: Reliability, stability, prediction (pp. 387-409). Hillsdale, NJ: Erlbaum. Gunnar, M. R., & Nelson, C. A. (1994). Event-related potentials in year-old infants predict negative emotionality and hormonal responses to separation. Child Development, 65, 80-94. Howe, M. L. (2000). Historical and future trends in studying the development of long-term retention. Commentary on Bauer et al., Monographs of the Society for Research in Child Development, 65(4, Serial No. 263). Howe, M. L., & Courage, M. L. (1993). On resolving the enigma of infantile amnesia. Psychological Bulletin, 113, 305-326. Hudson, J. A. (1986). Memories are made of this: General event knowledge and development of autobiographic memory. In K. Nelson (Ed.), Event knowledge: Structure and function in development (pp. 97-118). Hillsdale, NJ: Edbaum. Hudson, J. A. (1990). The emergence of autobiographical memory in mother-child conversation. In R. Fivush & J. A. Hudson (Eds.), Knowing and remembering in young children (pp. 166-196). Cambridge, MA: Cambridge University Press. Hudson, J. A., & Fivush, R. (1991). As time goes by: Sixth graders remember a kindergarten experience. Applied Cognitive Psychology, 5, 346-360. Hudson, J. A., & Sheffield, E. G. (1998). D6jh vu all over again: Effects of reenactment on toddlers' event memory. Child Development, 69, 51-67. Hutt, C. (1978). Biological bases of psychological sex differences. American Journal of Diseases in Children, 132, 170-177. Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of memory. Journal of Memory and Language, 30, 513-541. Janowsky, J. S., Shimamura, A. P., & Squire, L. R. (1989). Memory and metamemory: Comparisons between patients with frontal lesions and amnesic patients. Psychobiology, 17, 3-11. Jetter, W., Poser, U., Freeman, Jr., R. B., & Markowitsch, H. J. (1986). A verbal long-term memory deficit in frontal lobe damaged patients. Cortex, 22, 229-242. Kail, R. V., & Salthouse, T. A. (1994). Processing speed as a mental capacity. Acta Psychologica, 86, 199-225. Kandel, E. R., & Squire, L. R. (2000). Neuroscience: Breaking down scientific barriers to the study of brain and mind. Science, 290, 1113-1120. Kapur, S., Craik, F. I. M., Jones, C., Brown, G. M., Houle, S., & Tulving, E. (1995). Functional role of the prefrontal cortex in memory retrieval: A PET study. Neuroreport, 6, 1880-1884. Kleinknecht, E. E., & Beike, D. O. R. (2002). How knowing and doing inform an autobiography: Relations among preschoolers' theory of mind, narrative, and event memory skills. Manuscript in review. Kleinknecht, E. E., Rademacher, J., & Bauer, P. J. (2002). Pleasure gives way to interest and persistence: Changing relations between temperament and memory at 13 and 20 months. Manuscript in review.

150

Bauer et aI.

Kroupina, M. G., Parker, S. W., Bruce, J., Gunnar, M. R., & Bauer, P. J. (2000, July). Assessing the cognitive development of internationally adopted children. Poster presented at the International Conference on Infant Studies, Brighton, England. Lepage, M., & Richer, F. (1996). Inter-response interference contributes to the sequencing deficit in frontal lobe lesions. Brain, 119, 1289-1295. Mandler, J. M. (1984). Representation and recall in infancy. In M. Moscovitch (Ed.), Infant memory: Its relation to normal and pathological memory in humans and other animals (pp. 75-101). New York: Plenum. Mandler, J. M. (1990). Recall of events by preverbal children. In A. Diamond (Ed.), The development and neural bases of higher cognitive functions (pp. 485-516). New York: New York Academy of Science. Mandler, J. M. (1998). Representation. In D. Kuhn & R. S. Siegler (Vol. Eds.) Cognition, perception, and language, Vol. 2; W. Damon (Editor-in-Chief), Handbook of child psychology, Fifth Edition (pp. 255-308). New York: Wiley. Mandler, J. M., & McDonough, L. (1995). Long-term recall of event sequences in infancy. Journal of Experimental Child Psychology, 59, 457-474. Markowitsch, H. J. (1995). Which brain regions are critically involved in the retrieval of old episodic memory? Brain Research Reviews, 21, 117-127. Martin, R. P. (1988). The temperament assessment battery for children. Brandon, VT: Clinical Psychological. McDonough, L., Mandler, J. M., McKee, R. D., & Squire, L. R. (1995). The deferred imitation task as a nonverbal measure of declarative memory. Proceedings of the National Academy of Sciences USA, 92, 7580-7584. Meltzoff, A. N. (1985). Immediate and deferred imitation in fourteen- and twenty-four-month-old infants. Child Development, 56, 62-72. Meltzoff, A. N. (1988a). Infant imitation after a 1-week delay: Long-term memory for novel acts and multiple stimuli. Developmental Psychology, 24, 470-476. Meltzoff, A. N. (1988b). Infant imitation and memory: Nine-month-olds in immediate and deferred tests. Child Development, 59, 217-225. Meltzoff, A. N. (1990). The implications of cross-modal matching and imitation for the development of representation and memory in infants. In A. Diamond (Ed.), The development and neural bases of higher cognitive functions (pp. 1-37). New York: New York Academy of Science. Meltzoff, A. N. (1995). What infant memory tells us about infantile amnesia: Long-term recall and deferred imitation. Journal of Experimental Child Psychology, 59, 497-515. Moscovitch, M. (1992). Memory and working-with-memory: A component process model based on modules and central systems. Journal of Cognitive Neuroscience, 4, 257-267. Mullen, M. K. (1994). Earliest recollections of childhood: A demographic analysis. Cognition, 52, 55-79. Murray, E. A., & Mishkin, M. (1998). Object recognition and location memory in monkeys with excitotoxic lesions of the amygdala and hippocampus. Journal of Neuroscience, 18, 6568-6582. Nelson, C. A. (1995). The ontogeny of human memory: A cognitive neuroscience perspective. Developmental Psychology, 31, 723-738. Nelson, C. A. (1997). The neurobiological basis of early memory development. In N. Cowan (Ed.), The development of memory in childhood (pp. 41-82). Hove East Sussex: Psychology Press. Nelson, C. A., & Dukette, D. (1998). A cognitive neuroscience perspective on the relation between attention and memory development. In J. E. Richards (Ed.), Cognitive neuroscience of attention: A developmental perspective (pp. 327-362). Mahwah, NJ: Erlbaum. Nelson, C. A., & Webb, S. J. (2002). A cognitive neuroscience perspective on early memory development. In M. de Haan & M. H. Johnson (Eds.), The cognitive neuroscience of development. London: Psychology Press. In press.

