Mind, Body, World: Foundations of Cognitive Science

506 Pages·2013·4.052 MB·English

by  Michael R.W. Dawson

MIND, BODY, WORLD
OPEL (open paths to enriched learning)
Series Editor: Connor Houlihan
Open Paths to Enriched Learning (OPEL) reflects the continued commitment of Athabasca
University to removing barriers — including the cost of course materials — that restrict access to
university-level study. The OPEL series offers introductory texts on a broad array of topics, written
especially with undergraduate students in mind. Although the books in the series are designed for
course use, they also afford lifelong learners an opportunity to enrich their own knowledge. Like all
AU Press publications, OPEL course texts are available for free download at www.aupress.ca, as well
as for purchase in both print and digital formats.
series titles
Open Data Structures: An Introduction
Pat Morin
Mind, Body, World: Foundations of Cognitive Science
Michael R.W. Dawson
M I N D,
FOuNDatIONs OF COgNItIve sCIeNCe
B O D Y,
MIChaeL R. W. DaWsON
W O R L D
Copyright © 2013 Michael R. W. Dawson
Published by AU Press, Athabasca University
1200, 10011 – 109 Street, Edmonton, AB T5J 3S8
A volume in OPEL (Open Paths to Enriched Learning)
issn 2291-2606 (print) 2291-2614 (digital)
Cover design by Marvin Harder, marvinharder.com.
Interior design by Sergiy Kozakov.
Printed and bound in Canada by Marquis Book Printers.
library and archives canada cataloguing in publication
Dawson, Michael Robert William, 1959—, author
Mind, body, world: foundations of cognitive science / Michael R. W. Dawson.
(OPEL (Open paths to enriched learning), 2291-2606 ; 2)
Includes bibliographical references and index.
Issued in print and electronic formats.
isbn 978-1-927356-17-3 (pbk.) — isbn 978-1-927356-18-0 (pdf) — isbn 978-1-927356-19-7 (epub)
1. Cognitive science. I. Title. II. Series: Open paths to enriched learning ; 2
BF311.D272 2013 153 C2013-902162-0
C2013-902163-9
We acknowledge the financial support of the Government of Canada through the Canada Book Fund
(cbf) for our publishing activities.
Assistance provided by the Government of Alberta, Alberta Multimedia Development Fund.
This publication is licensed under a Creative Commons licence, Attribution-Noncommercial-No
Derivative Works 2.5 Canada: see www.creativecommons.org. The text may be reproduced for non-
commercial purposes, provided that credit is given to the original author.
To obtain permission for uses beyond those outlined in the Creative Commons licence, please contact
AU Press, Athabasca University, at [email protected].
Contents
List of Figures and Tables | ix
Preface | xiii
Who Is This Book Written For? | xiv
Acknowledgements | xv
Chapter 1. The Cognitive Sciences: One or Many? | 1
1.0 Chapter Overview | 1
1.1 A Fragmented Psychology | 2
1.2 A Unified Cognitive Science | 3
1.3 Cognitive Science or the Cognitive Sciences? | 6
1.4 Cognitive Science: Pre-paradigmatic? | 13
1.5 A Plan of Action | 16
Chapter 2. Multiple Levels of Investigation | 19
2.0 Chapter Overview | 19
2.1 Machines and Minds | 20
2.2 From the Laws of Thought to Binary Logic | 23
2.3 From the Formal to the Physical | 29
2.4 Multiple Procedures and Architectures | 32
2.5 Relays and Multiple Realizations | 35
2.6 Multiple Levels of Investigation and Explanation | 38
2.7 Formal Accounts of Input-Output Mappings | 40
2.8 Behaviour by Design and by Artifact | 41
2.9 Algorithms from Artifacts | 43
2.10 Architectures against Homunculi | 46
2.11 Implementing Architectures | 48
2.12 Levelling the Field | 51
Chapter 3. Elements of Classical Cognitive Science | 55
3.0 Chapter Overview | 55
3.1 Mind, Disembodied | 56
3.2 Mechanizing the Infinite | 59
3.3 Phrase Markers and Fractals | 65
3.4 Behaviourism, Language, and Recursion | 68
v
3.5 Underdetermination and Innateness | 72
