Handbook of Categorization in Cognitive Science · 2016-10-25 · HANDBOOK OF CATEGORIZATION IN...

HANDBOOK OF CATEGORIZATION IN COGNITIVE SCIENCE

i

This page intentionally left blank

ii

HANDBOOK OF CATEGORIZATION INCOGNITIVE SCIENCE

Edited by

HENRI COHENUniversite du Quebec a Montreal, Quebec, Canada

and

CLAIRE LEFEBVREUniversite du Quebec a Montreal, Quebec, Canada

Amsterdam ● Boston ● Heidelberg ● London ● New York ● Oxford ● ParisSan Diego ● San Francisco ● Singapore ● Sydney ● Tokyo

2005

iii

ELSEVIER B.V. ELSEVIER Inc. ELSEVIER Ltd ELSEVIER LtdSara Burgerhartstraat 25 525 B Street, Suite 1900 The Boulevard 84 Theobalds RoadP.O. Box 211 San Diego, CA 92101-4495 Langford Lane Kidlington London WC1X 8RR1000 AE Amsterdam USA Oxford OX5 1GB UKThe Netherlands UK

© 2005 Elsevier Ltd. All rights reserved.

This work is protected under copyright by Elsevier Lts, and the following terms and conditions apply to its use:

PhotocopyingSingle photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of thePublisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying foradvertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educationalinstitutions that wish to make photocopies for non-profit educational classroom use.

Permissions may be sought directly from Elsevier's Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44)1865 853333, e-mail: [email protected]. Requests may also be completed on-line via the Elsevier homepage(http: //www.elsevier.com / locate/permissions).

In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 RosewoodDrive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the CopyrightLicensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 207631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments.

Derivative WorksTables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resaleor distribution of such material. Permission of the Publisher is required for all other derivative works, including compilationsand translations.

Electronic Storage or UsagePermission of the Publisher is required to store or use electronically any material contained in this work, including any chap-ter or part of a chapter.

Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form orby any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of thePublisher.Address permissions requests to: Elsevier's Rights Department, at the fax and e-mail addresses noted above.

NoticeNo responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products lia-bility, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in thematerial herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drugdosages should be made.

First edition 2005

Library of Congress Cataloging in Publication DataA catalog record is available from the Library of Congress.

British Library Cataloguing in Publication DataA catalogue record is available from the British Library

ISBN: 0 08 044612-4

The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in The Netherlands.

Working together to grow libraries in developing countries

www.elsevier.com | www.bookaid.org | www.sabre.org

iv

v

PREFACE

The content of most of the chapters in this book was presented as part of the SummerInstitute on categorization that took place at the Université du Québec à Montréal(UQAM) for 10 days in June and July 2003. The objective of this Institute was toaddress the problem of categorization through the lens of all the disciplines that are atthe heart of the Cognitive Sciences: Cognitive Anthropology, Linguistics, Philosophy,Neuroscience, Psychology and Cognitive Computer Science. This book is a natural,concrete outcome of the Institute.

In planning this book, we wanted all aspects of categorization to be represented. Wetherefore filled the holes in the original program by soliciting contributions fromresearchers who had not been involved directly in the Summer Institute on categoriza-tion. In its present form, the book contains some 50 chapters. To our knowledge, it isthe first time in history that the problem of categorization has been considered from somany angles within a single book.

We would like to thank the members of the scientific committee of the SummerInstitute for their contributions, which have had a major impact on the contents of thisbook (Henrietta Cedergren for Linguistics, Pierre Poirier for Philosophy, Henri Cohenfor Neuroscience, Stevan Harnad for Psychology, Bernard Lefebvre for CognitiveComputer Science, and Claire Lefebvre as the Director of the Institute). We would liketo thank Robert Proulx, Dean of the Faculty of Human Sciences, for his supportthroughout. The financial contribution from the Faculty enabled us to prepare thisexceptionally large manuscript. Last but not least, we are grateful to Zofia Laubitz,Marlene Busko, and Sanja Obradović for making an important contribution by copy-editing and finalizing the manuscript.

Henri CohenClaire Lefebvre


vi

vii

TABLE OF CONTENTS

See Color Plate Section at the back of this book

Preface v

Bridging the Category DivideHENRI COHEN and CLAIRE LEFEBVRE

1. Introduction 22. Organization of the book 23. Major common themes 7

3.1. The notions of category and categorization 73.2. The nature of categories: Discrete, vague, or other 93.3. Are there modality effects on categories? 103.4. Are there universal categories? Are there innate categories? 11

4. Bridging the category divide 13References 15

PART 1 – CATEGORIZATION IN COGNITIVE SCIENCE

Chapter 1To Cognize is to Categorize: Cognition is CategorizationSTEVAN HARNADAbstract 201. Sensorimotor systems 212. Invariant sensorimotor features (“affordances”) 213. Categorization 214. Learning 225. Innate categories 236. Learned categories 237. Supervised learning 248. Instrumental (operant) learning 249. Color categories 25

10. Categorical perception 2511. Learning algorithms 2612. Unsupervised learning 2713. Supervised learning 2714. Vanishing intersections? 2815. Direct sensorimotor invariants 29

16. Abstraction and hearsay 3017. Abstraction and amnesia 3018. Invariance and recurrence 3119. Feature selection and weighting 3220. Discrimination versus categorization 3221. Recoding and feature selection 3322. Learned categorical perception and the Whorf hypothesis 3423. Uncertainty reduction 3524. Explicit learning 3625. Categorization is abstraction 3626. Sensorimotor grounding: direct and derivative 3627. The adaptive advantage of language: hearsay 3728. Absolute discriminables and affordances 3929. Cognitive science is not ontology 3930. Cognition is categorization 40Appendix A. There is nothing wrong with the “classical theory” of categorization 40Appendix B. Associationism begs the question of categorization 41References 42

Chapter 2A Modular Approach to Grammatical Categories Evidence fromLanguage Diversity and ContactPIETER MUYSKENAbstract 461. Introduction 472. Modularity and mismatch 483. Grammaticalization: The case of pe in Sranan (Suriname) 524. Mismatches in complexity of representations:

The case of ku in Cuzco Quechua (Peru) 545. Lexical nondistinctness 55

5.1. The case of timap in Palikur 555.2. ‘For’ prepositions to become complementizers 565.3. Adjectives versus adverbs 57

6. Lexical overspecification: Dutch gender and definiteness 577. The status of null elements: Subjects in Sranan and Papiamentu 588. Partially overlapping categories: predicate adjectives in Creoles 599. Differences in lexical richness 60

10. Evidence from language contact 6110.1. Otomanguean-Spanish language contact 6210.2. Borrowing of content words in Salishan languages 65

11. Concluding remarks 67Appendix 67References 68

viii Table of Contents

Chapter 3Philosophical Analysis as Cognitive Psychology: The Case of Empty ConceptsGEORGES REYAbstract 721. Introduction 732. Misadventures of the Classical View 733. Terminological issues 75

3.1. Existential vs. (purely) intentional usage 753.2. Concepts as between representations and referents 77

4. The inadequacies of Externalism 775. The need for internal roles 82

5.1. The Quinean challenge 825.2. The analytic data 835.3. Rivals to the Analytic Explanation 85

6. Conclusion 87References 88

Chapter 4Categories and Cognitive AnthropologyJAMES S. BOSTERAbstract 921. Introduction 932. Cognition and culture, universalism and relativism 933. Paradigms and taxonomies 944. Kinship terminologies 1015. Color classification 1056. Ethnobiology 1097. Towards a science of the stimulus 114References 116

Chapter 5Categorization in Neuroscience: Brain Response to Objects and EventsCATHERINE HANSON and STEPHEN JOSÉ HANSONAbstract 1191. Introduction 1202. Representing object categories in the brain 121

2.1. Category-specific representation 1222.2. Feature-specific representation 1232.3. Process-specific representation 1252.4. Summary 125

3. Acquiring category knowledge 1263.1. Summary 130

Table of Contents ix

4. Categorizing actions and events 1304.1. The nature of event knowledge 1304.2. When categorization of action fails 1324.3. The perception of events 1334.4. Summary 136


Chapter 6Categorization in Cognitive Computer ScienceJOHN F. SOWAAbstract 1411. Computation in cognitive science 1422. The great categorization debates 1443. From local features to global structures 1484. Categorization and reasoning 1535. Levels of cognition 158References 161

PART 2 – SEMANTIC CATEGORIES

Chapter 7Semantic CategorizationBRENDAN S. GILLONAbstract 1671. Introduction 1682. The notional approach to lexical categories 1693. The notional approach to lexical subcategories 1704. Structural approach to semantic categories 1725. Coordinators and subordinators 1756. English nouns 1817. Conclusion 184Acknowledgments 184References 184

Chapter 8Emotion Categories across LanguagesJAMES S. BOSTERAbstract 1881. Introduction 1892. Methods of assessing cultural emotion systems 191

2.1. The Method of Translation 1912.2. The Method of Mapping 198

x Table of Contents

3. Theories of emotion 2104. Cross-cultural scenarios as a tool to compare emotion categories 2135. Conclusion 218References 220

Chapter 9The World Color Survey DatabaseRICHARD S. COOK, PAUL KAY and TERRY REGIERAbstract 2241. Introduction 2252. The WCS: History and methodology 2253. Data processing and analysis 2284. Cleaning the data 2325. Original format of the data and creation of the WCS Online Data Archive 2326. Uses of the WCS archive 234

