430
BOOK REVIEWS
physiological, psychophysical, cognitive-psychological, linguistic, ethnographic. This is the territory cognitive science must traverse. In these books on color naming the reader will find the journey to be well underway. References Berlin, B. and Kay, P. (1969), Basic Color Terms: Their Universality and Evolution, University of California Press. Dedrick, D. (1998), Naming the Rainbow: Colour Language, Colour Science, and Culture, Dordrecht: Kluwer Academic Publishers. Kay, P., Berlin, B., Maffi, L. and Merrifield, W., World Color Survey, (Forthcoming). Lucy, J. and Schweder, R.A. (1979), Whorf and his Critics: Lingusitic and Non-linguistic Influences on Color Memory, American Anthropologist 81, pp. 581–615. Rosch, E. (1975), Cognitive reference points, Cognitive Psychology 7, pp. 532–547. Sahlins, M. (1976), Culture and Practical Reason, University of Chicago Press. Saunders, B. A. C. and van Brakel, J. (1997), “Are their non-trivial constraints on color categorization”, Behavioral and Brain Sciences 20(2), pp. 167–179.
Department of Philosophy Concordia University Montreal, Quebec, Canada H3G1M8 E-mail:
[email protected]
DON DEDRICK
Andy Clark, Jesús Ezquerro, and Jesús M. Larrazabal (eds.), Philosophy and Cognitive Science: Catergories, Consciousness, and Reasoning, Philosophical Studies Series 69, Dordrecht: Kluwer, 1996, xviii + 270 pp., ISBN 0-7923-4068-X. This book is a collection of eleven papers that were presented at the Second International Colloquium on Cognitive Science held at San Sebastian on May 7–1, 1991. In the ten-page introduction by the editors that provides an overview of all the papers, the goal of the colloquium is set forth: namely to explore the philosophical foundation of the rapidly evolving field of cognitive science. These two factors pretty much set the stage for what this volume delivers. As this book grew out of a conference proceeding, the papers gathered here are loosely related at best. Despite the editors’ attempt to draw various threads together into a unifying braid in the introduction, the chapters themselves bear almost no relation to each other. Their themes cover diversity of topics including the purely philosophical issues of meaning and reference (Chapter 5 and 6), and the Al-related themes of how to model commonsense causal and default reasoning (Chapter 10), diagrammatic reasoning (Chapter 9) and uncertain reasoning (Chapter 11). On the positive side, this volume includes contributions from such well-known figures as John Perry, Keith Lehrer, Martin Davies, and others. On the negative side, though the conference was held in 1991, and the current volume was published in 1996, there seems to be little effort at post-colloquium editing and updating of the papers. There also does not seem to be any effort by different authors towards identifying
BOOK REVIEWS
431
some common themes by commenting on each other’s work in an effort to provide different perspectives on the same problem. A few chapters seem to end rather abruptly, as if the authors suddenly ran out of steam or space. Moreover, some of the chapters seem to be only slightly revised versions of papers that appeared somewhere else and pre-date this volume. All these factors, compounded with some minor proofreading and stylistic lapses, give this volume, if you overlook the nice binding, the appearance of a low-budget conference proceeding. The volume represents a philosophical perspective on some topics in cognitive science, and most of the contributors are philosophers. The topics touched upon are related to knowledge, reference, meaning, categorization, and reasoning. As many of these topics are philosophical issues themselves, about five of the eleven chapters (Chapter 1, 4–7) end up being philosophical arguments about philosophical problems. Three other chapters (Chapters 2, 3, 8) consider some cognitive science research and examine them from a philosophical perspective. The last three chapters (Chapters 9–11) are more directly related to cognitive modeling – in particular with reasoning – and explore reasoning with diagrams, commonsense reasoning, and uncertain reasoning, respectively. Having made these general observations, let us now take a closer look at the contents of this volume. I will group the chapters as noted above, and consider each group in turn.