3.6 Physical Symbol Systems | 75
3.7 Componentiality, Computability, and Cognition | 78
3.8 The Intentional Stance | 82
3.9 Structure and Process | 85
3.10 A Classical Architecture for Cognition | 89
3.11 Weak Equivalence and the Turing Test | 93
3.12 Towards Strong Equivalence | 97
3.13 The Impenetrable Architecture | 106
3.14 Modularity of Mind | 113
3.15 Reverse Engineering | 119
3.16 What is Classical Cognitive Science? | 122
Chapter 4. Elements of Connectionist Cognitive Science | 125
4.0 Chapter Overview | 125
4.1 Nurture versus Nature | 126
4.2 Associations | 133
4.3 Nonlinear Transformations | 139
4.4 The Connectionist Sandwich | 142
4.5 Connectionist Computations: An Overview | 148
4.6 Beyond the Terminal Meta-postulate | 149
4.7 What Do Output Unit Activities Represent? | 152
4.8 Connectionist Algorithms: An Overview | 158
4.9 Empiricism and Internal Representations | 159
4.10 Chord Classification by a Multilayer Perceptron | 162
4.11 Trigger Features | 172
4.12 A Parallel Distributed Production System | 177
4.13 Of Coarse Codes | 184
4.14 Architectural Connectionism: An Overview | 188
4.15 New Powers of Old Networks | 189
4.16 Connectionist Reorientation | 193
4.17 Perceptrons and Jazz Progressions | 195
4.18 What Is Connectionist Cognitive Science? | 198
Chapter 5. Elements of Embodied Cognitive Science | 205
5.0 Chapter Overview | 205
5.1 Abandoning Methodological Solipsism | 206
5.2 Societal Computing | 210
5.3 Stigmergy and Superorganisms | 212
5.4 Embodiment, Situatedness, and Feedback | 216
vi
5.5 Umwelten, Affordances, and Enactive Perception | 219
5.6 Horizontal Layers of Control | 222
5.7 Mind in Action | 224
5.8 The Extended Mind | 230
5.9 The Roots of Forward Engineering | 235
5.10 Reorientation without Representation | 239
5.11 Robotic Moments in Social Environments | 245
5.12 The Architecture of Mind Reading | 250
5.13 Levels of Embodied Cognitive Science | 255
5.14 What Is Embodied Cognitive Science? | 260
Chapter 6. Classical Music and Cognitive Science | 265
6.0 Chapter Overview | 265
6.1 The Classical Nature of Classical Music | 266
6.2 The Classical Approach to Musical Cognition | 273
6.3 Musical Romanticism and Connectionism | 280
6.4 The Connectionist Approach to Musical Cognition | 286
6.5 The Embodied Nature of Modern Music | 291
6.6 The Embodied Approach to Musical Cognition | 301
6.7 Cognitive Science and Classical Music | 307
Chapter 7. Marks of the Classical? | 315
7.0 Chapter Overview | 315
7.1 Symbols and Situations | 316
7.2 Marks of the Classical | 324
7.3 Centralized versus Decentralized Control | 326
7.4 Serial versus Parallel Processing | 334
7.5 Local versus Distributed Representations | 339
7.6 Internal Representations | 343
7.7 Explicit Rules versus Implicit Knowledge | 345
7.8 The Cognitive Vocabulary | 348
7.9 From Classical Marks to Hybrid Theories | 355
Chapter 8. Seeing and Visualizing | 359
8.0 Chapter Overview | 359
8.1 The Transparency of Visual Processing | 360
8.2 The Poverty of the Stimulus | 362
8.3 Enrichment via Unconscious Inference | 368
8.4 Natural Constraints | 371
8.5 Vision, Cognition, and Visual Cognition | 379
8.6 Indexing Objects in the World | 383
vii
8.7 Situation, Vision, and Action | 390
8.8 Scaffolding the Mental Image | 394
8.9 The Bounds of Cognition | 397
Chapter 9. Towards a Cognitive Dialectic | 399
9.0 Chapter Overview | 399
9.1 Towards a Cognitive Dialectic | 400
9.2 Psychology, Revolution, and Environment | 406
9.3 Lessons from Natural Computation | 412
9.4 A Cognitive Synthesis | 417
References | 425
Index | 485
viii
List of Figures and Tables
Figure 2-1. (A) An electrical switch, labelled x. (B) Switches x and y in series. (C)
Switches x and y in parallel. | 31
Figure 2-2. A relay, in which a signal through an electromagnetic gate controls a switch
that determines whether the current from the source will flow through the