6.1. Universals of color naming 2346.2. Variation in color naming 237


Chapter 10Atoms, Categorization and Conceptual ChangePAUL THAGARD and ETHAN TOOMBSAbstract 2431. Introduction 2442. Theories of concepts 2453. The ancient concept of an atom 2464. Revival of the concept of the atom 2485. Modern development of the concept of an atom 2496. Theories and meaning 2527. Conclusion 253References 253

Chapter 11Relations between Language and Thought: Individuation and the Count/MassDistinctionANNA PAPAFRAGOU Abstract 2561. Introduction 2572. Strong discontinuity proposals 260

2.1. Quine 2602.2. Abstract individuation in language and thought 262

Table of Contents xi

3. Weak discontinuity proposals 2633.1. Crosslinguistic studies 2633.2. Language-on-language effects 266

4. Material and shape cues in labeling and categorization 2685. Conclusion 271References 271

Chapter 12Definitions in Categorization and Similarity JudgmentsSERGE LAROCHELLE, DENIS COUSINEAU and ANNIE ARCHAMBAULTAbstract 2781. Introduction 2792. Importance rating and property selection 284

2.1. Method 2842.2. Results 286

3. Categorization judgments 2883.1. Method 2883.2. Results 2903.3. Discussion 296

4. Similarity judgments 2974.1. Method 2974.2. Results 298

5. General discussion 300References 302

Chapter 13Why (Most) Concepts aren’t CategoriesRUTH GARRETT MILLIKANAbstract 3051. Introduction 3062. Species are not categories 3063. Three kinds of (Aristotelian) “substances” 307

3.1. Historical kinds 3073.2. Eternal kinds 3083.3. Individuals 309

4. Concepts of individuals 3105. Concepts of substances more generally 3116. Substances encountered through language 312References 315

xii Table of Contents

PART 3 – SYNTACTIC CATEGORIES

Chapter 14Lexical, Functional, Crossover, and Multifunctional CategoriesLISA DEMENA TRAVISAbstract 3201. Introduction 3212. Categories as feature bundles 321

2.1. The system 3212.2. Natural classes 3222.3. Unnatural classes 324

3. Categories and phrase structure 3253.1. Lexical and functional categories 3263.2. Articulation of functional categories 3273.3. Articulation below N and V 3293.4. Crossover and multifunctionality 330

4. Where do categorial distinctions reside? 3365. Conclusions 344References 345

Chapter 15Isolating-Monocategorial-Associational LanguageDAVID GILAbstract 3481. Introduction 3492. What IMA Language is Like 349

2.1. Isolating 3502.2. Monocategorial 3502.3. Associational 351

3. Where IMA Language Is Found 3543.1. Semiotics 3543.2. Phylogeny 3563.3. Ontogeny 358

4. Typology 3594.1. Riau Indonesian: overview 3604.2. Riau Indonesian: analysis 3644.3. Riau Indonesian: A Relative IMA Language 374

5. Cognition 375Acknowledgments 377References 377

Table of Contents xiii

Chapter 16Categories in Quebec Sign Language: Reflections on Categorization across ModalitiesDENIS BOUCHARD, COLETTE DUBUISSON and ANNE-MARIE PARISOTAbstract 3811. The categories of lexical items 3822. Traditional categorization applied to LSQ 384

2.1. Nouns and verbs 3872.2. Pronouns and definite determiners 387

3. Pronouns in oral languages and in sign languages 3883.1. The effects of perceptual substances on linguistic forms 3893.2. Explaining the different properties 390

4. Consequences for linguistic categorization and universals 396References 398

Chapter 17Syntactic Categories in Signed versus Spoken LanguagesDIANE LILLO-MARTINAbstract 4021. Introduction 4032. Lexical categories 4033. Grammatical structures 405

3.1. Subordination 4053.2. “Spatial syntax” 4063.3. Word order 413

4. Conclusion 417Acknowledgments 418Appendix. Notational conventions 418References 419

Chapter 18On Syntactic CategoriesMARK C. BAKER 423

PART 4 – ACQUISITION OF CATEGORIES

Chapter 19The Acquisition of Grammatical Categories: the State of the ArtMARIE LABELLEAbstract 4331. Grammatical categories 434

xiv Table of Contents

2. Two-word utterances and their analysis 4353. A semantic approach to grammatical categorization:

Semantic bootstrapping 4364. Distributional learning 439

4.1. Word order 4394.2. Inflection and inflectional class 4414.3. Function words 4454.4. Word classes 4474.5. Other cues to grammatical category learning 449

5. Models of distributional learning 4496. Constraining the search space 4507. Conclusion 451References 452

Chapter 20Semantic Categories in AcquisitionEVE VIVIENNE CLARKAbstract 4591. Introduction 4602. Space 4613. Shape 4654. Adding common ground 4665. Conceptual domains and lexical options 4676. Adding meaning in the course of

conversation 4727. Universals in mapping? 4738. Conclusion 476References 477

Chapter 21Early Syntactic Categories in Infants' LanguageRUSHEN SHIAbstract 4811. Introduction 4822. The acquisition of grammatical categories and the earliest binary

distinction of function words and content words 4823. Input speech and the categorization of function words

and content words 4864. Function words and language acquisition 4885. Conclusions 492Acknowledgment 492References 493

Table of Contents xv

Chapter 22Acquiring Auditory and Phonetic CategoriesMARTIJN GOUDBEEK, ROEL SMITS, ANNE CUTLER and DANIEL SWINGLEYAbstract 4971. Introduction 4982. Testing category learning 5003. Learning of nonspeech categories 5024. Learning of speech categories 5065. Conclusion 510References 511

Chapter 23Syntactic Categories in Second Language AcquisitionLYDIA WHITEAbstract 5151. Introduction 5162. Lexical and functional categories 5163. Lexical categories in L2 acquisition 5174. Functional categories in acquisition: Issues of evidence 5195. Functional categories in the L2 initial state and in L2 development 522

5.1. Morphology-before-syntax 5225.2. Syntax-before-morphology 523

6. Acquiring versus losing categories and features 5247. Discussion 529References 530

Chapter 24The Development of Categories in the Linguistic and Nonlinguistic Domains: the Same or Different?DIANE POULIN-DUBOIS 535

PART 5 – NEUROSCIENCE OF CATEGORIZATION AND CATEGORY LEARNING

Chapter 25Multiple Systems of Perceptual Category Learning: Theory and Cognitive TestsF. GREGORY ASHBY and VIVIAN V. VALENTINAbstract 5481. Introduction 5492. Two Category-Learning Tasks 5503. COVIS 5504. The COVIS explicit system 552

4.1. Switching attention in the explicit system 5544.2. Long-term storage of explicit category knowledge 556

xvi Table of Contents

5. The COVIS procedural-learning system 5576. Competition between the COVIS explicit and implicit systems 5607. Dissociations between rule-based and information-integration

category learning 5618. Conclusions 563Appendix A 564

A.1. Network implementation of the explicit system 564A.2. Network implementation of the implicit system 566

Acknowledgment 568References 568

Chapter 26The Neuropsychology of Perceptual Category LearningW. TODD MADDOX and J. VINCENT FILOTEOAbstract 5741. Introduction 5752. Competition between verbal and implicit systems (COVIS) 5753. Testing a priori Predictions of COVIS 5784. Perceptual category learning in neurological patients 581

4.1. Nonlinear information-integration category learning in amnesia 5824.2. Nonlinear information-integration category learning in striatal-damaged patients 5854.3. Rule-based category learning in PD 5874.4. Further study of information-integration category learning in PD 591


Chapter 27Neural Regions Associated with Categorical Speech Perception and ProductionSUSAN M. RAVIZZAAbstract 6011. Introduction 6022. Evidence for categorical speech processing 6023. Prefrontal regions and motor speech categories 6054. Temporal–parietal regions and acoustic speech categories 6095. Cerebellar contributions to categorical production and perception 6116. Concluding remarks 612References 613

PART 6 – CATEGORIES IN PERCEPTION AND INFERENCE

Chapter 28Situated ConceptualizationLAWRENCE W. BARSALOUAbstract 620

Table of Contents xvii

1. Introduction 6211.1 Conceptual systems 6211.2 Semantic memory 621

2. Grounding the conceptual system in the modalities 6222.1. Modal reenactments of perception, action, and introspection 6232.2. Simulators and simulations 6242.3. Situated conceptualizations 6262.4. Inference via pattern completion 628

3. Empirical evidence 6293.1. Behavioral evidence for a modal nonmodular conceptual system 6293.2. Neural evidence for a modal nonmodular conceptual system 6373.3. Evidence for situated conceptualizations 639

4. Conclusion 6444.1. Important issues for future research 645

Acknowledgment 647References 647

Chapter 29Perceptual and Semantic Reorganization during Category LearningROBERT L. GOLDSTONE, BRIAN J. ROGOSKY, RACHEL PEVTZOWand MARK BLAIRAbstract 6521. Introduction 6532. Concept learning and perception 653

2.1. Object segmentation 6542.2. Experiment 1 6552.3. Experiment 2 6572.4. Conclusions on perceptual reorganization 662

3. Semantic reorganization during category learning 6643.1. Integral versus separable dimensions 6643.2. Experiment 3 6653.3. Experiment 4 669

4. Conclusions on semantic reorganization 6715. Integrating perceptual and semantic reorganization 671

5.1. Characterizing psychological features 6725.2. Characterizing featural change 6735.3. Prospects for synthesizing perceptual and semantic reorganization 675