Developments in Early Recall

151

Nelson, K., & Gruendel, J. (1981 ). Generalized event representations: Basic building blocks of cognitive development. In M. E. Lamb & A. L. Brown (Eds.),Advances in developmentalpsychology (Vol. 1, pp. 131-158). Hillsdale, NJ: Erlbaum. Nelson, K., & Gruendel, J. (1986). Children's scripts. In K. Nelson (Ed.), Event knowledge: Structure and function in development (pp. 21-46). Hillsdale, NJ: Erlbaum. Piaget, J. (1952). The origins of intelligence in children. New York: International Universities Press. Pillemer, D. B., & White, S. H. (1989). Childhood events recalled by children and adults. In H. W. Reese (Ed.), Advances in child development and behaviour (Vol. 21, pp. 297-340). San Diego: Academic Press. Posner, M. I., Petersen, S. E., Fox, P. T., & Raichle, M. E. (1988). Localization of cognitive operations in the human brain. Science, 240, 1627-1631. Ratner, H. H. (1980). The role of social context in memory development. In M. Perlmutter (Ed.), New directions for child developmentmChildren's memory (pp. 49-67). San Francisco: Jossey-Bass. Reese, E., Haden, C. A., & Fivush, R. (1993). Mother-child conversations about the past: Relationships of style and memory over time. Cognitive Development, 8, 403-430. Reinisch, J. M., Rosenblum, L. A., Rubin, D. B., & Schulsinger, M. E (1991). Sex differences in developmental milestones during the first year of life. Journal of Psychology and Human Sexuality, 4,19. Rose, S. A., Feldman, J. E, & Wallace, I. F. (1992). Infant information processing in relation to six-year cognitive outcomes. Child Development, 63, 1126-1141. Rothbart, M. K., Ahadi, S. A., & Hershey, K. L. (1994). Temperament and social behavior in childhood. Merrill-Palmer Quarterly, 40, 21-39. Rothbart, M. K., & Bates, J. E. (1998). Temperament. In W. S. E. Damon & N. V. E. Eisenberg (Vol. Eds.), Social, emotional and personality development, Vol. 3; W. Damon (Editor-in-Chief), Handbook of child psychology, Fifth Edition (pp. 105-176). New York: Wiley. Rothbart, M. K., Derryberry, D., & Posner, M. I. (1994). A psychobiological approach to development of temperament. In J. Bates & T. Wachs (Eds.), Temperament: Individual differences at the interface of biology and behavior. Washington, DC: American Psychological Association. Rothbart, M. K., & Posner, M. I. (2001). Mechanism and variation in the development of attentional networks. In C. A. Nelson & M. Luciana (Eds.), Handbook of developmental cognitive neuroscience (pp. 353-363). Cambridge, MA: MIT Press. Rovee, C., & Fagen, J. W. (1976). Extended conditioning and 24-hour retention in infants. Journal of Experimental Child Psychology, 21, 1-11. Rubin, D. C. (1982). On the retention function for autobiographical memory. Journal of Verbal Learning and Verbal Behavior, 21, 21-38. Schacter, D. L., & Moscovitch, M. (1984). Infants, amnesics, and dissociable memory systems. In M. Moscovitch (Ed.), Infant memory: Its relation to normal and pathological memory in humans and other animals (pp. 173-216). New York: Plenum. Schneider, W., & Bjorklund, D. E (1998). Memory. In D. Kuhn & R. S. Siegler (Vol. Eds.) Cognition, perception, and language, Vol. 2; W. Damon (Editor-in-Chief), Handbook of child psychology, Fifth Edition (pp. 467-521). New York: Wiley. Serres, L. (2001). Morphological changes of the human hippocampal formation from midgestation to early childhood. In C. A. Nelson & M. Luciana (Eds.), Handbook of developmental cognitive neuroscience (pp. 45-58). Cambridge, MA: MIT Press. Sheffield, E. G., & Hudson, J. A. (1994). Reactivation of toddlers' event memory. Memory, 2, 447465. Squire, L. R. (1982). The neuropsychology of human memory. Annual Review of Neuroscience, 5, 241-273. Squire, L. R. (1986). Mechanisms of memory. Science, 232, 1612-1619.

152

Bauer et al.