drain. | 35
Figure 3-1. The starting configuration for a five-disc version of the Tower of Hanoi
problem. | 62
Figure 3-2. An intermediate state that occurs when MoveStack () is applied to a five-
disc version of the Tower of Hanoi. | 63
Figure 3-3. The root of the Sierpinski triangle is an equilateral triangle. | 64
Figure 3-4. The second step of constructing a Sierpinski triangle. | 64
Figure 3-5. The Sierpinski triangle that results when the recursive rule is applied four
times to Figure 3-4. | 65
Figure 3-6. A phrase marker for the sentence Dogs bark. | 66
Figure 3-7. Phrase markers for three noun phrases: (A) the dog, (B) the cute dog, and (C)
the cute brown scruffy dog. Note the recursive nature of (C). | 67
Figure 3-8. How a Turing machine processes its tape. | 69
Figure 3-9. How a finite state automaton processes the tape. Note the differences
between Figures 3-9 and 3-8. | 70
Figure 3-10. Results of applying MDS to Table 3-1. | 88
Figure 3-11. Unique features pop out of displays, regardless of display size. | 101
Figure 3-12. Unique combinations of features do not pop out. | 102
Figure 3-13. The Müller-Lyer illusion. | 111
Figure 4-1. A distributed memory, initially described by James (1890a) but also part of
modern connectionism. | 136
Figure 4-2. (A) Pattern space for AND; (B) Pattern space for XOR. | 143
Figure 4-3. A Rosenblatt perceptron that can compute the AND operation. | 144
Figure 4-4. A multilayer perceptron that can compute XOR. | 146
Figure 4-5. A typical multilayer perceptron has no direct connections between input
and output units. | 147
Figure 4-6. Probability matching by perceptrons. Each line shows the perceptron
ix
activation when a different cue (or discriminative stimulus, DS) is presented.
Activity levels quickly become equal to the probability that each cue was
reinforced (Dawson et al., 2009). | 154
Figure 4-7. A small piano keyboard with numbered keys. Key 1 is C. | 162
Figure 4-8. The C major scale and some of its added note chords. | 162
Figure 4-9. The circle of fifths. | 163
Figure 4-10. The two circles of major seconds. | 167
Figure 4-11. The four circles of major thirds. | 168
Figure 4-12. The hidden unit space for the chord classification network. H1, H2, and H3
provide the activity of hidden units 1, 2, and 3 respectively. | 171
Figure 4-13. An example of output unit partitioning of the hidden unit space for the
chord classification network. | 172
Figure 4-14. Any input pattern (dashed lines) whose vector falls in the plane orthogonal
to the vector of connection weights (solid line) will be a trigger feature for a
hidden value unit. | 174
Figure 4-15. An example of banding in a jittered density plot of a hidden value unit in a
network that was trained on a logic problem. | 175
Figure 4-16. Coordinates associated with each output note, taken from an MDS of the
Table 4-8 correlations. Shading reflects groupings of notes as circles of
major thirds. | 197
Figure 8-1. Underdetermination of projected movement. | 364
Figure 8-2. The aperture problem in motion perception. | 365
Figure 8-3. An example Sudoku puzzle. | 372
Figure 8-4. The “there can be only one” constraint propagating from the cell labelled 5 |
372
Figure 8-5. The “last available label” constraint. | 373
Figure 8-6. The “naked pair constraint.” | 374
Figure 8-7. The motion correspondence problem. | 376
Figure 8-8. Pylyshyn’s theory of preattentive visual indexing provides referential links
from object files to distal objects in the world. | 390
Figure 9-1. Word cloud generated from the text of Chapter 3 on classical cognitive
science. | 401
Figure 9-2. Word cloud generated from the text of Chapter 4 on connectionist cognitive
science. | 402
Figure 9-3. Word cloud generated from the text of Chapter 4 on embodied cognitive
science. | 402
x