Acknowledgments 676References 676

Chapter 30The Return of Concept EmpiricismJESSE J. PRINZAbstract 679

xviii Table of Contents

1. Introduction 6802. Concept empiricism 680

2.1. Representing and doing: Two faces of concepts 6802.2. Variable mechanisms 6842.3. Perceptual vehicles 6862.4. Innateness 6882.5. Summary 691

3. The abstract ideas objections 692References 694

PART 7 – GROUNDING, RECOGNITION, AND REASONING INCATEGORIZATION

Chapter 31Categorization, Reasoning, and Memory from a Neo-logical Point of ViewSERGE ROBERTAbstract 7001. Introduction 7012. Order, Information, and Categories 7013. Inferences, Arguments, and Information 7014. Logical and Creative Arguments 7035. Types of Creative Arguments 7036. Two Rather Incompatible Views in Cognitive Science 7057. Experiments on Memory and Logical Competence 7068. Logical Weakness and Classification of Errors 7099. A New Theory of Logical Error and Logical Competence 710

10. The Cognitive Functions of Logic 71111. Corrective Inferences and Cognitive Progress 71312. The Fundamental Cognitive Function of Logical Reasoning 71513. The Mind as a Dynamic System: Inference and Memory 71514. General Conclusions: Categorization, Reasoning, and Memory 716References 717

Chapter 32Approaches to Grounding Symbols in Perceptual and Sensorimotor Categories ANGELO CANGELOSIAbstract 719

1. Cognitive symbol grounding 7201.1. The symbol grounding problem 7201.2. Grounding symbols in cognition 721

2. Linking vision and language: connectionist approaches to category learning and symbol grounding 7242.1. Connectionist modeling of category learning and naming 7242.2. Connectionist modeling of symbol grounding transfer 727

Table of Contents xix

3. Linking vision, action and language: embodied approaches to language learning and evolution 7293.1. Grounding symbols in simulated agents: The symbolic theft hypothesis 7293.2. The emergence of language in robots 731

4. Discussion and conclusion 733References 735

Chapter 33Embodied CategorizationPIERRE POIRIER, BENOIT HARDY-VALLÉE andJEAN-FRÉDÉRIC DEPASQUALEAbstract 7401. Introduction: Embodied categorization 7412. Purely reactive categorizers 743

2.1. The perceptual aliasing problem [Whitehead and Ballard (1991)] 7442.2. Type I versus II problems [Clark and Thornton (1997)] 744

3. Reactive categorizers that learn 7474. Representing categorizers 7495. Emulating and simulating categorizers 752

5.1. Emulating categorizers 7535.2. Simulating categorizers 755

6. Analogizing categorizers 7597. Linguistic categorizers 7618. Conclusion 761References 762

Chapter 34Categorization of Objects, Scenes, and Faces through TimeÉRIC MCCABE, CAROLINE BLAIS and FRÉDÉRIC GOSSELINAbstract 7681. A model of categorization 7692. Basic-level literature 7723. Discrete processing cycles 773

3.1. A Bubbles primer 7733.2. Fossilized discrete processing cycles 7753.3. What can temporal bubbles reveal about a SLIP categorizer? 777

4. The need for flexibility and a paradox 7784.1. Limited processing capacity 7784.2. The need for flexibility 7794.3. Back to the paradox 780

5. Categorization as an iterative process 7805.1. Compulsory feedforward processing sweeps 7815.2. Flexible iterative processing sweeps 782


xx Table of Contents

Chapter 35Adaptive Categorization and Neural NetworksROBERT PROULX and SÉBASTIEN HÉLIEAbstract 7941. The problem of divergence 7982. The solution: dual Hebbian/anti-Hebbian learning 799

2.1. Stabilization 8002.2. Oscillation 8012.3. Linearity 8012.4. Additional properties of the learning rule 801

3. The Eidos model 8024. The letter classification task 803

4.1. Methodology 8034.2. Results 804

5. The problem of convergence 8066. The solution: Unlearning 8087. The letter-classification task revisited 809

7.1. Methodology 8097.2. Results 809

8. Current trends: Elimination of spurious attractors 8129. Conclusion 813References 814

Chapter 36A Grounded Mind in a Robotic BodySTEVAN HARNAD 817

PART 8 – MACHINE CATEGORY LEARNING

Chapter 37Concept Learning and Nonmonotonic ReasoningPETER GÄRDENFORSAbstract 8241. The role of concepts 8252. Three kinds of cognitive representations 8253. Learning in symbolic systems 8264. Learning in connectionist systems 8275. Conceptual spaces as a representational framework 8276. The origin of quality dimensions 8297. Properties and concepts 8318. Prototypes and conceptual spaces 8329. Learning in conceptual spaces 834

10. The role of similarity in learning 83611. Nonmonotonic aspects of concepts 838

Table of Contents xxi

11.1. Change from general category to subordinate 83911.2. Context effects 840


Chapter 38Categorization in Symbolic Data AnalysisEDWIN DIDAYAbstract 8461. Introduction 8472. Categories, concepts, and symbolic data 848

2.1. From individuals to concepts 8482.2. Categories in a database 8492.3. From categories to concepts: reification of a category in a concept 8492.4. Sources of symbolic data 851

3. Symbolic data tables and their background knowledge, concepts,and categories 8523.1. Symbolic data tables 8523.2. Building a symbolic data table by reification of categories in concepts 8523.3. Description of concepts when the individuals are described by fuzzy data 8533.4. Adding conceptual variables, joining concepts, and the DB2SO

module of SODAS 8544. Modeling concepts by “symbolic objects,” with certain

philosophical aspects 8554.1. Kinds of concepts and intuitive introduction of “symbolic objects” 8554.2. Modeling concepts with four spaces: “individuals,” “concepts,”

“descriptions,” and “symbolic objects” 8554.3. Extent of concepts and symbolic objects 8564.4. Syntax of symbolic objects in the case of “assertions” 8584.5. Extent of a symbolic object 8584.6. Concepts: Four approaches 858

5. Tools for symbolic objects 8595.1. Order between symbolic objects 8595.2. Finding a unique description for a concept: “T-norm of descriptive generalization” 8595.3. Finding several descriptions for a concept 8605.4. Dissimilarities between concepts 8615.5. Finding prototypes from a concept 861

6. Underlying structures of symbolic objects 8616.1. A generalized conceptual lattice 8616.2. Mathematical framework of a symbolic data analysis 863

7. Steps and tools for Symbolic Data Analysis 863

xxii Table of Contents

7.1. Main steps 8637.2. Descriptive SDA in SODAS 864

8. Overview of SODAS 8658.1. Some advantages of the use of concepts modeled by symbolic objects 8658.2. Overview of SODAS software 865

9. Final remarks 865References 866

Chapter 39Category Formation in Self-organizing Embodied AgentsSTEFANO NOLFIAbstract 8691. Introduction 8702. The method 8703. Categories emerging from the interaction between

the agent and the environment 8713.1. Finding and remaining in favorable environmental areas 8713.2. Discriminating objects with different shapes on the basis of tactile information 8733.3. Behavior emerging from the dynamic interaction between the agent and its environment 876

4. Action-mediated sensory states 8784.1. Discriminating larger and smaller cylindrical objects 8794.2. Navigating toward a target area of the environment 881

5. Integrating sensorimotor information over time and the emergence ofcomplex internal categories 8835.1. The self-localization problem 884

6. Conclusions 887Acknowledgments 888References 888

Chapter 40An Information-based Discussion of Vagueness:Six Scenarios Leading to VaguenessDIDIER DUBOIS and HENRI PRADE,FRANCESC ESTEVA and LLUIS GODOAbstract 8921. Introduction 8932. The information framework 8943. Classical vs. gradual properties 895

3.1. Graduality and partial preorderings 8953.2. Membership functions as total preorders 8963.3. Fuzzy sets and similarity to prototypes 8973.4. Set-theoretic operations 897

Table of Contents xxiii

3.5. Graduality is a useful form of vagueness 8984. Precisely defined vs. poorly defined properties 899

4.1. Classification ambiguity 9004.2. Vagueness as limited perception 9004.3. Supervaluations 9014.4. Ill-known partial membership 901

5. Refining precisely defined properties using closeness relations 9016. Single agent vs. multiple agents 9027. Ill-known attribute values and twofold sets 9048. Approximately described sets 9059. Concluding remarks 906References 907

PART 9 – DATA MINING FOR CATEGORIES AND ONTOLOGIES

Chapter 41A Smooth Introduction to Symbolic Methods for Knowledge DiscoveryAMEDEO NAPOLI Abstract 9141. Introduction 9152. Methods for KDD 916

2.1. An introductory example 9162.2. Data mining methods 917

3. Lattice-based classification 9184. Frequent itemset search and association rule extraction 920

4.1. Frequent itemset search 9214.2. Association rule extraction 923

5. Applications 9235.1. Mining chemical reaction database 9245.2. An experiment in biology 9275.3. An introduction to Web mining 928

6. Discussion 9307. Conclusion 930References 931

Chapter 42Genre-Specific Text Mining and Extensional Inductive Concept Recognition:A Pseudocognitive ApproachYVES KODRATOFFAbstract 9361. Introduction and definition of text mining (TM) 937

1.1. Text mining 9371.2. Our approach 938

2. Text retrieval 939

xxiv Table of Contents

3. Standardization 9404. Grammatical tagging 940

4.1. Why expert rules at the tagging stage? 9414.2. A tagging language 9424.3. Our approach to grammatical tagging 9434.4. Automatic learning of new tagging rules 943

