PREFACE

This book was a long time in the making. There are parts of it that date back to about 1987 and other parts that are very recent indeed. It started out as an attempt to bring my own peculiar philosophical background (which is an unusual one in cognitive science circles) into contact with what was at that time (and to some extent still is) the mainline view of the mind in analytic philosophy of mind: the Computational Theory of Mind (CTM for short). I came to the study of cognitive science with three kinds of relevant background, each of which is at least a little bit off center with respect to the contemporary scene in the philosophy of mind. Perhaps the most prosaic of these is that I worked much of my way through graduate school teaching courses in computer programming, software design, and artificial intelligence. It would be a most heinous exaggeration if I were to describe myself as either a computer scientist or a hacker, but I learned some theory, did some programming, and made my way through most of the three-volume History of Artificial Intelligence with a class of undergraduate students. I knew computers in theory and in practice before I began to think about them as a philosopher.

This, however, was not my first exposure to computer models of the mind. I had studied as an undergraduate with Stephen Grossberg of Boston University, one of the few people doing continuous work in neural network models from the 1960s up until the present, even through the two decades when it was not a particularly popular thing to be doing. So whereas most people in the philosophy of cognitive science came to cognitive science by way of the symbol-processing paradigm embracing

Turing, Minsky, Newell, Colby, Winograd, and Marr (to name but a few), I cut my teeth on neural network models. In a way, I have been in exactly the opposite position of most philosophers doing cognitive science in recent years: whereas they have had to learn about the "new" neural network paradigm (which is in fact as old as the symbol-processing paradigm), I had to do exactly the opposite in the late 1980s.

The third element in my background, of course, was my philosophical training. My first philosophical love was speech act theory, on which I wrote an undergraduate thesis in 1981 with Bruce Fraser. The major philosophical writers on the subject at that juncture were Grice, Austin, Strawson, Vendler, Searle, Bach, and Harnish. With the possible exception of the last two, these philosophers practiced their trade in a style markedly different from that commonly found in cognitive science today. At the same time, I was beginning to read some of the works of Edmund Husserl, which was to have a profound influence on how I came to view philosophy. In spite of praise from people like Chisholm, Sellars, Dreyfus, Føllesdal, Haugeland, and (more recently) Putnam, Husserl is not adequately appreciated among American analytic philosophers. Much of what has transpired since his day in philosophy of language is already present in the first "Logical Investigation," and no one was more keenly aware than he of the difficulties and pitfalls of coming to a philosophical understanding of the mind. Husserl's focus on the centrality of intentionality made what seems to have been a permanent impression on me. Somewhere in 1982 I became convinced that the study of speech acts could not progress further without a study of intentionality. (It was gratifying to see a year later that John Searle had come to a similar conclusion.) That focus became central to my philosophical thinking for the next ten years, and it still occupies an important (though no longer central) place for me in the scheme of important philosophical problems.

As a graduate student, I worked with Kenneth Sayre at Notre Dame, one of the first philosophers to write about artificial intelligence in the early 1960s, and a longtime proponent of an alternative vision of the mind centering around the Mathematical Theory of Communication articulated by Shannon and Weaver. It was, in fact, only after I started working with Sayre that I began to read what most people consider "mainstream" artificial intelligence and philosophy of cognitive science, so the symbol-processing paradigm was actually the third paradigm I was exposed to in cognitive science. Along the way in my philosophical studies I felt some influence from the writings of Plato, Aristotle, Kant, and Wittgenstein.

In short, I came to the contemporary scene in cognitive science with a very different list of philosophical and scientific heroes from those of most of my colleagues in the field; and much of what I found in the "mainstream" initially struck me as eminently wrongheaded. But it is one thing to think that something is wrongheaded; it is of course quite another to understand why people would believe it and to identify just where you think the problem lies. This book is in large measure the product of a long process of trying to do two things: first, to understand the project from the inside, as it were, in terms that its own advocates would embrace; and, second, to articulate what seem to me its major flaws in a fashion that does not depend too much upon an alternative philosophical viewpoint and which might be accessible to someone who does not share my own philosophical leanings.

