Building Object Categories in Developmental Time, 1st
Edition
Visit the link below to download the full version of this book:
https://medipdf.com/product/building-object-categories-in-developmental-time-1st
-edition/
Click Download Now
BUILDING OBJECT
CATEGORIES IN
DEVELOPMENTAL TIME
Edited by
Lisa Gershkoff-Stowe
Indiana University
David H. Rakison
Carnegie Mellon University
32nd Carnegie Mellon Symposium
on Cognition
LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS
2005 Mahwah, New Jersey London
Copyright © 2005 by Lawrence Erlbaum Associates, Inc.
All rights reserved. No part of this book may be reproduced in
any form, by photostat, microform, retrieval system, or any other
means, without the prior written permission of the publisher.
Lawrence Erlbaum Associates, Inc., Publishers
10 Industrial Avenue
Mahwah, New Jersey 07430
www.erlbaum.com
Cover art bv Rachel Wu
Cover design by Kathryn Houghtaling Lacey
Library of Congress Cataloging-in-Publication Data
Carnegie Symposium on Cognition (32nd : 2002 : Carnegie-Mellon University)
Building object categories in developmental time / edited by Lisa Gershkoff-Stowe
and David H. Rakison.
p. cm.
"32nd Carnegie Mellon symposium series on cognition."
Includes bibliographical references and index.
ISBN 0-8058-4490-2
ISBN 0-8058-4491-0 (pbk.)
1. Categorization (Psychology) in children—Congresses. I. Gershkoff-Stowe, Lisa.
II. Rakison, David H., 1969- III. Title.
BF723.C27C37 2005
155.4'13—dc22 2004052068
CIP
Books published by Lawrence Erlbaum Associates are printed on acid-free paper,
and their bindings are chosen for strength and durability.
Printed in the United States of America
10 9 8 7 6 5 4 3 2 1
To my children, Ben and Sarah.
—L. G.-S.
To those closest to me,
Caro, Michael, Jon, and Nic.
—D. H. R.
This page intentionally left blank
Contents
Contributors ix
Preface xi
1 The Segregation of Face and Object Processing 1
in Development: A Model System of Categorization?
Charles A. Nelson and Kelly Snyder
2 Building Knowledge From Perception in Infancy 33
Scott P. Johnson
3 Categories, Kinds, and Object Individuation in Infancy 63
Fei Xu
4 Bubbles: A User's Guide 91
Frederic Gosselin and Philippe G. Schyns
5 Young Infants' Categorization of Humans Versus
Nonhuman Animals: Roles for Knowledge Access
and Perceptual Process 107
Paul C. Quinn
vii
viii CONTENTS
6 The Perceptual to Conceptual Shift in Infancy
and Early Childhood: A Surface or Deep Distinction? 131
David H. Rakison
7 Emerging Ideas About Categories 159
Linda B. Smith
8 Imposing Equivalence on Things in the World:
A Dynamic Systems Perspective 175
Lisa Gershkoff-Stowe
9 Why Can't You "Open" a Nut or "Break"
a Cooked Noodle? Learning Covert Object
Categories in Action Word Meanings 209
Melissa Bowerman
10 The Development of Relational Category Knowledge 245
Dedre Centner
11 Demystifying Theory-Based Categorization 277
Woo-kyoung Ahn and Christian C. Luhmann
12 Can Our Experiments Illuminate Reality? 301
Brian MacWhinney
13 Knowledge, Categorization, and the Bliss of Ignorance 309
Frank C. Keil
14 A Parallel Distributed Processing Approach
to Semantic Cognition: Applications to
Conceptual Development 335
Timothy T. Rogers and James L. McClelland
15 Abstraction as Dynamic Interpretation in
Perceptual Symbol Systems 389
Lawrence W. Barsalou
16 Models of Categorization: What Are the Limits? 433
Robert Siegler
Author Index 441
Subject Index 455
Contributors
Woo-kyoung Ahn, Department of Psychology, Yale University, 2 Hillhouse Avenue,
New Haven, CT 06511-6814
Lawrence W. Barsalou, Department of Psychology, Emory University, 532 N. Kilgo
Circle, Atlanta, GA 30322
Melissa Bowerman, Max Planck Institute for Psycholinguistics, Postbus 310, 6500
AH, Nijmegen, The Netherlands
Dedre Gentner, Department of Psychology, Northwestern University, 2029 Sheri-
dan Road, Evanston, IL 60208