1. Philosophical Papers The papers in this group are concerned with the problem of meaning and reference. Alfonso Garcia Suarez in ‘Reference without sense: An examination of Putnam’s semantic theory’ (Chapter 6) argues that Putnam’s theory is really a variant of traditional semantic theory. Examining Putnam’s arguments against the classical Fregean semantics (Putnam, 1975), and his ‘division of linguistic labor’ and ‘indexicality’ theses, Suarez first argues that the division of linguistic labor is a very reasonable thesis to maintain, but is quite consistent with the Fregean semantics. Secondly, and more importantly, Suarez takes apart Putnam’s indexicality thesis to reveal that it commits us to a kind of externalism where the world is already readymade, complete with objective, mind-independent essences that are waiting out to be discovered. Of course, Putnam himself later developed what he refers to as ‘internal realism’ as a way of having the cake and eat it too (Putnam, 1981). It would have been much more interesting if Suarez had set his sights at Putnam’s internal realism but, unfortunately, he left it for ‘another day’. Juan J. Acero in ‘Attitudes , content and identity: A dynamic view’ (Chapter 7) takes on the classical Fregean puzzle of morning star and evening star, where different names have the same referent. Acero’s proposal is a dynamic belief system that allows the content of the previously believed sentences to change as new information is received. Thus, the proposal explicitly recognizes the context as playing a key role in determining the content of sentences. Indeed, this dynamism of belief
432
BOOK REVIEWS
systems has been duly recognized by cognitive scientists, and some interesting approaches have been developed to model it. Two notable ones are Dinsmore’s partitioned representation (Dinsmore, l99l) and the Sneps system developed at University at Buffalo by Shapiro, Rapaport and others (Shapiro and Rapaport, 1987 Martins & Shapiro, 1988; Shapiro and Rapaport, 1991). What is perhaps significant about Acero’s paper is the philosophical arguments to show that the dynamism of beliefs is compatible both with the post-Kripke-Putnam theory of reference and with externalism. Martin Davies’s ‘Externalism and Experience’ (Chapter l) is a slightly updatedand-revised version of Davies (1992). It concerns the content of a perceptual experience and whether arguments a la Putnam concerning externalism can be applied to it. Davies first argues that the perceptual content is distinct from the conceptual content of a belief or a judgement, and so the externalism of conceptual content cannot be automatically transferred to perceptual content. Then he articulates the externalist claim by contrasting it with individualist claim as proposed by Burge (1986) and refines it by distinguishing between the constitutive and the modal externalist claims. Finally, a variation of McGinn’s (1989) argument is used to show that externalism can be maintained for perceptual content. In ‘Skepticism, Lucid Content and the Metamental Loop’ (Chapter 4), Keith Lehrer concerns himself with epistemology from a cognitive point of view. Duly noting that personal justification is a necessary but not a sufficient condition for knowledge, Lehrer proposes that the truth condition of knowledge is derived from (1) the trustworthiness of the information, and (2) the trustworthiness of the mental operations that are applied to the information. In this way, the key to Lehrer’s cognitive epistemology is the correctness of the mental evaluation process that determines the trustworthiness of information, which is referred to as the metamental loop. In ‘Evading the Slingshot’ (Chapter 5) John Perry takes aim at the thesis that sentences designate (or refer to) truth values. The arguments against this thesis were actually made by Barwise and Perry quite some years ago (Barwise and Perry, 1981), and were some of the prime motivating factors behind the development of situation semantics. What is new in this chapter is Perry’s discussion of Ken Olson’s (1987) argument, namely that the situation semantics theory implies a ‘block universe’ view, so that one cannot individuate a single fact without bringing in the whole world. Perry’s tentative rejoinder is to propose an incremental version so that a situation determines some facts, given some other facts. The proposal is indeed tentative, for it is not fully developed. For example, it is not clear how the process would bootstrap itself.