5. Terminology 9446. Concept recognition in texts 944

6.1. Polysemy 9456.2. General versus local collocations 9456.3. Terms and collocations 9456.4. ACT as a friendly interface helping the expert 9466.5. ACT as an inductive program 9476.6. Validation 951

7. Conclusion 951Acknowledgments 952References 953

Chapter 43Classification and Categorization in Computer-AssistedReading and Text AnalysisJEAN GUY MEUNIER, DOMINIC FOREST and ISMAIL BISKRIAbstract 9561. Introduction 957

1.1. CARAT: General presentation 9571.2. Difficulties with the technology 9581.3. The nature of reading and analyzing a text 959

2. Definitions of classification and categorization for CARAT 9613. Text classification and categorization 962

3.1. Text classification 9623.2. Text categorization 9623.3. Computer text classification and categorization 963

4. Methodology for text classifying and categorizing 9634.1. Steps 1, 2, and 3: From a text to a matrix 9644.2. Steps 4 and 5 9674.3. Step 6: Navigation 9684.4. Step 7: Evaluation 968

5. Applications in CARAT 9695.1. Thematic analysis 9695.2. Categorical exploration of philosophical texts 9705.3. Content analysis 972

6. The computer design: SATIM 9746.1. The workshop 9746.2. The laboratory 975

Table of Contents xxv

6.3. Applications 9757. Conclusion 976References 976

Chapter 44Graph Matching, System Design and Knowledge ModelingGUY W. MINEAUAbstract 9791. Introduction 9802. Knowledge represented as graph structures 9823. Learning heuristic knowledge 9844. Viability conditions 9855. The complexity of learning 9866. Categorization of knowledge in layers 9887. Conclusion 989References 989

PART 10 – THE NATURALIZATION OF CATEGORIES

Chapter 45Nominalism and the Theory of ConceptsCLAUDE PANACCIOAbstract 9931. Nominalism 9942. Ockham’s cleaver 9953. Motivations 9994. Nominalistic constraints for the theory of concepts 1001

4.1. Represented things as singular 10024.2. Representations as singular 1004

References 1006

Chapter 46Why do We Think Racially?EDOUARD MACHERY, LUC FAUCHERAbstract 10101. Introduction 10112. Is racialism a mere social construct? 1012

2.1. Racial skepticism 10122.2. Races are interactive kinds 10132.3. Races are transient kinds 10132.4. Merits and problems 1015

3. Is racialism a by-product of a human kind module? 1016

xxvi Table of Contents

3.1. The nature of racialism 10163.2. The human kind module 10173.3. Empirical evidence 10173.4. Merits and problems 1019

4. Are races mere coalitions? 10214.1. Races and coalitions 10214.2. Empirical evidence 10214.3. Merits and problems 1022

5. Is racialism a by-product of an evolved ethnic cognitive system? 10245.1. “Ethnies” are not mere coalitions 10245.2. An adaptive scenario: Ethnic cognition and the exaptation of human folk biology 10255.3. Empirical evidence 10265.4. Merits and problems 1027


Chapter 47Neurosemantics and CategoriesCHRIS ELIASMITHAbstract 10361. Introduction 1037

1.1. Why “neuro”? 10371.2. The explanandum 1039

2. Mental representations as neural codes 10402.1. Representations 10402.2. Transformation 10412.3. A representational hierarchy 1042

3. The meaning of neural representations: Neurosemantics 10433.1. The representation relation 10433.2. A neurosemantic theory 10443.3. Discussion 1049

4. Misrepresentation 10505. Conclusion 1052References 1052

Chapter 48 Conceptual Analysis and Philosophical NaturalismELISABETTA LALUMERAAbstract 10551. Introduction 10562. What is intuitive about conceptual analysis? 10573. Cognitive privileges, metaphysical privileges, and the Transparency Thesis 10584. Against privileges 1059

Table of Contents xxvii

5. The inward approach 10616. Conceptual truths or truths about concepts? 10627. The outward approach 10648. ‘Bachelors are unmarried men’ is about facts 10659. Explaining away the illusion 1067

10. A “mixed bag” 106811. Conclusion 1070References 1070

Chapter 49Crisis! What Crisis? 1073PIERRE POIRIER

Index 1081

xxviii Table of Contents

xxix

LIST OF CONTRIBUTORS

Archambault, AnnieDépartement de Psychologie, C.P. 6128, Succ. Centre-ville, Montreal, QC, H3C 3J7Canada.

Ashby, F. GregoryDepartment of Psychology, University of California, Santa Barbara, CA 93106, USA.

Baker, Mark C.Department of Linguistics, Rutgers University, 18 Seminary Place, New Brunswick, NJ08901, USA.

Barsalou, Lawrence W.Department of Psychology, Emory University, Atlanta, GA, 30329, USA.

Biskri, IsmailDépartement de mathématique et d’informatique, Université du Québec à Trois-Rivières, 3351 Boul. des Forges, Trois-Rivières, G9A 5H7 Canada.

Blair, MarkPsychology building, 1101 E 10th Street, Indiana University, Bloomington, IN 47405-7007, USA.

Blais, CarolineDépartement de Psychologie, Université de Montréal, C.P. 6182, Succ. Centre-ville,Montreal, QC, H3C 3J7 Canada.

Boster, James S.Department of Anthropology, U-2176, 354 Mansfield Road, University of Connecticut,Storrs, CT 06269-2176, USA.

Bouchard, DenisDépartement de Linguistique, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Cangelosi, AngeloAdaptive Behaviour and Cognition Research Group, School of Computing,Communications and Electronics, University of Plymouth, Portland Square A316,Plymouth PL4 8AA, UK.

Clark, Eve V.Department of Linguistics, Margaret Jacks Hall (Bldg 460), Stanford University,Stanford, CA 94305-2150, USA.

Cohen, HenriCognitive Neuroscience Center, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Cook, Richard S.University of California at Berkeley, 1203 Dwinelle Hall, Berkeley, CA 94720-2650,USA.

Cousineau, DenisUniversité de Montréal, C.P. 6128, Succ. Centre-ville, Montreal, QC, H3C 3J7 Canada.

Cutler, AnneMax Planck Institute for Psycholinguistics, PO Box 310, 6500 AH Nijmegen, TheNetherlands.

De Pasquale, Jean-FrédéricDépartement d’informatique, U. du Québec à Montréal, C.P. 8888 Succ. Centre-ville,Montreal, QC, H3C 3P8 Canada.

Diday, EdwinParis IX Dauphine University, Place du Maréchal de Lattre de Tassigny, Bureau B633,75775 Paris cedex 16, France.

Dubois, DidierCNRS, Porte 308, IRIT, Université Paul Sabatier, 118 route de Narbonne, 31062, Toulouse,cedex 4, France.

Dubuisson, ColetteDépartement de Linguistique, Université du Québec à Montréal, C.P. 8888 Succ.Centre-ville, Montreal, QC, H3C 3P8 Canada.

Eliasmith, ChrisDepartment of Philosophy, University of Waterloo, Waterloo, ON, N2L 3G1 Canada.

xxx List of Contributors

Esteva, FrancescIIIA – Institut d’Investigació en Intel.ligència Artificial, CSIC – Spanish ScientificResearch Council, Campus Universitat Autonoma de Barcelona, 08193 Bellaterra,Catalonia, Spain.

Faucher, LucDépartement de philosophie, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Filoteo, J. Vincent UCSD/VA, 3350 La Jolla Village Drive, San Diego, CA 92161-116A, USA.

Forest, DominicDépartement de Philosophie, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Gärdenfors, PeterCognitive Science, Kungshuset, Lundagardt, Lund University, SE-222 22 Lund, Sweden.

Gil, DavidDepartment of Linguistics, Max Planck Institute for Evolutionary Anthropology,Deutscher Platz 6, Leipzig 04103, Germany.

Gillon, Brendan S.Department of Linguistics, McGill University, 1085 Doctor Penfield Avenue, Montreal,QC, H3A 1A7, Canada.

Godo Lacasa, LluísInstitut d’Investigació en Intel.ligència Artificial (IIIA), Consejo Superior deInvestigaciones Científicas (CSIC), 08193 Bellaterra, Spain.

Goldstone, Robert R.Psychology building, 1101 E 10th Street, Indiana University, Bloomington, IN 47405-7007, USA.

Gosselin, FredericDépartement de Psychologie, Université de Montréal, C.P. 6182, Succ. Centre-ville,Montreal, QC H3C 3J7 Canada.

Goudbeek, MartijnMax Planck Institute for Psycholinguistics, P.O. Box 310, 6500 AH Nijmegen, TheNetherlands.

Hanson, CatherinePsychology Department, Rutgers University, 101 Waren St., Newark, NJ 07102, USA.

List of Contributors xxxi

Hanson, Stephen JosePsychology Department, Rutgers University, 101 Waren St., Newark, NJ 07102, USA.

Hardy-Vallée, BenoîtInstitut Jean-Nicod, UMR 8129, 1bis, Avenue de Lowendal, F-75007 Paris, France.

Harnad, StevanCognitive Neuroscience Center, Université du Québec à Montréal, C.P. 8888 Succ.Centre-ville Montreal, QC, H3C 3P8 Canada.

Hélie, SébastienLaboratorie d’Études en Intelligence Naturelle et Artificielle, Université du Québec àMontréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Kay, PaulInternational Computer Science Institute, Berkeley, ICSI, 1947 Center St, Suite 600Berkley, CA 94704-1198, USA.