Squire, L. R. (1992). Memory and the hippocampus: A synthesis from findings with rats, monkeys, and humans. Psychological Review, 99, 195-231. Squire, L. R., Knowlton, B., & Musen, G. (1993). The structure and organization of memory. Annual Review of Psychology, 44, 453-495. Squire, L. R., & Zola-Morgan, S. (1991). The medial temporal lobe memory system. Science, 253, 1380-1386. Tessler, M., & Nelson, K. (1994). Making memories: The influence of joint encoding on later recall by young children. Consciousness and Cognition, 3, 307-326. Thomas, A., & Chess, S. (1977). Temperament and development. New York: Brunner Mazel. Van Abbema, D. L. (2002). Autobiographical memory in childhood: Development of a personal past. Unpublished doctoral dissertation. University of Minnesota. Vargha-Khadem, E, Gadian, D. G., Watldns, K. E., Connelly, A., Van Paesschen, W., & Mishkin, M. (1997). Differential effects of early hippocampal pathology on episodic and semantic memory. Science, 277, 376-380. Vecera, S. P., Rothbart, M. K., & Posner, M. I. (1991). Development of spontaneous alternation in infancy. Journal of Cognitive Neuroscience, 3, 351-354. Wachs, T. D., Morrow, J., & Slabach, E. H. (1990). Intra-individual variability in infant visual recognition memory performance: Temperamental and environmental correlates. Infant Behavior and Development, 13, 397-403. Wenner, J. A., & Bauer, P. J. (1999). Bringing order to the arbitrary: One- to two-year-olds' recall of event sequences. Infant Behavior and Development, 22, 585-590. Wenner, J. A., Lynch, J. S., Wilson, S., Bramhall, M. A., & Galbraith, S. (2002, June). Partners in story: The relation between children's narrative skill and parents' reminiscing style. Paper presented at the Society for Text and Discourse Conference, Chicago, Illinois. West, T. A., & Bauer, P. J. (1999). Assumptions of infantile amnesia: Are there differences between early and later memories? Memory, 7, 257-278. Yasuno, E, Hirata, M., Takimoto, H., Taniguchi, M., Nakagawa, Y., Ikejiri, Y., Nishikawa, T., Shinozaki, K., Tanabe, H., Sugita, Y., & Takeda, M. (1999). Retrograde temporal order amnesia resulting from damage to the fornix. Journal of Neurology, Neurosurgery, and Psychiatry, 67, 102-105. Zola-Morgan, S., & Squire, L. R. (1990). The primate hippocampal formation: Evidence for a timelimited role in memory storage. Science, 250, 288-290. Zola-Morgan, S., Squire, L. R., & Amaral, D. G. (1986). Human amnesia and the medial temporal region: Enduring memory impairment following bilateral lesion limited to field CA1 of the hippocampus. The Journal of Neuroscience, 6, 2950-2967.

INTERSENSORY REDUNDANCY

GUIDES EARLY

PERCEPTUAL AND COGNITIVE DEVELOPMENT

Lorraine E. Bahrick and Robert Lickliter DEPARTMENT OF PSYCHOLOGY FLORIDA INTERNATIONAL UNIVERSITY MIAMI, FLORIDA 33199

I. INTRODUCTION: HISTORICAL OVERVIEW AND PERSPECTIVES ON PERCEPTUAL D E V E L O P M E N T II. A M O D A L RELATIONS AND THE M U L T I M O D A L NATURE OF E A R L Y EXPERIENCE III. U N I M O D A L - M U L T I M O D A L D I C H O T O M Y IN D E V E L O P M E N T A L RESEARCH IV. N E U R A L AND B E H A V I O R A L EVIDENCE FOR INTERSENSORY INTERACTIONS V. INTERSENSORY R E D U N D A N C Y HYPOTHESIS: TOWARD AN INTEGRATED T H E O R Y OF P E R C E P T U A L D E V E L O P M E N T A. PREDICTIONS OF THE INTERSENSORY R E D U N D A N C Y HYPOTHESIS AND THE I M P O R T A N C E OF INCREASING SPECIFICITY IN E A R L Y DEVELOPMENT B. DIRECT EMPIRICAL SUPPORT FOR THE INTERSENSORY R E D U N D A N C Y HYPOTHESIS C. ON W H A T BASIS DOES INTERSENSORY R E D U N D A N C Y FACILITATE PERCEPTUAL DISCRIMINATION AND LEARNING? VI. S U M M A R Y AND DIRECTIONS FOR F U T U R E STUDY OF PERCEPTUAL D E V E L O P M E N T REFERENCES

I. Introduction: Historical Overview and Perspectives on Perceptual Development The world provides a richly structured, continuous flux of multimodal stimulation to all of our senses. Objects and events can be seen, heard, smelled, and 153 ADVANCESIN CHILDDEVELOPMENT AND BEHAVIOR,VOL.30

Copyright 2002, Elsevier Science (USA). All rights reserved.

0065-2407/02 $35.00

1b4

Bahrick and Lickliter

felt, and as we move around and interact with the people, places, and objects in our environment we produce continuous changes in proprioceptive and visual feedback from our exploratory activities. Our senses provide overlapping and redundant information for objects and events in our environment. Dating as far back as Aristotle's De Anima and De Sensu, scientists have been intrigued and challenged by issues arising from the specificity of stimulation from the different senses and the nature of the overlap among them. How are objects and events experienced as unitary when they stimulate receptors that give rise to such different forms of information? How are disparate forms of stimulation bound together? Aristotle postulated a "sensus communis"--an amodal or common sense--which he thought was responsible for perceiving the qualities of stimulation that were general and not specific to single senses ("common sensibles"). According to Aristotle, common sensibles included motion, rest, number, form, magnitude, and unity. These properties are remarkably similar to those characterized as amodal by contemporary perceptual theorists (Bahrick & Pickens, 1994; J. J. Gibson, 1966, 1979; Marks, 1978; Stoffregen & Bardy, 2001). Centuries later, Locke (1690/1971) and Berkeley (1709/1910), among others, took a different approach to intersensory perception, proposing that perceivers must learn to interpret and integrate sensations before meaningful perception of objects and events could be possible. Following this "constructionist" approach, most modem theories of perception have been founded on the assumption that the different forms of stimulation from the various senses must be integrated or organized in the brain and therefore pose a "binding" problem for perception. It was thought that sensory stimulation had to be united by mechanisms that translate information from different codes and channels into a common language (Muller's "Law of Specific Energies," 1838). The constructionist view permeated thinking about the development of perception during most of the 20th century (Birch & Lefford, 1963; Friedes, 1974; Piaget, 1952), with most investigators assuming that we must learn to coordinate and integrate the separate senses. From this view, information had to be integrated across the senses through a gradual process of association across development in order to perceive unified objects and events. This integration was thought to occur by interacting with objects, experiencing concurrent feedback from different senses, and associating, assimilating, or calibrating one sense to another. For example, according to Piaget (1952, 1954) not until well into the first half-year following birth do vision and touch begin to be integrated. Through acting on objects, tactile feedback gradually endows the two-dimensional visual image of an object with three dimensionality. The attainment of perceptual abilities such as size and shape constancy, visually guided reaching, and object permanence were thought by Piaget and his colleagues (e.g., Piaget & Inhelder, 1967) to be slow to emerge and to depend on the gradual development of sensory integration. Prior to this integration, the visual world of the infant was thought to consist of images shrinking, expanding, changing shape, and disappearing and then reappearing capriciously.