2. Philosophy and Cognitive Science I now turn to the three chapters in this volume that fall right on the intersection of philosophy and cognitive science. William Ramsey in ‘Conceptual Analysis
BOOK REVIEWS
433
and the Connectionist Account of Concepts‘ (Chapter 2) considers the impact of cognitive science research on philosophy, by focusing on the implications of categorization research in cognitive modeling on the way analytic philosophy is done. He points out that analytic philosophy has mostly sought to find ‘simple’ definitions of abstract concepts that do not admit any intuitive counterexamples. But the categorization research, including Rosch’s prototype theory and connectionist modeling has shown that both these requirements cannot be satisfied simultaneously. We could either have simple category definitions that would admit some intuitive counterexamples; or, if we want to have watertight definitions, we must put up with category definitions that are complex, with many disjunctive clauses and several qualifications. In ‘Of Norms and Neurons’ (Chapter 3), the shortest paper in this volume (14 pages), Andy Clark applies philosophy to cognitive science. In particular, he concerns himself with the issue of whether internal structure of cognitive systems that display certain complex behavior patterns has any relevance in ascribing intentionality to the system. Retracting his own earlier response to this issue, where he maintained that internal structure has no relevance whatsoever, and only the overt behavior pattern matters, Clark now reexamines the problem by applying the generality and normative criteria to a variety of cognitive systems. He concludes that it is unreasonable to confer intentionality to cognitive systems based on giant look-up tables or on purely statistical example-based methods; yet, certain complex neural net systems, for example ones that include ‘skeletonization’ (Mozer and Smolensky, 1989), and non-representational systems a la Brooks (1987) do seem to satisfy these criteria. Peter Gärdenfors, in ‘Conceptual spaces as a basis for cognitive semantics’ (Chapter 8), contrasts realistic semantics with cognitive semantics. Whereas realistic semantics is concerned with truth conditions, cognitive semantics is concerned with meanings independent of truth conditions. He then argues for a topological framework for cognitive semantics and goes on to show how metaphor and prototype effects can be modeled in this framework. This work is essentially along the line of Fauconnier (1985) and Lakoff (1987), and a more extensive version of this paper appeared in Synthese at about the same time as the publication of this volume (Gärdenfors, 1996).
3. Formal Approaches to Various Types of Reasoning The final three papers in the volume are concerned with modeling different types of reasoning, something in which practicing cognitive science and Al researchers are interested. In ‘the cognitive impact of diagrams’ (Chapter 9), Keith Stenning considers diagrammatic reasoning. He confronts a sort of pessimism about cognitive value of diagrams in reasoning that can be traced back to Berkeley at the beginning of the eighteenth century, but has been more recently reincarnated by Pylyshyn (1973). The pessimism comes from noticing that any diagram overdetermines the
434
BOOK REVIEWS
concept it is supposed to instantiate, and so the cognitive agent needs to reintroduce abstraction over the information provided by any diagrammatic scheme. Stenning retorts that "The abstractions that can be reintroduced by conventions of interpretation are not all of the abstractions which have to be given up in resorting to diagrams" (p. 183, emphasis author’s). In fact, he argues that it is the specificity of the diagram schema that reduces the expressive complexity, thereby allowing for an efficient mode of reasoning. Therein lies the true cognitive value of diagrams. Kurt Konolige, in ‘What’s happening: Elements of commonsense causation’ (Chapter 10), outlines a model of causation that is designed to provide explanations and predictions by integrating causal and default reasoning. The model assumes that the primitive causation relation between propositions is given; as in ‘if the car lights are left on, the battery goes dead’. This, together with a correlation theory that captures normal conditions constitutes a ‘default causal net’. Using a simple example, Konolige explains how predictions can be made, and explanations and excuses can be derived for observable data. The last and the longest chapter of the volume (38 pages), ‘Principles of uncertain reasoning’ by Jeff Paris and Alena Vencovska (Chapter 11), is an overview of the mathematical framework of probabilistic reasoning. This chapter may seem overtly mathematical at a casual glance, but is in fact quite easy to read and only a rudimentary knowledge of probability theory is required. The authors have also provided ample motivation and intuitive explanations of mathematical definitions and theorems. 4. Summary The diversity of the subject matters, and the fact that most of the articles have a somewhat tentative nature and some seem to have been written in haste, prevent this volume from rising above the level of a conference proceeding. Of course, conference proceedings have the advantage that their speedy publication provides quick and broad access to work in progress. However, the five-year delay between this conference and the publication of this volume has all but wiped out this advantage. It can be plainly evidenced from the fact that more thorough and comprehensive versions of many chapters have been published elsewhere and pre-date this volume. So, for example, Gardenfors (1996) might be a better source than Chapter 8, and Stenning and Oberlander (1994, 1995) might be better than Chapter 7. Extensive post-conference editing of the papers to provide some common themes connecting different papers might have added some value to the collection, but this, unfortunately, does not seem to have happened here. References Barwise, J. and Perry, J. (1981), ‘Semantic innocence and uncompromising situations’, Midwest Studies in Philosophy VI, pp. 387–403.