Kodratoff, YvesCNRS, LRI, Université Paris-Sud XI, 91405 Orsay, France. Tel.: 33-1 69156904; Fax:33-1 69156586; E-mail: [email protected]

Labelle, MarieDépartement de linguistique, Université du Québec à Montréal, C.P. 8888, Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Lalumera, ElisabettaDipartimento di Discipline della Comunicazione, Università degli Studi di Bologna, ViaA. Gardino 23, 40131, Bologna, Italy.

Larochelle, SergeDépartement de Psychologie, Université de Montréal Université de Montréal, C.P. 6128, Succ. Centre-ville, Montreal, QC, H3C 3J7 Canada.

Lefebvre, ClaireDépartement de linguistique et didactique des langues, Université du Québec à Montréal,C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Lillo-Martin, DianeDepartment of Linguistics, University of Connecticut, 337 Mansfield Road, Unit 1145,Storrs, CT 06269-1145, USA.

Machery, EdouardDepartment of History and Philosophy of Science, University of Pittsburgh, CL 1017,Pittsburgh, PA 15260, USA.

xxxii List of Contributors

Maddox, W. ToddDepartment of Psychology and Institute for Neuroscience, 1 University Station A 8000,University of Texas, Austin, TX, 78712, USA.

McCabe, ÉricDépartement de Psychologie, Université de Montréal, C.P. 6182, Succ. Centre-ville,Montreal, QC, H3C 3J7 Canada.

Meunier, Jean-GuyDépartement de Philosophie, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Millikan, Ruth* GerrettDepartment of Philosophy, U-2054, University of Connecticut, Storrs, CT 06269-2054,Office: 202 Manchester Hall, USA.

Mineau, GuyFaculty of Sciences and Engineering, Department of Computer Science and SoftwareEngineering, Université Laval, Québec, QC, G1K 7P4 Canada.

Muysken, PieterLinguistics, Radboud University, Nijmegen. Postbus 9103, Nijmegen, The Netherlands.

Napoli, AmedeoLORIA, LORIA, B.P. 239, 54506 Vandoeuvre les Nancy, France.

Nolfi, StefanoInstitute of Cognitive Sciences and Technologies, National Research Council (CNR),Viale Marx 15 00137 Rome, Italy.

Panaccio, ClaudeDépartement de Philosophie, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Papafragou, Anna108 Wolf Hall, University of Delaware, Newark, DE 19716, USA.

Parisot, Anne-MarieDépartement de Linguistique, Université du Québec à Montréal, C.P. 8888 Succ.Centre-ville, Montreal, QC, H3C 3P8 Canada.

Pevtzow, RachelPsychology building, 1101 E 10th Street, Indiana University, Bloomington, IN 47405-7007, USA.

List of Contributors xxxiii

* Important: Pending upon Elsevier’s permission to reproduce

Poirier, PierreDépartement de Philosophie, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3C 3P8 Canada.

Poulin-Dubois, DianePsychology Department, Concordia University, 7141 Sherbrooke Street West,Montreal, QC, H4B 1R6 Canada.

Prade, HenriCNRS, porte 307, IRIT, Université Paul Sabatier, 118 route de Narbonne, 31062,Toulouse, cedex 4, France.

Prinz, Jesse J.Department of Philosophy, CB#3125, Caldwell Hall, University of North Carolina, ChapelHill, NC 27599, USA.

Proulx, RobertFaculté des Sciences Humaines, Université du Québec à Montréal, C.P. 8888 Succ.Centre-ville, Montreal, QC, H3C 3P8 Canada.

Ravizza, Susan M.Department of Psychology, UC Davis Imaging Research Center, 4701 X Street,Sacramento, CA 95817, USA.

Regier, TerryDepartment of Psychology, University of Chicago, 5848 S. University Avenue, Office:Green 414, Chicago, IL 60637, USA.

Rey, GeorgesDepartment of Philosophy, University of Maryland, College Park, MD 20742, USA.

Robert, SergeDepartment of Philosophy, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC H3C 3P8 Canada.

Rogosky, Brian J.Psychology building, 1101 E 10th Street, Indiana University, Bloomington, IN 47405-7007, USA.

Shi, RushenDépartement de Psychologie, Université du Québec à Montréal, C.P. 8888 Succ. Centre-ville, Montreal, QC, H3c 3P8 Canada.

xxxiv List of Contributors

Smits, RoelMax Planck Institute for Psycholinguistics, Wundtlaan 1, P.O. Box 310, 6500 AHNijmegen, The Netherlands.

Sowa, John F.VivoMind LLC, 21 Palmer Avenue, Croton-on-Hudson, NY 10520, USA. Tel.:914-271-5557; E-mail: [email protected]

Swingley, DanielDepartment of Psychology, University of Pennsylvania, 3401 Walnut Street 302 C,Philadelphia, PA 19104, USA.

Thagard, PaulPhilosophy Department, University of Waterloo, Waterloo, ON, N2L 3G1 Canada. Tel.:519-888-4567, ext 3594; Fax: 519-746-3097; E-mail: [email protected]

Toombs, EthanPhilosophy Department, University of Waterloo, Waterloo, ON, N2L 3G1 Canada.

Travis, Lisa deMenaMcGill University, 1085 Dr. Penfield Avenue, Department of Linguistics, Montreal, QCH3A 1A7 Canada.

Valentin, Vivian V.Department of Psychology, University of California Santa Barbara, CA 93106, USA.

White, LydiaDepartment of Linguistics, McGill University, 1085 Dr. Penfield Avenue, Montreal,QC, H3A 1A7 Canada.

List of Contributors xxxv


xxxvi

BRIDGING THE CATEGORY DIVIDE

HENRI COHEN AND CLAIRE LEFEBVRE

Université du Québec à Montréal

Contents

1. Introduction 22. Organization of the book 23. Major common themes 7

3.1. The notions of category and categorization 73.2. The nature of categories: Discrete, vague, or other 93.3. Are there modality effects on categories? 103.4. Are there universal categories? Are there innate categories? 11

4. Bridging the category divide 13References 15

1

Handbook of Categorization in Cognitive Science, Edited by Henri Cohen and Claire Lefebvre© 2005 Elsevier Ltd., All rights reserved

1. Introduction

Categorization is the mental operation by which the brain classifies objects and events.This operation is the basis for the construction of our knowledge of the world. It is themost basic phenomenon of cognition, and consequently the most fundamental problemof cognitive science. Cognitive science is concerned with the kinds of knowledge thatunderlie human cognition, the details of human cognitive processes, and the computa-tional modeling of these processes.

This book presents the study of categories and the process of categorization as viewedthrough the lens of the founding disciplines of the cognitive sciences: cognitive anthro-pology, cognitive computer science, linguistics, neuroscience, philosophy, and psychol-ogy. The study of categorization has long been at the core of each of these disciplines.

The literature on categorization reveals that there is a plethora of definitions, theo-ries, models, and methods to apprehend this central object of study. The contributionsin this handbook reflect this diversity. For example, the notion of category is not uni-form across these contributions and there are multiple definitions of the notion of con-cept. Furthermore, the study of category and of categorization is approached differentlywithin each discipline. For some authors, the categories themselves constitute the objectof study, whereas for others, it is the process of categorization, and for others still, it isthe technical manipulation of large chunks of information. Finally, yet another contrasthas to do with the biological versus artificial nature of agents or categorizers.

Recently, since cognitive science came on the scene, there has been a concertedeffort to establish connections between disciplines. As a result, our understanding ofhuman cognition has been profoundly altered. This book constitutes a major effort tobring the various disciplines together, for the first time, around a single theme: catego-rization. We hope that this collective work will result in the cross-fertilization of meth-ods and ideas, and that it will contribute significantly to our understanding ofcategorization in particular, and of human cognition in general.

In the sections that follow, we present the contents of the book. We then summarizethe major themes and issues that are raised, and we point out some of the controversiesand similarities that emerge between the authors and disciplines represented here.

2. Organization of the book

The book is organized in 10 parts. The first one, Categorization in Cognitive Science,contains six chapters introducing the reader to the notion of category/categorizationfrom the point of view of the six disciplines involved. From the perspective of psy-chology, Harnad considers that categorization is the most fundamental cognitiveprocess and concludes that to cognize is to categorize. From the point of view of lin-guistics, Muysken shows that grammatical categories are not unitary notions, butemerge at the interface of different components of our human cognitive and commu-nicative capacities. Rey discusses how philosophers approach the notion of concept and

2 Henri Cohen and Claire Lefebvre

argues for the importance of empty concepts. Boster provides an overview of the rep-resentations of category structure by cognitive anthropologists in kinship terminologies,color classification, and ethnobiology. From a neuroscientific perspective, Hanson andHanson look at how the brain represents category knowledge and at what informationis acquired during category learning. Finally, Sowa surveys theories of categorizationand reasoning in cognitive science that have been implemented and tested in computersystems. He also shows that, while most of the ideas emerged before the advent of com-puters, this technology did provide the tools for developing these ideas in a way that hadpreviously been impossible.

The other nine parts are organized by theme rather than by discipline. These themesare semantic categories; syntactic categories; acquisition of categories; neuroscience ofcategorization and category learning; categories in perception and inference; ground-ing, recognition, and reasoning in categorization; machine category learning; data min-ing for categories and ontologies; and finally, the naturalization of categories. Ingeneral, the first chapter of each part constitutes an introduction to the theme.