Lisa Gershkoff-Stowe, Department of Speech and Hearing Sciences and Cognitive
Science Program, Indiana University, 200 South Jordan Avenue, Bloomington,
IN 47405-7002
Frederic Gosselin, Department of Psychology, University of Montreal, C.P. 6128,
succ. Centre-ville, H3C 3J7, Montreal, QC Canada
Scott P. Johnson, Department of Psychology, New York University, 6 Washington
Place, New York, NY 10003
Frank C. Keil, Department of Psychology, Yale University, 2 Hillhouse Avenue, New
Haven, CT 06511-6814
Christian C. Luhmann, Department of Psychology, Vanderbilt University, 301 Wil-
son Hall, 111 21st Avenue South, Nashville, TN 37203
Brian MacWhinney, Department of Psychology, Carnegie Mellon University, 5000
Forbes Avenue, Pittsburgh, PA 15213
ix
X CONTRIBUTORS
James L. McClelland, Center for the Neural Basis of Cognition and Department of
Psychology, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA
15213
Charles A. Nelson, Institute of Child Development, 51 East River Road, Minneapo-
lis, MN 55455-0345
Paul C. Quinn, Department of Psychology, University of Delaware, Wolf Hall, New-
ark, DE 19716-2577
David H. Rakison, Department of Psychology, Carnegie Mellon University, 5000
Forbes Avenue, Pittsburgh, PA 15213
Timothy T. Rogers, Department of Psychology, University of Wisconsin, 524 WJ
Brogden Hall, Madison, WI 53706
Philippe G. Schyns, Department of Psychology, University of Glasgow, Room 416,
58 Hillhead Street, Glasgow, Scotland, UK G12 8QB
Robert Siegler, Department of Psychology, Carnegie Mellon University, 5000
Forbes Avenue, Pittsburgh, PA 15213
Linda B. Smith, Department of Psychology and Cognitive Science Program, Indiana
University, 1101 E. 10th Street, Bloomington, IN 47405
Kelly Snyder, Department of Psychology, University of Denver, Frontier Hall, 2155
S. Race Street, Denver, CO 80208
Fei Xu, Department of Psychology, University of British Columbia, 2136 West Mall,
Vancouver, BC Canada V6T 1ZA
Preface
In 1987, the 21st Carnegie Symposium on Cognition was organized by col-
leagues David Klahr and Kenneth Kotovsky. Its purpose was to examine and
celebrate the far-reaching impact of H. A. Simon's work on the field of cog-
nitive psychology. As they noted in the accompanying volume that followed
from the conference, Herb Simon played an influential role in the scien-
tific agenda of nearly all of the symposia since its inception in 1965 (Klahr
& Kotovsky, 1989). In the years since that observation was made, Herb's au-
thority and vision have continued to inspire the many outstanding scientists
who regularly meet in Pittsburgh to present their work, reflect on issues of
common concern, and identify remaining challenges in the study of hu-
man cognition. Herb Simon died before the start of the 32nd Carnegie
Symposium held in June 2001, but his impact continues to sharpen and
deepen our understanding of human behavior, as this volume attests.
This book concerns the way in which developmental processes work. It
presents the current state of research in the area of early category develop-
ment. The contributors to the volume attempt to answer the question of
how children build object categories, beginning with the basic architecture
of the brain and with the constraints or biases that provide the foundation
of early perceptual experience. These advances are further considered in
view of subsequent growth of language and other higher-order cognitive
processes that guide the formation of coherent systems of knowledge.
The importance of categorization and concept formation to cognitive de-
velopment, and to cognition in adults, cannot be underestimated. Categori-
zation is the primary means of coding experience, which in turn reduces de-
xi
xii PREFACE
mands on inherently limited memory storage and perceptual and reasoning
processes. Categorization is also inextricably linked to language processes.