BOOK REVIEWS
435
Brooks, R. (1987), ‘Planning is just a way of avoiding figuring out what to do next’, MIT Artificial Intelligence Laboratory Working Paper 103. Burge, T. (l986), ‘Individualism and psychology’, Philosophical Review 95, pp. 3–45. Davies, M. (1992), ‘Perceptual content and local supervenience’, Proceedings of the Aristotelian Society 92, pp. 21–45. Dinsmore, J. (1991), Partitioned representations, Dordrecht: Kluwer Academic Publishers. Fauconnier, G. (1985), Mental spaces, Cambridge, MA: MIT Press. Lakoff, G. (1987), Women, fire and dangerous things, Chicago: Univ. of Chicago Press. Gärdenfors, P. (1996), ‘Mental representations, conceptual spaces and metaphors’, Synthese 106, pp. 21–47. Martins, J.P. and Shapiro, S.C. (1988), ‘A model for belief revision’, Artificial Intelligence 35, pp. 25–79. McGinn, C. (1989), Mental Content, Oxford: Basil Blackwell. Mozer, M. and Smolensky, P. (1989), ‘Using relevance to reduce network size automatically’, Connection Science 1, pp. 3–17. Olson, K. (1987), An essay on facts. Stanford: CSLI/Univ. of Chicago Press. Putnam, H. (1975), Mind, language and reality: Philosophical papers, Vol. 2, Cambridge: Cambridge Univ. Press. Putnam, H. (1981), Reason, truth and history, Cambridge: Cambridge University Press. Pylyshyn, Z.W. (1973), ‘What the mind’s eye tells the mind’s brain’, Psychological Bulletin 80, pp. 1–24. Shapiro, S.C. and Rapaport, W.J. (1987), ‘SnePS considered as a fully intensional propositional semantic network’, in N. Cercone and G. McCalla (eds.) The knowledge frontier: Essays in the representation of knowledge, New York: Springer-Verlag, pp. 262–315. Shapiro, S.C. and Rapaport, W.J. (1991), ‘Models and Minds: Knowledge representation for naturallanguage competence’, in R. Cummins and J. Pollock (eds.) Philosophy and AI: Essays at the interface, Cambridge: MIT Press, pp. 215–259. Stenning, K. and Oberlander, J. (1994), ‘Spatial inclusion as an analogy for set membership: a case study of analogy at work’, in K. Holyoak and J. Barnden (eds.) Analogical Connections, Hillsdale, N.J.: Erlbaum, pp. 446–486. Stenning, K. and Oberlander, J. (1995), ‘A cognitive theory of graphical and linguistic reasoning: logic and implementation’, Cognitive Science 19, pp. 97–140.
Department of Computer, Information and Communication Sciences, Tokyo University of Agriculture and Technology, 2-24-16 Nakacho, Koganei, Tokyo 184-8588, Japan E-mail:
[email protected]
BIPIN INDURKHYA
Mark H. Bickhard and Loren Terveen, Foundational Issues in Artificial Intelligence and Cognitive Science: Impasse and Solution, Advances in Psychology, Vol. 109, Amsterdam: North-Holland/Elsevier Science B.V., 1995, ix + 384 pp., $152.50 (cloth), ISBN 0-444-82048-5; xii + 384 pp., $92.00 (paper), ISBN 0-444-82520-7. This book revitalizes dialogue regarding the status of computational representations of knowledge in a theoretical account of cognition, displacing the grounding of symbols as a secondary concern relative to the content and function of representation. Inspired by developmental psychology, Bickhard and Terveen focus