In the first chapter on semantic categories in Part 2, Gillon addresses the question ofwhat a semantic category is. After presenting the two views from which linguists haveaddressed this question – the structural and the notional – he argues that only the struc-tural view has provided sound answers to linguistic questions. The next chapter, byBoster, discusses the degree to which emotions are biologically endowed or culturallyconstructed. It explores the nature of the similarities and differences among culturalemotion systems, and compares methods of carrying out that exploration. Cook, Kay,and Regier present the history of the World Color Survey, and show how this databasehas been used to test the universality of color naming across languages. On the basis ofthe observation that the concept of [ATOM] has undergone numerous changes through-out the history of chemistry, Thagard and Toombs use this concept to discuss the rela-tionship between categorization and conceptual change. Papafragou considers thequestion of whether crosslinguistic differences in marking the count/mass distinction insyntax affect the nonlinguistic individuation criteria used by speakers of different lan-guages. Larochelle, Cousineau, and Archambault examine the role of definitions incategorization and similarity judgments. They show that, while definitions and charac-teristic descriptions are treated alike when making similarity judgments, definitionsappear to be treated in a more unitary fashion when making categorization judgments.Finally, Millikan takes up the challenge of explaining why most concepts are notcategories.

Part 3 on syntactic categories begins with Travis’s introduction to types of syntacticcategories: lexical or major categories, functional or minor categories, crossover cate-gories (e.g., gerunds), and multifunctional categories. While her main goal is to providesome background on the issues raised by the study of syntactic categories, she defendsthe need for categorial information in lexical entries. Although most languages displaya number of syntactic categories, there appear to be languages that are almost deprivedof them. Based on the study of Riau Indonesian, Gil argues that languages of the lattertype share three basic properties. They are morphologically isolating and present no

Bridging the Category Divide 3

word-internal morphological structure, they are monocategorial, and they are semanti-cally associational. From variation in syntactic categories between spoken languages, weturn to differences between languages that may be due to differences in modality. Basedon data from Quebec Sign Language, Bouchard, Dubuisson, and Parisot show that cate-gorization is modality-dependent, and that therefore, some categories are not the same insigned and in spoken languages. Lillo-Martin further explores the similarities and dif-ferences between spoken and signed languages. She shows that, while there are profounddifferences between spoken and signed languages (e.g., the use of spatial information toconvey information about space), in some cases, morphosyntactic properties that at firstglance seemed to show effects of the linguistic modality, are in fact underlyingly similaracross the modes. Both chapters explore the consequences of the modality effect on syn-tactic categories for the notion of linguistic universals. This part of the book ends with adiscussion by Baker, who takes a stand in favor of the universality of grammatical cate-gories both across languages and across modalities (oral and sign).

The fourth part, dedicated to the acquisition of categories, is introduced by Labelle’spresentation of the state of the art. She reviews recent research on the acquisition ofgrammatical categories, focusing on three aspects of the problem – parts of speech,inflection, and subcategories of words – and she reviews the various proposals suggestedto account for the fact that children appear to master the complexity of syntactic cate-gories quite early in their development of language. Considering that both universal andlanguage-specific meanings play a role in children’s acquisition of semantic categories,Clark weighs the respective contributions of universal conceptual categories and of theconventions of the language community in the acquisition of these categories. Shi con-siders early syntactic categories in infants, proposing that infants can derive the distinc-tion between content words and function words on the basis of a constellation ofacoustic/phonetic and phonological cues in the input. Goudbeek, Smits, Swingley, andCutler consider the acquisition of auditory and phonetic categories. This learning isviewed as the formation of categories in a multidimensional psychophysical space.White’s chapter bears on second language acquisition, and more specifically on howsyntactic categories (lexical and functional) are acquired and represented in the interlan-guage grammars of second language (L2) speakers. Several issues are addressed, includ-ing whether L2 learners simply adopt the syntactic categories that are represented in theirL1. In her discussion of the chapters on acquisition, Poulin-Dubois shows how the issuesthat are relevant to the development of categories in the language domain both overlapwith, and differ from, those that are raised in the nonlinguistic domain.

Part 5 focuses on the neuroscience of categorization and category learning. With datafrom animal, lesion, neuropsychological, and computer-modeling studies, neurosciencehas recently witnessed a wealth of results that are homing in on the neural mechanismsand structures that mediate category learning. In the first chapter in this part, Ashby andValentin present COVIS, a well-developed theoretical model of perceptual categorylearning, in which two learning systems are set in opposition. The frontal-based explicitsystem depends on working memory and executive attention, and is mediated primarilyby the anterior cingulate, the prefrontal cortex, and the head of the caudate; the basal


ganglia-mediated system uses procedural learning and requires a dopamine reward sig-nal. Their colleagues Maddox and Filoteo investigate how the predictions made byCOVIS accord with data from neurological patients. In the last chapter, Ravizza showsthat the ease with which we produce and comprehend speech belies the large number ofneural areas supporting these skills. She presents neuropsychological and neuroimagingevidence corroborating the idea that speech processes are categorical. These three con-tributions highlight the role of the specific brain structures involved in categorization.

Part 6 addresses questions related to categories in perception and inference. Ourconceptual system is made up of category representations and is central to memory, lan-guage, and thought. Drawing on psychology and cognitive neuroscience, Barsaloumakes the case that a given concept can produce different situated conceptualizations,each tailored to different instances in different settings. In a somewhat complementaryvein, Prinz takes sides in the debate between rationalists and empiricists, and defendsthe view that conceptual representations are perceptually based. In the last chapter, welearn that our representation of objects not only influences, but is influenced by the con-cepts that we learn. Goldstone, Rogosky, Petzow, and Blair present evidence thatdemonstrates how categorization experience alters the descriptions of objects. In cate-gory learning, we create the elements of categorized objects’ descriptions and, at thesame time, associate those elements with categories.

Grounding, recognition, and reasoning in categorization are the topics covered in Part7. Much of our knowledge is the result of categorization, and reasoning plays an impor-tant role in this process. In the first chapter, Robert investigates the relations between cat-egorization and reasoning, with some discussion of memory, to gain a betterunderstanding of the mechanisms involved in our capacity to build categories. The issueof intrinsically linking the symbols used by a cognitive agent to their correspondingmeanings has been called “the symbol grounding problem.” Cangelosi presents connec-tionist and embodied modeling approaches for the grounding of language in perception,cognition, and action. Poirier, Hardy-Vallée, and DePasquale further explore the embod-ied nature of categorization by studying the categorizing abilities of architecturally sim-ple embodied agents, then turn to living systems, switching from AI and robotics toneuroscience and psychology. Their concern is to show what these living systems have incommon with their artificial counterparts, in an effort to determine how embodimentinfluences categorization. In the next chapter, McCabe, Blais, and Gosselin address someaspects of categorization that have been neglected in the literature: categorization throughtime, limited processing capacities, and the paradox that this creates. Finally, connection-ist networks of categorization rely on Hebbian learning to convert the stimulus space intoa feedback subspace sufficient to categorize new stimuli. The Hebbian learning rulespecifies by how much the weight of the connection between two units should beincreased or decreased in proportion to the product of their activation. The rule builds onHebb’s (1949) learning rule, postulating that the connections between two neurons mightbe strengthened if they fire simultaneously. Proulx and Hélie test a new learning/unlearn-ing procedure applied to an existing model, and review a new approach that reduces thenumber of spurious attractors. This part of the book ends with a discussion by Harnad.


Part 8 addresses issues related to machine category learning. Gärdenfors argues thatthere are aspects of cognitive phenomena for which neither symbolic representation norassociationism seems to offer appropriate modeling tools. In the first chapter of this sec-tion, he advocates a third form of representing information, which employs geometricstructures rather than symbols or associations, outlining how conceptual spaces can beused to model some of the fundamental aspects of how we learn and reason with con-cepts. Diday adopts a computer sciences-based approach to categories and concepts asthey emerge from databases. He attempts to show how these notions from cognitive sci-ence can improve knowledge discovery in text and data mining and how, in return, sym-bolic data analysis can extract categories from concepts. Nolfi focuses on howcategories might emerge, through artificial evolution, from the dynamic interactionbetween situated agents and their environment, and on the relation between categoriesand behavior. Finally, it seems that vagueness can be precisely defined: a concept isvague as soon as it partitions the universe of discourse. In their chapter, Dubois, Esteva,Godo, and Prade discuss and investigate a series of information scenarios concerningvagueness from an AI perspective by focusing on knowledge representation.

Part 9 deals with data mining for categories and ontologies. Extracting informationfrom databases is like searching for gold in riverbeds. In the first chapter, Napoli intro-duces the knowledge discovery process, encompassing the requisite steps for efficientmining operations in very large databases. To solve the problem of concept recognition intexts, Kodratoff defines a new form of learning called extensional induction. The goal ofthese inductive procedures is not to define a concept but to help field experts recognize itspresence in a text. Meunier, Forest, and Biskri also work closely with specialist readers,and show how computer-assisted reading and analysis can help them sift through the con-tent, themes, and concepts of large textual corpora. Modeling complex systems can belikened to a well-honed art where the designer must first ask whether the modeling activ-ities interact with the world with sufficient complexity. In the graph-matching problem,Mineau explores this question and illustrates the relationship between knowledge-basedtechnology and the modeling required to produce such systems.