Words have the power to expand and sharpen category boundaries, to en-
courage category inductions beyond surface similarity, and to shift attention
to less salient but potentially relevant properties of objects. The early classifi-
cation abilities of infants and young children, and the mechanisms underly-
ing those abilities, are of intense interest not only to developmental scientists,
but also to investigators in fields as wide-ranging as neuroscience, visual per-
ception, computational modeling, and linguistics.
As the title of this volume suggests, the ultimate aim of this collection is
to account for the progressive elaboration of categorization in developmen-
tal time—age-related changes that appear globally stable over the time span
of months and years. To investigate development over the long term, how-
ever, many of the contributors have focused their lens on the days, minutes,
and even seconds of time over which change also occurs. A basic message of
the 32nd Annual Symposium is that both local and global levels of analysis
are essential to understanding the processes by which children's knowledge
of categories develops.
There is no single point of view that inhabits the chapters of this vol-
ume. Rather, the chapters reflect the diverse backgrounds and rich theo-
retical perspectives of the particular set of invited speakers. Topics range
from the neural substrates for face and object recognition to basic percep-
tual skills to the role of language and conceptual development in the
transformation of emerging categories. The methods for exploring these
topics are varied as well, and include both behavioral and neural ap-
proaches and computational models. Despite these differences, the
reader will find much common ground across the chapters, particularly in
their recurring reference to landmark studies in the developmental and
adult categorization literature.
The book is divided into three Parts. Part I comprises six chapters and a
commentary, which center on the fundamental processes of perception,
memory, and attention. In the opening chapter, Charles Nelson and Kelly
Snyder explore the neural bases for the development of categorization of
face and nonface objects. They develop a cognitive neuroscience framework
for categorization based on the human and nonhuman animal literature.
They then apply this framework, in conjunction with Nelson's well-estab-
lished model of face recognition, to the development of object processing
more generally. Their aim is to show that face recognition offers a model
system for other aspects of development.
In the next chapter, Johnson examines a critical precursor to object cate-
gorization, namely the ability to represent objects. Johnson tackles one of
the thorniest debates in the developmental literature—the relative role of
innate versus learned knowledge—by focusing on evidence from studies on
PREFACE xiii
object unity and occlusion. He proposes the sensitivity-first hypothesis,
which stipulates that object knowledge arises from ontogenetic processes
over developmental time, processes that are rooted in relatively low-level,
perceptual-cognitive capacities. In support of this hypothesis, Johnson pre-
sents compelling behavioral and imaging data from studies with infants,
adults, and nonhuman animals.
In the third chapter, Xu investigates the notion of individuation as an-
other important precursor to object categorization. Individuation, as de-
fined by Xu, is the process of representing numerically distinct individuals
that can be tracked through time and space. She suggests a theory of kind
representations that accounts for both categorization and individuation. Fur-
thermore, she proposes that not all categories are created equal; some cate-
gories are kinds, and some kinds are "essence sortals." To support her view,
Xu presents evidence of both an early developing object-based system of in-
dividuation and a relatively late developing kind-based system, the latter
driven, perhaps, by advances in language learning.
Next, Gosselin and Schyns provide a user's guide to Bubbles, a technique
that reveals the information that drives a measurable response. Bubbles is a
technique that can resolve the credit assignment problem of attributing the
determinants of a response to the parametric subspace of a carefully speci-
fied information search space. Gosselin and Schyns illustrate the technique
with a complete example, the Face Inversion Effect, and discuss the six ba-
sic decisions that must be made to set up a Bubbles experiment (i.e., the
stimulus set, the generation space, the "bubbles," the task, the group of ob-
servers, and the response). They describe methods to analyze the data and
provide practical advice for the researcher intending to use the technique.
In his chapter, Quinn describes recent research with the familiarization/
novelty preference procedure that investigates young infants' categoriza-
tion of humans and nonhuman animals. At issue is whether category repre-
sentations are formed online or constructed prior to the experiment. The
evidence Quinn presents reveal that young infants use an online summary
representation to categorize nonhuman animals at a basic level of exclu-
sivity. In contrast, they use an exemplar representation, formed through
previous experience, to categorize humans at a global level of exclusivity.
In the former instance, infants appear to rely on part information to cate-
gorize; in the latter case, they rely on holistic—configural information.