lntersensory Redundancy

155

Until the activity-based achievement of sensory integration, infants were thought to perceive only unrelated patterns of visual, acoustic, or tactile stimulation, expressed by the well-known description of the world of the newborn infant by William James (1890) as a "blooming, buzzing confusion." Not until J. J. Gibson's (1966, 1979) seminal work on the "ecological" view of perception was the integration perspective on perceptual development seriously questioned. In a sharp break from traditional views, Gibson recognized that the existence of different forms of sensory stimulation was not a problem for the perception of unitary events but instead provided an important basis for it. He argued that the senses interact and work together to pick up invariant aspects of stimulation and should be considered as a "perceptual system." One important type of invariant information is amodal information that is common across the senses. As pointed out by Aristotle, amodal information is not specific to a particular sensory modality but is information common to several senses. Temporal and spatial aspects of stimulation are typically conveyed in multiple senses. As a case in point, the rhythm or rate of a ball bouncing can be conveyed visually or acoustically and is completely redundant across the two senses. The sight and sound of hands clapping likewise share temporal synchrony, a common tempo of action, and a common rhythm. We now know from a prolific body of research conducted over the last 25 years of the 20th century, inspired in large part by Gibson's ecological approach to perception, that even young infants are adept perceivers of arnodal stimulation (see Bahrick & Pickens, 1994; Lewkowicz, 2000; Lickliter & Bahrick, 2000; Walker-Andrews, 1997). Infants detect the temporal aspects of stimulation such as synchrony, rhythm, tempo, and prosody that unite visual and acoustic stimulation from single events, as well as spatial colocation of objects and their sound sources, and changes in intensity across the senses (see Lewkowicz & Lickliter, 1994, for a review). These competencies provide the foundation for the perception of meaningful and relevant aspects of stimulation in social and nonsocial events, and they are described in more detail in subsequent sections of this chapter. In our view, detection of amodal information in early development provides a radical and efficacious solution to the so-called "binding" problem (see J. J. Gibson, 1979). That is, detection of amodal information in early development does away with the notion of perceivers having to coordinate and put together separate and distinct sources of information. By detecting higher order information common to more than one sense modality, even relatively naive perceivers can explore a unitary multimodal event in a coordinated manner. The task of development becomes to differentiate increasingly more specific information from the global array through detecting invariant patterns of both multimodal and unimodal stimulation (E. J. Gibson, 1969; J. J. Gibson, 1979; Stoffregen & Bardy, 2001). Results from contemporary research on infant perception indicate that the fact that our senses provide overlapping and redundant information for certain properties of objects and events poses no problem for perception. Rather, as we

156

Bahrick and Lickliter

argue later, this redundancy is a cornerstone of perceptual development, allowing optimal deployment of attention and the discovery of higher order perceptual structure. In this chapter we present a framework, the "intersensory redundancy hypothesis," that provides a way of conceptualizing the role of redundancy across the senses for promoting and organizing early perceptual and cognitive development. The intersensory redundancy hypothesis makes predictions about what aspects of stimulation will be attended to and processed more readily as a function of whether available stimulation for an object or event is multimodal or unimodal. Specifically, the intersensory redundancy hypothesis proposes that in early infancy information that is simultaneously available across two or more sensory modalities (amodal properties such as tempo, rhythm, and intensity) is highly salient and is therefore attended, learned, and remembered better than when the same information is presented in only one modality. Conversely, processing of some information is facilitated by unimodal stimulation. When modality-specific properties (such as pitch, color, pattern, or orientation) are presented in a single sensory modality, they are attended, processed, and remembered better than when the same properties are presented in the context of multimodal stimulation. We review a growing body of research that supports this framework and synthesize findings from human and animal as well as neural and behavioral studies that demonstrate the important role of intersensory redundancy in the development of perception, cognition, and communication.

II. Amodal Relations and the Multimodal Nature of Early Experience The young infant encounters a world of richly structured, changing, multimodal stimulation through his or her interactions with objects, events, people, places, and the self. This stimulation is experienced through a unified perceptual system that is sensitive to invariant aspects of stimulation across the senses (E. J. Gibson, 1969; E. J. Gibson & Pick, 2001). Several researchers have argued that amodal information can initially guide infant attention and perceptual learning in a manner that is economical, veridical, and adaptive (e.g., Bahrick, 1992, 1994, 2001; E. J. Gibson, 1969; E. J. Gibson & Pick, 2001; Walker-Andrews, 1997). As we have already described, amodal information is not specific to a particular sense modality but is redundant or invariant across two or more senses. Across the visual and tactile modalities, shape, texture and substance are amodal and specifiable in either modality. Any changes in intensity and temporal and spatial aspects of stimulation are amodal, including temporal synchrony, common rhythm, and tempo of action, which unite the movements and sounds of most audible and visible events. If the perceiver detects amodal information, then attention is, by definition, focused on a unitary, multimodal event. By detecting these higher order relations that encompass