As a result, the first two chapters of this book attempt to lay out computationalism in its historical context and to explain to the reader why one might very sensibly think that it is offering some tempting philosophical fruit. At the same time, I have tried to emphasize elements in the historical context and connections with other strands of philosophical psychology that seem important yet are often passed over by those who consider themselves to be within the computationalist camp. I hope that these chapters will serve as a good introduction to the computational theory for a wide philosophical audience. I suspect that they may also prove useful for the initiate who wishes to read the critical sections of the book, as they attempt to lay out computationalism with more exactitude than is normally done and with a minimum of rhetoric. In a sense, the moral of the book is just this: if you are not extremely careful about how you use words like 'computer', 'symbol', 'syntax', and 'meaning', you are likely to stumble into some pernicious confusions about computation and the mind and to be tempted by some subtly fallacious arguments that seem to deliver philosophical results but in fact mislead.

The rest of the book grew gradually. It started as a purely critical project of debunking claims that CTM provides an account of the intentionality of the mental and a "vindication" of realism about mental states. Once I had satisfied myself that I had proven my case there to my own satisfaction, I began to be more interested in what could be said in a positive way about the importance of computational psychology as a way of understanding the mind, and how far away from the views articulated by writers like Jerry Fodor and Zenon Pylyshyn one would have to go in order to do so. Unlike some critics of CTM, I do not believe that empirical research in cognitive science stands or falls with the philo-

sophical claims of CTM. The several chapters of the book that explore alternative ways of looking at computation and the mind were first drafted at the National Endowment for the Humanities Summer Seminar on Mental Representation held at the University of Arizona in the Summer of 1991. Rob Cummins, who directed that institute, has gone far beyond the call of duty by reading three separate drafts of the book in the three years since that time. Chapter 6, which contains some of the most provocative material in the entire book, was written after the rest of the manuscript to respond to criticisms that Rob raised about an earlier draft. Rob has my undying gratitude for his responses along the way, and not least of all for occasionally admitting that I had convinced him about something. Thanks to him and to the NEH, which helped finance that extremely productive summer.

Major thanks also go to Ken Sayre, who forced me to take CTM seriously and on its own terms, and forced me to a higher standard of clarity and exactitude than I might otherwise have attained. Thanks to Ken also for the useful way he engaged in the process of helping me hammer out views that were strongly related to his own yet contrary to his own formulations.

Jay Garfield of Hampshire College read and commented upon the entire manuscript very late in the game, made some very supportive comments, and also made a number of very important suggestions that have ultimately made the final product a much better and more readable book than it otherwise might have been. The chapter on naturalization in particular is much expanded as a result of Jay's (deservedly) pitiless attack upon a former incarnation of the same. (I fear it still does not meet with his entire approval—the credit for its expansion is his, any residual faults are my own.)

Richard DeWitt of Fairfield University also went beyond the call of duty in reading multiple drafts of four or five chapters, and has been supportive of the project via numerous e-mail exchanges since we met at Cummins's NEH seminar. Likewise, my Wesleyan colleague Sanford Shieh made some very helpful suggestions on the chapters in Part II, and probably saved me from some grave embarrassments in my use of terms that had technical meanings in logic of which I was blessedly unaware.

Many other people read or commented on all of part of the manuscript along the way. All of the following people were at least so kind as to agree to read parts of the manuscript for me somewhere along the line. Many made important contributions to the present form of the work by their comments and criticisms: Michael Anderson, Robert Audi, Lynne

Rudder Baker, David Burrell, Hubert Dreyfus, Aaron Edidin, Brian Fay, Tim Fischer, Pat Franken, Bruce Fraser, Heather Gert, Ruth Ginzberg, Victor Gourevitch, Robert Losonsky, Vaughn McKim, Chris Menzel, Mark Moes, Hans Müller, Shelly Park, Bill Ramsey, Bill Robinson, and Joe Rouse.

Finally, I should like to thank the many people whose love and friendship over the years have made it possible for me to pursue something as demanding as a book in philosophy. In particular, I wish to thank my parents, who have provided ample support throughout my life, and who have been in my corner for many years while I worked on a project whose merits they could only take on faith. Plato somewhere describes intellectual creations as a kind of progeny. I hope that they will be pleased with their first grandchild.

MIDDLETOWN, CONNECTICUT
OCTOBER 1994