The role of philosophers in cognitive science is illustrated in the chapters in the lastpart of the book, dealing with the naturalization of categories. In his defense of nomi-nalism – the claim that only singular objects and individuals exist – Panaccio explainshow a nominalistic outlook can bear upon the way we think about concepts, making usaware of certain important problems and distinctions, and avoid widespread confusionamong the things we are talking about when we talk about concepts. For Eliasmith, fun-damental assumptions about semantics remain implicit in most discussions about cate-gorization. He argues that categorization is a universal neural phenomenon, andattempts to outline a semantics for computational neuroscience where neural states arerepresentations. In Machery and Faucher’s view, the time has come to bridge the gapbetween social constructionism and cognitive evolutionary theories of racial catego-rization. Drawing on several disciplines, they attempt to explain why, of all the ways ofdividing the world, we find that some categories, but not others, are natural. In the lastchapter, Lalumera contrasts two positions that a naturalistically minded philosopher


may take toward the intuition of special-status contents. She defends an alternative viewaccording to which relations among concepts mirror the relations among the real-worldproperties they refer to. Poirier concludes this part of the book with an overview of thecontribution of philosophers to cognitive science.

3. Major common themes

A number of common themes run through the chapters in this book. They include thenotions of category and categorization, the nature of categories – whether discrete,vague, or other – whether there is a modality effect on categories, and finally, whetherthere are categories that are universal and categories that are innate. In this section, wesummarize what our authors have to say about these themes.

3.1. The notions of category and categorization

The first question that one may ask is whether the notion of category is uniform acrossdisciplines and authors. Looking back at the contents of Part 1, one may already sur-mise the answer to this question. Harnad focuses on cognitive states and processes,Muysken on grammatical categories, Rey on concepts and empty concepts, Boster onstructural arrangements of categories, Hanson and Hanson on how the brain – and thedamaged brain – represents category knowledge, and Sowa on artificial system cate-gories. In the following paragraphs, we present the various ways that the notions of cat-egory and categorization are being used by the various authors in this book.

The linguists refer to phonetic, phonological, syntactic, semantic, and lexical cate-gories. All these categories may be described in terms of feature bundles. For example,major lexical categories are defined by a combination of the major features [�/–N(oun)],[�/–V(erb)], yielding the four major lexical categories in (1) [Chomsky (1970)]:

(1) [� N, – V]: Nouns[– N, � V]: Verbs[� N, � V]: Adjectives[– N, – V]: Pre-/postpositions

Minor or functional categories are defined in terms of minor features such as[�/–Det(erminer)], [�/–T(ense)], [�/–P(lural)], etc. Likewise, semantic categoriesmay be defined by features such as [�/–human], [�/–abstract], and so on. Althoughthere is debate concerning some cases (e.g., the status of prepositions as major- orminor-category lexical items), there is a general consensus among linguists as to whatconstitutes a grammatical category.

For cognitive anthropologists, the notion of category is not far removed from that oflinguists. In their view, categories (e.g., color terms, kinship terms, etc.) reveal themselvesthrough the lexicon, and therefore, the basic task of the investigator consists in exploringthe cognitive organization of the lexicon for a given domain, with componential analysis


as an important tool. Several authors, whose studies are reported by Boster, make use offeature bundles in referring to folk categories.

In contrast to linguists and cognitive anthropologists, other contributors focus on con-cepts or the process of categorization as their object of study. As Rey discusses at length,there is no consensus in the literature on philosophy concerning the definition of theterm. The philosophers in this volume, however, appear to share the general view thatconcepts and categories are grounded in experience. For Lalumera, the relations betweenconcepts mirror the relationships between the real-world properties that they refer to. ForPanaccio, concepts represent singular things. For Machery and Faucher, social controlfactors impact on the conditions surrounding the formation of categories and concepts.These authors consider that social categories require a more integrated approach to cat-egorization. This may help explain why people classify humans on the basis of theirphysical properties; this is situated cognition with an added cultural construction.

The definitions offered by certain authors as to what constitutes a category some-times reflect their object of study. For example, categories are defined by some as dif-ferent classes of environmental situations that emerge from the interaction between theagent and the environment. This is especially the case with authors who adopt a situ-ated or embodied view of cognition, such as Cangelosi, Harnad, Nolfi, and Poirier et al.It is also the case with others, such as Gärdenfors, for whom concepts constitute bridgesbetween perception, reasoning, and action. In this general view, concepts and categoriescan be considered as abstracted from experience with the world, and they are ever-evolving rather than fixed.

In his highly detailed account of concept and category representation, Barsalou con-trasts two ways of thinking about concepts. In the first one, set within a semantic mem-ory perspective, the properties and exemplars of a category are integrated into a generaldescription that is relatively detached from the goals of specific agents. In the second,a concept can be viewed as an agent-dependent instruction manual that delivers spe-cialized packages of inferences to guide an agent’s interactions with particular categorymembers in specific situations. For Barsalou, our conceptual system is a collection ofcategory representations, widely distributed in the brain, and rich with knowledgeacquired during one’s life span. It is a dynamic system, as it anticipates, categorizes,and provides inferences following categorization, which constitutes expertise about theworld. Thus, knowledge about a category provides a great deal of detailed informationabout the diverse range of its instances.

A number of contributors adopt, implicitly or explicitly, a position that is in line withBarsalou’s view of concepts and categories. A strong view is expressed by Prinz, forwhom concepts have their basis in perception and represent categories by reliablecausal relations to category instances. In this view, concepts are not fixed but vary fromoccasion to occasion. In Prinz’s approach, as with most psychologists, concepts consti-tute the tools for categorizing. They must be built up from features, and in contrast toword-like entities, they cannot be unstructured.

Goldstone et al. share similar views and their work complements that of Barsalou,Nolfi, and Prinz. For them, category learning not only depends upon perceptual and


semantic representations but leads to the generation of these representations. In short,categorization experience not only uses descriptions of objects, it also alters thesedescriptions. Here again, the emphasis is on action-mediated states used as the basis forconstructing and elaborating categories. Concepts are complex databases and theyallow us to represent, predict, and interact with the world. Many contributors hold theview that representing and doing are intimately connected.

Other authors are more interested in categorization processes than in categories perse. This is the case of neuroscientists such as Hanson and Hanson, for whom the mainfocus of study is not categories or concepts as such but the study of the processesinvolved in categorization and the localization of these operations in the brain.Categories, however, play a central role in these studies, as they are used to test theassumptions of the category-learning systems, as is the case in Ashby and Valentin’sand in Maddox and Filoteo’s contributions. In this context, categories are user-defined,and generally represent classes of stimuli that share similar attributes (e.g., segmentedlines with a particular orientation and length).

In Ravizza’s work, the notion of category is closer to that of linguists and of psy-cholinguists, as categorization relies on the perception of a number of distinctive features(e.g., acoustic, articulatory) that reflect the unique attributes of a particular phoneme. Inaddition, Ravizza holds that motor speech representations and information about acousticfeatures must be maintained for successful categorical perception and production.

Authors who work on data mining for categories and ontologies are rather vague aboutwhat constitutes a category or a concept. Although the aim of data mining is to extract cat-egories and concepts from large (and sometimes noisy) databases, there is no proper def-inition of what exactly constitutes a category. Categories and concepts are constructsborrowed from cognitive science to help with the extraction process. They are consideredto be given, and the expert user is assumed to already possess good exemplars of what tolook for. In data mining, the main concern is not with categories and concepts per se, butwith the operations and procedures that contribute to extracting them from large data-bases. Meunier et al., however, consider that it is in the selection of the labels, expressingsome aspect of the “semantic” content of a class of textual entities, that the cognitive andstructural dimensions of categorization are called upon in text mining.

It therefore appears that the notion of category is not uniform across disciplines andis not always a central concern. For some researchers, it is the object of study. For oth-ers, categories are the end result, and it is the process and mechanisms of categoriza-tion that are important.

3.2. The nature of categories: Discrete, vague, or other

Another topic that emerges from some of the contributions in this book pertains to thenature of categories, whether discrete, vague, or other.

Grammatical categories may be discrete or nondiscrete. Examples of discrete gram-matical categories are nouns and verbs, which are defined by opposite features [see (1)].Grammatical categories can also be mixed, in the sense that they draw some of their


properties from one category and some from another. This is the case of adjectives, whichdraw some of their properties from nouns and some from verbs. The features that definethem in (1), [�N, �V], are a means of representing these mixed properties. Gerunds, dis-cussed by Travis, also have mixed properties. They have the characteristics of both nounsand verbs; furthermore, there are several types of them, which differ as to how nominal orverbal they are. Grammatical categories can also be multifunctional. Some lexical itemscan be used with more than one function (e.g., the English word phone, which can be usedas a noun – a phone – or as a verb – to phone) and have different argument structuresdepending on the syntactic realization (as a noun, phone has no argument structure, but asa verb, it has two arguments: the phon-er and the phon-ee) (see Travis). Multifunctionalcategories are also extensively documented by Gil on the basis of Riau Indonesian, wherethe bulk of the lexicon is claimed to be multifunctional. Multifunctional categories are alsofound in sign languages. Bouchard et al. report that in Quebec Sign Language, verb rootsare not distinguished from noun roots and that pronouns are not distinguished from deter-miners. Furthermore, a great deal of Muysken’s chapter is dedicated to showing thatgrammatical categories are not unitary notions, but emerge at the interface of differentcomponents of human cognitive and communicative capacities.

Contributors to this book from other disciplines are concerned with the discrete natureof category representation. For Harnad, the innate ability to build discrete and hierarchi-cally ordered representations of the environment (i.e., categories) is the basis of all higher-order cognitive abilities, including language. Categorical perception is a representationalprocess, resulting in the compression of within-category differences between members ofthe same category, and the expansion of between-category distances among members ofdifferent categories, as with the categorical perception of some speech features. Categoricalperception has been shown to occur in animals and human subjects, as well as in artificialsystems, as shown by Nolfi and Poirier et al. McCabe et al. present novel evidence that weapprehend the world via discrete processing cycles. They suggest that the discrete parti-tioning of our experience with the world extends to the visual domain. This contrasts withthe general view that our everyday experience of time is continuous in nature.