Intersensory Redundancy

157

multiple forms of sensory stimulation, the problem of how infants come to integrate stimulation across the senses is effectively eliminated. For example, detection of amodal temporal synchrony, rhythm, and tempo may focus an infant's attention on the sights and sounds of a person speaking or on the visual and acoustic impacts of a bouncing ball. Consequently, the person or ball would be perceived as a unitary entity. Sensitivity to amodal relations can also act as a buffer against forming inappropriate associations across the senses, as infants would not readily relate the sounds of speech with other objects that do not share the temporal structure of the speech sounds. Researchers have demonstrated that infants perceive a variety of amodal relations across multiple senses (see Lewkowicz & Lickliter, 1994). For example, infants can perceive the relation between movements of a face and the sounds of a voice on the basis of temporal synchrony (Dodd, 1979), their common emotional expression (Walker, 1982; Walker-Andrews, 1997), and spectral information common to the shape of the mouth and a vowel sound (Kuhl and Meltzoff, 1984). Young infants can relate moving objects and their impact sounds on the basis of temporal synchrony (Bahrick, 1983, 1987, 1988; Lewkowicz, 1992, 1996), their common tempo of action (Bahrick, Flom, & Lickliter, 2002; Lewkowicz, 1985; Spelke, 1979), rhythm (Bahrick & Lickliter, 2000; Mendelson & Ferland, 1982), and collocation (Fenwick & Morrongiello, 1998; Morrongiello, Fenwick, & Nutley, 1998). They can also detect temporal information common to visual and acoustic stimulation specifying the substance and composition of moving objects (Bahrick, 1983, 1987, 1988, 1992) and the changing distance of moving objects (Pickens, 1994; Walker-Andrews & Lennon, 1985). In the area of visual-tactile perception, young infants can detect the common shape and substance of objects across vision and touch (E. J. Gibson & Walker, 1984; Hernandez-Reif & Bahrick, 2001; Meltzoff & Borton, 1979; Rose, 1994; Streri, 1993). Detection of these amodal relations guides selective attention and exploration of objects and events in the environment and promotes the perception of unitary multimodal events. Infants not only detect amodal relations, they also participate in temporally coordinated, co-regulated interactions with adult caretakers. Much early perceptual and cognitive development emerges in the context of close face-to-face interaction with caretakers. Adults regularly scaffold infants' attention and provide a rich interplay of concurrent visual, vocal, tactile, vestibular, and kinetic stimulation. The movements and vocal rhythms of infants have also been shown to contain a burstpause, turn-taking pattern that is intercoordinated with the temporal characteristics of adult communication (Jaffe, Beebe, Feldstein, Crown, & Jasnow, 2001; Sander, 1977; Stem, 1985; Trevarthan, 1993; Tronick, 1989). This sensitivity to temporal, spatial, and intensity information in human interaction promotes affective attunement between caregivers and young infants (Stem, 1985) and provides a vehicle for the development of intersubjectivity and shared perspective (Rochat & Striano, 1999; Trevarthan, 1993). Infants thus create, participate in, and respond to amodal

158

Bahrick and Lickliter

information in their interactions with adult caretakers in a mutually co-regulated manner. This lays the foundation for later milestones of social and communicative functioning. Exploration of the self also provides one of the earliest and most potent and reliable sources of multimodal stimulation, as proprioceptive feedback always accompanies self-generated visual, vocal, and tactile stimulation (see Rochat, 1995). Infants engage in active, self-directed, intermodal exploration of their bodies (e.g., Butterworth & Hopkins, 1988; Rochat, 1993; Van de Meer, Van der Weel, & Lee, 1995) and the temporal and spatial contingencies between their movements and those of the multimodal objects and events in their environment (e.g., Bahrick, 1995; Bahrick & Watson, 1985; Rochat & Morgan, 1995; Schmuckler, 1995). In sum, a large body of converging evidence highlights the fact that infants are adept at perceiving, generating, and responding to a host of amodal relations uniting stimulation across visual, auditory, vestibular, tactile, and proprioceptive stimulation in the first months of life.

III. Unimodal-Multimodal Dichotomy in Developmental Research Despite the fact that the infant's world is inherently multimodal, and that virtually all perception, learning, memory, and social and emotional development emerges in this multimodal context, the majority of research in developmental psychology has focused on development in only a single sense modality at a time (see Kuhn & Siegler, 1998, for an overview of this type of research). This state of affairs likely resulted from the historical concern with sensory integration, the apparent intractability of the binding problem, and a lack of appreciation of the complex interdependencies among the senses. Scientists have traditionally specialized in unimodal areas such as vision, audition, or olfaction research, with subspecializations within each sensory area. As a result of this "unimodal" approach, the development of a specific skill or competence has been typically investigated detached from the rich multimodal context in which it occurs. For example, theories of speech and language development have typically focused on the unimodal speech stream, detached from the moving face and person that produce the speech. Research on infant memory and categorization has often focused on responsiveness to a unimodal visual display. Theories of face perception have been primarily based on studies of a unimodal, visual facial display devoid of movement and speech. Studies of the development of joint attention typically present the visual behavior detached from the auditory and tactile stimulation that typically co-occur (for further discussion, see Lickliter & Bahrick, 2001, and Walker-Andrews & Bahrick, 2001). The growth of the field of developmental psychology in general and the study of perceptual development in particular have tended to reflect this compartmentalization. Although research on the development of intersensory perception has

Intersensory Redundancy

159

grown more prominent, its theories and findings are for the most part segregated from research on the same questions explored with unimodal stimulation. Largely due to historical tradition, it has been placed alongside the other "senses" with "intersensory perception" constituting a separate area of inquiry (see Kellman & Arterberry, 1998, for an example). Thus, investigations of the development of a particular competence (e.g., aspects of memory, categorization, attention, speech perception) are likely to be conducted in separate studies of unimodal versus multimodal perception and to be undertaken by separate investigators. Consequently, research findings from the two areas are not well integrated and studies of unimodal and multimodal perception are difficult to compare, as they typically employ different methods and measures. Furthermore, few investigators actually compare responsiveness in one sense modality to responsiveness in two or more sensory modalities concurrently. Thus it is not known how perception of unimodal events such as the speech stream or moving faces generalizes to the multimodal world where speech occurs in the context of moving faces and vice versa. Importantly, research findings are consistent with the view that the senses interact in complex ways (e. g., King & Carlile, 1993; Lickliter, 2000; Lickliter & Hellewell, 1992; Stein & Meredith, 1993) and that different results are obtained when perception and cognition are investigated in the context of multimodal as compared with unimodal stimulation (Lickliter & Bahrick, 2001; Walker-Andrews & Bahrick, 2001). Research from the areas of unimodal and multimodal perception needs to be integrated if we are to develop a unified, ecologically relevant theory of the nature of perceptual development. Studying the single sensory system alone can, in many cases, result in a distortion of normally occurring patterns of sensory experience and consequently result in findings of limited generalizability. More studies are needed with both humans and animals that examine the development of skills and capabilities in a multimodal context and directly compare responsiveness to unimodal versus multimodal events in single research designs. Furthermore, just as unimodal and multimodal research is not well integrated, neither is behavioral research well integrated with research in the neurosciences. The proliferation of research on multisensory functioning in the neurosciences (see Calvert, Spence, & Stein, 2002) makes findings in this area of central importance to a biologically plausible theory of the development of intersensory perception.