Quinn uses these data to develop further his perceptual learning model of
early categorization.
Finally, Rakison examines the phenomenon known as the perceptual to
conceptual shift. This phrase refers to the findings that younger children and
infants tend to rely on perceptual properties as the basis for categorization,
whereas older children weigh more heavily the nonobvious or conceptual
properties of objects in category membership decisions. Based on a review of
xiv PREFACE
the literature, Rakison concludes that there is little evidence to support the
notion that representational development undergoes some kind of
transitional shift either in infancy or the preschool years. Rather, he pre-
sents a theoretical model and evidence that suggests that perceptual informa-
tion provided by the different sensory input systems—including language—is
continually added to the earliest representations through associative learn-
ing, and as sensitivity to different sorts of information emerges, so this grad-
ual process gives the appearance of qualitative, representational change.
Smith's commentary on these chapters summarizes the field of early cat-
egory development and points out that the field is very much in transition.
She addresses the major themes highlighted by the chapters by focusing on
a number of key questions: How are categories represented? Are general or
special processes involved? Are categories based on percepts or concepts?
Smith suggests that these issues are irresolvable but points out that the field
is moving in new directions, toward fundamentally different ideas about
what categories and category development might be. She describes these
new directions in relation to several of the chapters presented throughout
the volume.
Part II of the book comprises four chapters plus commentary, concern-
ing how linguistic and conceptual advances transform children's develop-
ing category knowledge. Gershkoff-Stowe applies the principles of dynamic
systems theory to the study of object category development. At the core of
this approach is the idea that change is a function of converging forces that
operate at multiple levels of time. Gershkoff-Stowe offers a mechanistic ac-
count of how large-scale developments in children's category behavior are
specified in the real-time processes of seeing, manipulating, and naming
objects. Her findings support the view that children are inherently flexible
and adaptive in their ability to classify objects in response to variations in
context.
Bowerman's chapter deals with covert or action-related object catego-
ries, for example, verbs like open, cut, or break that apply to events involving
objects of a certain kind. She offers a cross-linguistic perspective, showing
how different languages partition events differently according to the prop-
erties of the objects involved. The developmental question Bowerman asks
is how children acquire covert object categories through the words they se-
lect to encode them. Her analysis of the errors young children produce re-
veal that extension patterns vary according to the language they learn and
that these variations are systematically related to the semantic structure of
the target language.
Gentner's chapter provides new insights into the acquisition of rela-
tional categories such as predator as well as the relational aspects of ordinary
object categories such as tiger. She elaborates her theory of structural align-
ment as a general learning mechanism available to young children—one
PREFACE XV
that promotes the learning of relational structure through the process of
comparison. According to Gentner, the role of language is to encourage
such comparison by guiding attention beyond surface-level commonalities
to deep conceptual similarity. Such relational commonalities are thought
to provide the "theory-like aspects of concepts and categories" that Ahn and
Luhmann investigate in the next chapter.
Ahn and Luhmann analyze the rich conceptual structures of natural
kinds and artifacts within which feature relations exist. They show that the
critical role of features is not their specific content, for example, the shape
of an object or its color, but rather the feature's position within a given con-
ceptual structure. In particular, adults and children show a preference to
attend to causes over effects. This processing bias is referred to as the causal
status hypothesis and presents a plausible alternative to theory-based ac-
counts of early category development, and in addition, to the enduring de-
bate about whether children prefer form or function as a basis for object
name generalization.
MacWhinney's commentary on these four chapters centers on the ques-
tion of how extensively do the methods and findings reported here reflect
real-world category learning in the everyday lives of children. Using exam-
ples from Web-related sources such as the Child Language Data Exchange
System (CHILDES) and TalkBank, he advocates the need for researchers to
explore, in addition, the rich social context in which concept and word
learning is clearly embedded.
Part III is organized around the formation of system-based category
knowledge, and in particular, how complex conceptual structure might be
the emergent result of the fundamental cognitive processes discussed in
previous chapters. In the first of the three chapters that comprise this last
section, Keil remarks on the failing of prevalent theories of cognitive devel-
opment in accounting for the nature and development of representational
states that enable categorization. He describes empirical work showing how
individuals track and use abstract causal patterns in category reasoning.