Categories and concepts can also be vague. Vagueness is expressed as an overlapbetween categories. That is, members for which it cannot be cognitively decidedwhether they belong in one category or another end up placed at the periphery of cate-gories, such that adjacent categories gradually merge. But, as is argued by Diday,vagueness should not be interpreted as a weakness, and it should be distinguished fromnonspecificity and ambiguity. Furthermore, graduality, or the graded membership ofexemplars in a particular category, can be a useful form of vagueness as it helps to bet-ter capture the idea of typicality, and to interface linguistic categories with a continuumof attribute values without introducing arbitrary discontinuities.

3.3. Are there modality effects on categories?

Are there modality effects on categories? This question is pertinent for most contribu-tors to the book. For linguists, modality refers to the oral or signed nature of languages.


Oral languages are primarily sequential and use the vocal/auditory channel, whereassign languages are primarily spatial and use the manual (corporeal)/visual channel.Both Lillo-Martin and Bouchard et al. address the question of whether this differencein modality of transmission has an effect on categories. Bouchard et al. conclude that“Oral and sign languages are actually very similar in the fundamental principles of theirsyntax, but important physico-perceptual differences between their modalities deter-mine the surface realizations of these principles in ways that make them appear verydifferent.” As for Lillo-Martin, she concludes that morphosyntactic categories thatappear to show the effects of the linguistic modality at first glance can be argued to besimilar at a deeper level upon a second examination. Baker further explores this issueand presents a strong argument in favor of categorial similarity across modalities.

In psychology and neuroscience, and for some authors in philosophy, modality refersto the level of sensory experience (e.g., auditory, visual, tactile, etc.). Some authors see aclear association between situated and embodied cognition and a modal representation ofcategories. Barsalou speaks of a modal reenactment. Enactment represents the notion that,when people act, they bring structures and events into existence and set them in action. InBarsalou’s view, modal simulations underlie conceptual processing. The modal reenact-ment of perceptual, motor, and introspective states is assumed to be quite similar to thereenactment process underlying mental imagery. In his approach, the conceptual systemshares fundamental mechanisms with modality-specific systems. Thus, in situated con-ceptualization, many different specialized representations can be constructed for a givenconcept, each tailored to situation-specific goals and constraints. This conceptual systemis distributed in the brain, where “convergence zones” capture the patterns of activationevoked in the scaffolding of these representations. Barsalou’s model is an ecological con-ceptual system, in touch with an agent’s experience and anticipated actions. It is adynamic system that feeds on the interactions between an agent and the environment.Although not clearly stated, the notion of emergence – well developed in Nolfi’s chapter– is an important element of this conceptual system. This general view of modal repre-sentation appears to be shared by many authors in the book, though not always explicitly.

Modality thus appears to be of central importance to most authors interested inhuman and artificial cognition. For linguists, modality effects on categories reduce tothe visual and auditory domains. For most others, the sensorium plays an important partin the shaping of our categories and concepts.

3.4. Are there universal categories? Are there innate categories?

Several chapters of this book address the question of whether there are categories thatare universal, and of whether there are innate categories. As will be seen below, peopletend to hold strong views about these issues.

Like Chomsky, linguists assume that there is a language acquisition device calledUniversal Grammar (UG), which is part of the human biological endowment. Byhypothesis, UG comprises the principles and parameters that define the form of naturallanguages (as opposed to artificial languages). Principles and parameters need not be


learned for they are hypothesized to be innate. What needs to be learned are the prop-erties that are language specific: the language-specific lexical properties and the para-metric values of particular languages. The problem that linguists face is distinguishingthe properties of language that are universal from those that are language-specific, andthose that are innate from those that are learned. For example, speaking of semantic cat-egories, Clark questions the extent to which children’s semantic categories are informedby universal conceptual categories or by the conventions of the language community.Some conceptual categories are assumed to be part of the universal human apparatusincluding notions of space, event cognition, quantity/number, causality, agency, andanimacy (see Papafragou). Are (some) grammatical categories universal? Competingviews on this issue are presented in several chapters of this book, for example, Travis,Gil, Bouchard et al., Lillo-Martin, Baker, Labelle, and Poulin-Dubois.

Cognitive anthropologists are also concerned with distinguishing between universalsand culture-specific categories, and between innate and learned ones. In the case of colorperception, categorical perception is claimed to be innate rather than learned. As Harnadstates, “color boundaries along the visible spectrum are a result of inborn feature detectorsrather than of learning to sort and name colors in particular ways.” Furthermore, buildingon Berlin and Kay (1969), Cook et al. show that universal crosslinguistic constraints oncolor naming exist, and that basic color terminology systems tend to develop in a partiallyfixed order, thus revealing the universality of color naming across languages. In the caseof emotions, Boster states his position in the following terms: “Rejecting both extremeuniversalism and relativism, I insist that emotions are at once absolutely universally bio-logically endowed and completely locally culturally constructed.”

Racialism has been claimed to be a by-product of a human kind module. In theirchapter on why we think racially, Machery and Faucher critically examine the evidenceconcerning the innateness hypothesis of racialism. Based on a large number of both the-oretical and empirical considerations, they reject the claim that racialism results froman innate and essentialist nature of racial categories.

Although many of the contributors in this book do not directly address the questionof the innate or learned nature of categories, it is possible to find a common viewbetween some of them. For those who consider that categories and concepts are percep-tually based, and that the relations between concepts mirror the relationships between thereal-world properties they refer to, the consensus would be that concepts are learned.Harnad, for one, thinks that categorization is a skill, and that, like all skills, it must belearned. Prinz makes a strong argument for the view that concepts are learned rather thaninnate. For authors in AI research, the construction of categories by artificial systems isdefinitely a product of the artificial agent’s interaction with the environment.

The view in situated and embodied conceptualization appears to be that the innateversus learned distinction is a moot point. For proponents of this position, it is possiblethat organisms are prepared to categorize their environment, but the perceptual basis ofmental representations is what is well established.

The question of whether categories are innate or acquired is thus a concern mainlyfor linguistics, cognitive anthropology, and philosophy. Nearly four centuries after


Descartes revived it, the debate between the rationalists and the empiricists is still anactive one. As the focus of study within some disciplines has shifted from categories toprocesses of categorization, this has become less of an issue for psychology and neuro-science. One issue, however, that needs further clarification is the relationship betweenwhat is universal and what is biologically endowed.

One is struck by the richness and variety of the views expressed in the book.Probably because each discipline focuses on particular objects of study, our represen-tation of categories is still diverse, we still approach the nature of categories from dif-ferent perspectives, and there is a great diversity in what constitutes the central objectof study. Except in rare cases, the views expressed are not so much competing as dif-ferent, and thus they hold great promise for complementarity. In the next section, wenow turn to the discussion of the bridges that may be built between these views.

4. Bridging the category divide

As we saw in the preceding section, most disciplines in cognitive science approach thestudy of categorization from well-established perspectives. Nonetheless, there has beensome effort to bring some of the contributing disciplines closer. In this section, we callattention to the bridges that have already been built, and point out avenues for furthercross-fertilization between disciplines.

Among the bridges between disciplines that are on firm foundations, the neuro-science of category learning and categorization reflects the solid contribution of threefounding disciplines: psychology, neuroscience, and cognitive computer science.Learning systems, brain anatomy, and neural network modeling have each contributedsignificantly to complementary aspects of human cognition, resulting in a better under-standing of the cognitive processes involved in categorization.

Another area of study which has greatly benefited from contributions by several dis-ciplines is grounding, recognition, and reasoning. Reflections and advances within phi-losophy, AI, psychology, neuroscience, and linguistics concerning concept and categoryformation have been well integrated. The authors refer to similar concepts and share thesame vocabulary. Their contributions, as well as those on the perception of categoriesand inference, all converge on the importance of situated and embodied aspects of cog-nition. This constitutes a significant breakthrough, considering that the authors fromthese disciplines have reached similar conclusions from different starting points.Studies in human cognition have thus taken a turn toward more complex and ecologi-cally valid representations of our perceptions.

Other disciplines where bridges already exist are linguistics and cognitive anthro-pology. Both disciplines share a nativist approach that goes back to Chomsky (1957)and Goodenough (1957). While linguists are interested in the knowledge that enablesspeakers of a given language to comprehend and use this language as a native speaker,cognitive anthropologists are interested in the knowledge that enables members of asociety to comprehend this society and behave in it as a native. While linguists are


interested in a theory of grammar that enables speakers to learn and produce language,rather than in a simple description of their linguistic productions, cognitive anthropol-ogists are interested in a theory of how people conceptualize the world rather than in adescription of their behavior. Both disciplines are interested in discovering universals –of language and of culture, respectively – and in setting these apart from specifics – oflanguage and of culture, respectively. It appears that the advances of the last 50 yearsin both linguistics and cognitive anthropology have been significant in establishing thegoals and the framework for each of these disciplines.

There are other disciplines that benefit from each other’s input. This is the case oflinguistics and AI, with respect to data and text mining. This is also the case w

Date post:	13-Mar-2020
Category:	Documents
Upload:	others
View:	2 times
Download:	0 times

Handbook of Categorization in Cognitive Science · 2016-10-25 · HANDBOOK OF CATEGORIZATION IN...

Documents