IV. Neural and Behavioral Evidence for Intersensory Interactions Because the traditional view is that the different sensory modalities utilize separate and distinct neural pathways, neural "integration" of separate streams of sensory information has typically been viewed as necessary for adaptive perception and cognition. However, it has become increasingly clear that the separate

160

Bahrick and Lickliter

senses are not so separate at the level of the nervous system (Knudsen & Brainard, 1995; Meredith & Stein, 1986; Stein, 1998; Stein & Meredith, 1993). This appreciation of the multimodal nature of the brain calls into question the long-standing view that higher order perceptual processing and cognition is needed to achieve successful binding or integration across the sensory modalities. Evidence obtained from neuroimaging studies reveals that many areas of the cortex and subcortex previously thought to receive input from only one sensory modality respond reliably to multisensory stimulation (see Calvert, 2001 a, for a review). Furthermore, a number of empirical investigations have shown that both young and mature animals have well-organized inputs from different sensory modalities converging on the same target structure in the brain (e.g., Frost, 1984; Innocenti & Clarke, 1984). This body of evidence from the neurosciences has led investigators to a growing appreciation of the brain's sensitivity to multimodal information (see Calvert et al., 2002; Stein & Meredith, 1993), but such an appreciation is not yet widely held by developmental psychologists and has yet to be incorporated into our thinking about the nature and direction of early perceptual organization. Here we briefly review some of the available neural evidence informing the study of perceptual and cognitive development. The most investigated site of multimodal convergence is the superior colliculus, a midbrain structure known to play a fundamental role in attentive and orientation behaviors (reviewed in Stein & Meredith, 1993). Multisensory neurons have been found in the superior colliculus of cats (Meredith & Stein, 1983), monkeys (Jay & Sparks, 1984), and several species of rodents (Wallace, Wilkenson, & Stein, 1996). The multisensory neurons in the superior colliculus respond to input from several sensory modalities and provide a neural substrate for enhancing responsiveness to stimuli that are spatially and temporally aligned. For example, in guinea pigs, visual experience is required for the normal elaboration of the sensory map of auditory space in the superior colliculus (Withington-Wray, B inns, & Keating, 1990). Guinea pigs reared in darkness fail to develop an auditory map, supporting the view that normal development of a map of auditory space requires the coincident activation of neural activity deriving from the convergence of both auditory and visual input arising from common stimuli. The activity-based alignment of different sensory maps in the brain and the responsiveness of these areas to intersensory convergence is likely a critical feature of multisensory perception (Stein & Meredith, 1993). Sites of multisensory convergence have also been reported at the cortical level of the brain in cats (Wallace, Stein, & Meredith, 1992), monkeys (Mistlin & Perrett, 1990), rats (Barth, Goldberg, Brett, & Di, 1995), and humans (Calvert, 2001 a; Giard & Peronnet, 1999), suggesting that the mammalian brain is inherently multimodal in structure and function. Of particular interest to theories of intersensory functioning is the finding from a number of neuroanatomical and neurophysiological studies indicating that the temporal and spatial pairing of stimuli from different sensory modalities can elicit a

Intersensory Redundancy

161

neural response that is greater than the sum of the neural responses to the unimodal components of stimulation considered separately (the so-called "multiplicative or superadditive effect" reviewed in Stein & Meredith, 1993; Stein, Meredith, & Wallace, 1994). In other words, the activity of a neuron exposed to multisensory stimulation (i.e., simultaneous auditory and visual stimulation) differs significantly from the activity of the same cell when exposed to stimulation in any single modality (Meredith & Stein, 1986). Spatially coordinated and synchronous multimodal stimulus combinations have been shown to produce significant increases over unimodal responses in several extracellular measures of neural activity, including response reliability, number of impulses evoked, and peak impulse frequency. This superadditive effect of bimodal stimulation, in which the magnitude of neural effects resulting from bimodal stimulation consistently exceeds the level predicted by adding together responsiveness to each single-modality stimulus alone (i.e., neural enhancement) has also been reported in behavioral investigations. For example, Stein, Meredith, Honeycutt, and Wade (1989) demonstrated that the effectiveness of a visual stimulus in eliciting attentive and orientation behaviors in cats is dramatically affected by the presence of a temporally congruent and spatially collocated stimulus in the auditory modality. These findings provide further support for the notion of differential responsiveness to unimodal versus multimodal stimulation and indicate that spatially and temporally coordinated multimodal stimulation is highly salient at the level of neural responsiveness. There is also compelling neurophysiological and behavioral evidence of strong intermodal linkages in newborns, young infants, and adults from a variety of species, including humans (e.g., Carlsen & Lickliter, 1999; King & Carlile, 1993; King & Palmer, 1985; Knudsen & Brainard, 1991, 1995; Lewkowicz & Turkewitz, 1981; Lickliter & Banker, 1994; Massaro, 1998; Mellon, Kraemer, & Spear, 1991; Withington-Wray et al., 1990). Experimental manipulations with animal subjects that augment or attenuate sensory stimulation in one modality consistently lead to significant effects on the development of perception in other sensory modalities and on the development of intersensory functioning during both the prenatal and postnatal periods (Lickliter & Banker, 1994; Lickliter & Hellewell, 1992; Radell & Gottlieb, 1992; Sleigh & Lickliter, 1997). For example, Lickliter and Lewkowicz (1995) showed the importance of prenatal tactile and vestibular stimulation to the successful emergence of species-typical auditory and visual responsiveness in bobwhite quail chicks. Hein and colleagues (Hein, 1980; Hein & Diamond, 1983; Held & Hein, 1963) demonstrated that visual stimulation provided by young kittens' own locomotion was necessary for the development of eye-paw coordination and visually guided behavior. Eye-paw coordination was found to develop normally in kittens allowed to simultaneously walk and look at objects and events, but did not develop normally when kittens could only look at things while being moved passively. Kittens denied visual feedback from locomotion also consistently failed to develop visually guided reaching. Of course, under normal developmental