The data reveal that not only children, but also adults have a shallowness of
theory with which they think about the world. Keil interprets this "illusion
of explanatory depth" as support for a revised concepts-as-theories view in
which structural properties of causal systems are encoded and linked im-
plicitly to domain-specific categories of knowledge.
Rogers and McClelland describe a number of simulations with a simple
PDP model of semantic memory. They show that representations are organ-
ized with respect to certain especially useful or salient properties, regardless
of whether such properties can be directly observed. They claim that in
many cases this salience may be acquired through domain-general learning
mechanisms that are sensitive to the high-order coherent covariation of di-
rectly observed stimulus properties across a breadth of experience. When
xvi PREFACE
trained with backpropagation to complete queries about the properties of
different objects, the model's internal representations differentiate in a
coarse-to-fine manner. As a consequence, different sets of properties come
to be especially "salient" to the model at different points during develop-
ment. Such dynamics, they claim, provide a simple account of the acquisi-
tion of conceptual structure from perceptual experience.
Barsalou's chapter considers the idea of abstraction as a central con-
struct in cognitive science. Challenging established notions of how knowl-
edge is represented, Barsalou presents an alternative theory known as Dy-
namic Interpretation in Perceptual Symbol Systems (DIPSS). According to the
theory, conceptual structure is captured in loose collections of properties
and relations of specific simulators. These simulators are applied dynami-
cally across different contexts in an online and highly flexible manner.
Barsalou provides convincing evidence consistent with this general ac-
count, applying his theory to a variety of abstraction phenomenon in cate-
gorization and learning.
The final commentary is offered by Siegler. He observes that all three
chapters in this last section establish conclusions regarding categorization
that are applicable to other cognitive domains including language, mem-
ory, and scientific thinking. As suggested by Newell (1973), a major goal in
cognitive science is to develop broad, encompassing models that integrate
results across a wide range of experimental tasks and that specify a small but
precise set of mechanisms. Siegler commends these authors for making
contributions toward this goal. In addition, however, he suggests that fu-
ture work be directed toward constraining such models to identify the
boundary conditions to which they apply and to search for new mecha-
nisms beyond those currently specified.
ACKNOWLEDGMENTS
We are grateful to the many contributors who generously agreed to partici-
pate in the 32nd Annual Symposium on Cognition and whose work in-
creases our understanding of the complex processes involved in building
object categories in developmental time.
Thanks are due to the support staff and students in the department of
psychology at Carnegie Mellon University for giving generously of their
time: Kathy Majors, Bridget Boring, Margaret Kinsky, Tim Carroll, Annika
Fasnacht, and Erin Hahn. We would also like to extend a special note of ap-
preciation to Rochelle Sherman for her outstanding job of coordinating
the behind-the-scenes details, and to Audrey Russo for her invaluable assis-
tance in handling the many arrangements for the conference.
PREFACE xvii
Marlene Behrmann provided valuable advice on how to organize the sym-
posium. Jay McClelland and Heidi Feldman kindly opened up their beautiful
home for a delightful evening of food and conversation. David Klahr pro-
vided priceless counsel in his role as series editor, and the quality of this vol-
ume owes much to his input. Finally, we would like to thank Bill Webber, edi-
tor at Lawrence Erlbaum Associates, for his guidance and especially his
patience in bringing the book to publication. Sarah Wahlert, Senior Produc-
tion Editor at LEA, was also fundamental in transforming many typewritten
pages into the book that you now hold in your hands.
Funding for the 32nd Annual Symposium was provided by the psychol-
ogy department at Carnegie Mellon University, the National Institute of
Mental Health (1R13MH666370), and the National Science Foundation
(BCS-0132328); we gratefully acknowledge their support.
—Lisa Gershkqff-Stowe
David Rakison
REFERENCES
Klahr, D., & Kotovsky, K. (1989). Complex information processing: The impact of H. A. Simon.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Newell, A. (1973). You can't play 20 questions with nature and win: Projective comments on
the papers of this symposium. In W. G. Chase (Ed.), Visual information processing (pp.
283-308). New York: Academic.