162

Bahrick and Lickliter

conditions, convergence between multimodal visual and proprioceptive stimulation is a regular aspect of postnatal experience and such convergence appears to be an experiential requirement of normal perceptual development. Studies of neural and behavioral development have thus revealed strong intermodal interactions in newborns and infants, with stimulation in one sensory modality influencing and even calibrating responsiveness in other modalities in an ongoing manner. Research with human adults has also provided compelling support for the salience of intersensory congruence. For example, Sathian (2000) found that the visual cortex can be involved in tactile perception in adult humans. In this study, PET scans of blindfolded subjects performing a tactile discrimination task (determining the orientation of ridges on a surface) revealed increased activity in the visual cortex. Furthermore, when function of the visual cortex was interfered with by means of transcranial magnetic stimulation, tactile perception was significantly impaired. In a similar vein, Calvert (2001b) scanned the brains of adults when they smelled odors, looked at colors, or did both simultaneously. Olfactory areas of the brain became particularly active when the colors and scents were congruent (i.e., a red strawberry) as compared with incongruent (i.e., a blue strawberry). Calvert concluded that multisensory congruence enhances neural responsiveness, whereas incongruence serves to suppress neural responsiveness (for a similar view, see Stein, 1998). The potent intersensory interactions present in early development appear to continue to affect perceptual responsiveness in adulthood. Several perceptual illusions also underscore the existence of intersensory convergence and its role in guiding attention and perceptual discrimination. The wellknown McGurk effect (McGurk & MacDonald, 1976), an auditory-visual illusion, illustrates how perceivers merge information for speech across the senses. When we view the face of a person speaking a speech sound such as "ga," while hearing a different speech sound, "ba," the perception is of another sound, "da," a blend between the concurrently presented auditory and visual stimulation. Infants also show evidence of this effect in the first half-year following birth (Rosenblum, Schmuckler, & Johnson, 1997), indicating that visual input has significant auditory consequences, even during early development. Auditory input has also been shown to have dramatic consequences for visual perception. Scheir, Lewkowicz, and Shimojo (2002) demonstrated an audiovisual "bounce" illusion in young infants. Without sound, adults perceive two disks to be moving horizontally and passing through one another on a computer screen (streaming). When a discrete sound is added at the point of contact between the disks, adults report that the two disks appear to bounce against one another and change direction of motion. Young infants also appear to perceive the addition of sound to change the nature of the visual display from streaming to bouncing, indicating convergence across the modalities and demonstrating that sound can alter the perception of a visual event even during infancy. Shams (2000) reported a similar intersensory illusion in which sound can make adults see visual illusions.

Intersensory Redundancy

163

Adults hearing two beeps while seeing one flash of light reported that they saw two flashes. Furthermore, neural activity in the visual cortex (thought to be specific to visual processing) was found to be essentially equivalent whether the participant actually saw two flashes (with no beeps) or just one flash accompanied by two beeps, suggesting that neural and behavioral consequences are operating in parallel. These examples drawn from neural and behavioral studies indicate that intersensory convergence is integral to perceptual functioning. Inputs from our separate senses interact and influence one another more than we have acknowledged or appreciated for much of the 20th century. Furthermore, this influence results in the perception of emergent properties of stimulation qualitatively different from the perception of input from the separate sensory modalities. In our view, theories of behavioral development must be informed by knowledge of neural development and responsiveness, and vice versa. Simply put, our psychological theories of intersensory functioning must be biologically plausible. That is, they must be consistent with available findings on intersensory convergence from the neural level of analysis, the physiological level of analysis, and with the complex intersensory interactions known to exist in the very early stages of perceptual processing.

V. Intersensory Redundancy Hypothesis: Toward an Integrated Theory of Perceptual Development Intersensory redundancy refers to a particular type of multimodal stimulation in which the same information is presented simultaneously and in a spatially coordinated manner to two or more sensory modalities. For the auditory-visual domain, it also entails the temporally synchronous alignment of the information in each modality. Only amodal properties (e.g., tempo, rhythm, intensity) can be specified redundantly because, by definition, amodal information is information that can be conveyed by more than one sense modality. Thus, the sights and sounds of hands clapping provide intersensory redundancy in that they are synchronous, collocated, and convey the same rhythm, tempo, and intensity patterns across vision and audition. As depicted in Figure 1, intersensory redundancy is best viewed as arising from an interaction between the organism and its environment. Redundancy is not a property of the structure of the organism (its nervous system and sensory systems), nor is it a property of the structure of objects and events in the environment. Rather, it results from an interaction between a structured organism and a structured environment. Redundancy is experienced when an active perceiver explores multimodal events with multiple coordinated senses. For example, one might explore a person speaking by looking and listening. In this case, the perceiver would experience redundantly specified information for the tempo, rhythm, and intensity patterns of auditory-visual speech. However, when the perceiver looks away

164

Bahrick and Lickliter

Fig. 1. Intersensory redundancy results from the dynamic relationship between a structured organism and a structured environment. Redundancy arises from exploration by a nervous system specialized forpicking up different types of energy and their overlap andfrom unitary objects and events thatprovide a structured array ofmultimodal stimulation.

from the speaking person, or if the speaker leaves the room while talking, the perceiver no longer experiences redundantly specified information for tempo, rhythm, and intensity patterns. Rather, he or she perceives unimodal information for these speech properties. Thus, the perception of redundancy is dynamic in the sense that it can change from moment to moment as the relation between the nature of the organism's exploratory behavior changes and as the nature of the objects and events in the environment change. It is the convergence of information in two senses that makes amodal properties salient. Redundancy thus relies on both a nervous system specialized for different types of energy and the ability of the senses to provide overlapping information about objects and events that are unitary in the world. Redundancy is only apparent across different forms of stimulation and in this sense requires specific forms of energy from the different sensory modalities. As we describe in more detail in the sections that follow, intersensory redundancy is highly salient and can direct selective attention and facilitate perceptual learning in early development. In our view, intersensory redundancy is a particularly important and salient form of stimulation available to infants and plays a foundational role in early perceptual and cognitive development. Research with both animal and human infants indicates that different properties of stimuli are highlighted and attended to when redundant multimodal stimulation is made available to young organisms as compared with unimodal stimulation from the same events (see Bahrick, 2002; Bahrick & Lickliter, 2000; Lickliter & Bahrick, 2002). That is, young infants are especially adept at detecting amodal, redundant stimulation and detection of this information can organize early attention and provide a foundation for and guide and constrain perceptual development. We proposed an "intersensory redundancy hypothesis" to account for how this might be the case (Bahrick & Lickliter, 2000). The intersensory redundancy hypothesis describes how infants' attention will be allocated to different stimulus properties of objects and events as a function of the type of exploration (unimodal vs multimodal) afforded by the event. It also proposes consequences of this pattern of exploration for perception, learning, and memory.

IntersensoryRedundancy

165

One tenet of the intersensory redundancy hypothesis holds that in early development, information presented redundantly and in temporal synchrony to two or more sense modalities recruits infant attention and facilitates perceptual differentiation of the redundant information more effectively than does the same information presented to one sense modality at a time. From this view, detection of higher order amodal relations in multimodal stimulation from an object or event causes amodal stimulus properties to become "foreground" and other properties of the object or event to become "background." Thus, intersensory redundancy affects attentional allocation and this in turn can promote earlier processing of redundantly specified properties of stimulation (temporal and spatial aspects) over other stimulus properties. Because intersensory redundancy is readily available in the multimodal stimulation provided by our environment and our interaction with it, perception, learning, and memory of amodal properties likely precedes that for other stimulus properties. This "amodal processing precedence" in turn, can have long-range effects on perception, cognition, and social and emotional development. Because all our fundamental human capabilities emerge and develop in a multimodal context, rich with intersensory redundancy, these initial conditions can continue to influence the trajectory and organization of development. And because sensitivity to intersensory redundancy is present early in development and redundancy is so pervasive, it can create a cascading effect across development such that its consequences manifest in an ever-widening trajectory in a variety of domains (see Michel & Moore, 1995, and Moore, 1990, for examples of cascading effects in development). However, not all exploration of the objects and events in our environment makes multimodal stimulation available. In fact, intersensory redundancy is often not available for a particular event, either because the perceiver is not actively exploring that particular event with multiple senses, or because the event that is the focus of attention is not providing redundant simulation at that moment in time to the senses through which the perceiver is exploring (e.g., the perceiver is just looking at, but not touching, a stationary or silently moving object). In this case, amodal information for the event may be unavailable or available only in a single sense modality. For example, one might experience the rhythmic sounds of speech from a neighboring room, or the sight of a light blinking at a regular rate on a nearby appliance. The amodal properties of rhythm and rate would then be specified unimodally rather than redundantly. In this case, the amodal information of rhythm and rate would not be salient and there should be no amodal processing precedence. When only unimodal stimulation is provided for a particular property, there is no competition from intersensory redundancy. Therefore, attention is more likely to be recruited toward modality-specific properties of stimulation. Modality-specific properties are qualities specific to a particular sense modality. For example, color and pattern can only be perceived visually and pitch and timbre can only be perceived acoustically. According to a second tenet of the intersensory redundancy hypothesis, when only unimodal stimulation is available for a particular property,

Bahrick and Lickliter

166

attention to modality-specific properties should be facilitated relative to other stimulus properties. Thus, we hypothesize that unimodal exploration enhances perceptual differentiation of modality-specific and nonredundantly presented properties, as compared with the same properties presented in the context of multimodal, redundant stimulation (Bahrick & Lickliter, 2000). Optimal differentiation of visible qualities of an object or event should occur when there is no competition from auditory stimulation, which creates intersensory redundancy and recruits attention away from the visible qualities. For example, in early development, differentiation of the appearance of a person's face would be optimal when the individual was silent, differentiation of the nature of their particular voice would be optimal when their face was not visible, and differentiation of the prosody, rhythm, tempo, and timing of language would be optimal when viewing a speaking person. This observation is consistent with observations of the early emergence of sensitivity to the prosody in speech (Cooper & Aslin, 1989; Fernald, 1984). Thus, according to the intersensory redundancy hypothesis, the nature of exploration (unimodal vs multimodal) interacts with the type of property explored (amodal vs modality-specific) to determine the attentional salience and processing priority given to various properties of stimulation. As can be seen in Figure 2, bimodal exploration of amodal properties and unimodal exploration of modalityspecific properties receive priority in processing. In contrast, processing is relatively disadvantaged for bimodal exploration of modality-specific properties (e.g., listening to the pitch and timbre of a voice while also seeing the speaking face) and for unimodal exploration of amodal properties (e.g., seeing a rhythm displayed visually without sound, or hearing a rhythm in sound without visual

Stimulus Property Amodal

"~ m r

Multimodal (auditory-visual)

Modality..Specific

+