August 19, 1861, a cottage in Galloway, Scotland.

The young Clerk Maxwell is sitting in a garden deep in thought. On the table before him there is a sheet of paper on which he sketches various pictures of lines and circles and writes equations.

What is he thinking? Is he trying to cook up a model to go with the equations he has derived already by induction from the experimental data and electrical considerations alone? Is he concerned that his mathematical results are not quite right and so is thinking how to fudge his analysis to make it look right...

The new naturalism in science studies recognizes that people learn by active intervention in a world of objects and other people. Philosophy of science lacks resources to deal with new notions of reasoning and empirical access implied by the new image of scientific practice. In this chapter I look at the procedural stage-setting that scientists use to get concepts and phenomena off the laboratory bench and into their language and experience. It may be true, as Quine pointed out, that “most of the things, and most of the supposed traits of the so-called world, are learned through language”; the problem...

The very existence of this book suggests that the “cognitive turn” in the philosophy of science is already at an advanced stage. Indeed, not since the seminal work of Kuhn (1962) has there been more activity among philosophers directed at what we can loosely characterize as the “processes” of science, rather than at its “products.” In the present essay, I will seek to clarify some aspects of the cognitive turn. In particular, I hope to show that the oft-cited distinction between parallel and serial processes needs considerably more careful attention in its application to science than has so far been...

The contributors to this volume were charged to explore how research in cognitive science bears on issues discussed in the literature on the philosophy of science. Most took this as a challenge to show how results from cognitive psychology or artificial intelligence inform theories of the processes of theory development and theory choice. I focus on a different issue — the origin of scientific concepts.

Let me begin by settling some terminological matters. Byconcept, belief, andtheory, I refer to aspects of mental representations. Concepts are units of mental representation, roughly the grain of single lexical items, such as...

The simple working definition adopted in this essay for conceptual change is that it refers primarily to the notion of how a concept can change its meaning. Since a difference in meaning is difficult to define, one can think of it as a change in its categorical status: That is, changing the category to which the concept is assigned, since all concepts belong to a category. To assume that all concepts belong to a category is quite standard: For instance, A.R. White (1975) also assumed that a concept signifies a way of classifying something. Thus, this essay is not concerned...

The cognitive revolution in philosophy of science is well under way, and one point of this volume is to plan an agenda for the future. I believe that part of that agenda should be to rewrite the history of philosophy of science, for no revolution is complete until it has created an appropriate history of its own inevitability and rectitude.

Some kind of observation/theoretical distinction played a large part in the various positivist approaches to philosophy of science, but was swept away by the historicist revolution. Formal studies of foundations of measurement that had never quite connected with other aspects...

Foundationalist empiricism — foundationalism, briefly — is the view that knowledge¹ is based solely on the data of sensory cognition: perception, sensation, stimulation, etc.² Today, after decades of criticism, traditional versions of this view are virtually without defenders. It is ironic that many of the critics — notably Quine (1951, 1969) — have argued that epistemology should be naturalized and made scientific; for, as this essay will contend, when epistemology is naturalized and made scientific it suggests new interpretations of foundationalism that appear to be viable. My thesis is that a naturalized version of weak foundationalism is the proper framework...

A major concern in philosophy of science has been to provide norms to scientists that will assist them in their work. Ideally guidance would be provided at every step of the scientific process, but efforts to develop norms are predominantly restricted to thecontext of justification, where a scientist is attempting to prove that a hypothesis is true, and are seldom proposed for thecontext of discovery, where a new hypothesis is developed. This bias originated with the logical positivists, who argued that discovery was an inexplicable process that did not follow any logic or pattern, and therefore could not...

Understanding the growth of scientific knowledge has been one of the major tasks in philosophy of science in the last thirty years. No successful general model of scientific change has been found; attempts were made by, for example, Kuhn (1970), Toulmin (1972), Lakatos (1970), and Laudan (1977). A new approach is to view science as a problem-solving enterprise. The goal is to find both general and domain-specific heuristics (reasoning strategies) for problem solving. Such heuristics produce plausible, but not infallible, results (Nickles 1987; Thagard 1988).

Viewing science as a problem-solving enterprise and scientific reasoning as a special form of problem...

We apply in this essay a computational theory of explanatory coherence to an important case in the history of astronomy. The theory has been implemented in a connectionist computer program called ECHO that has been used to model the competition between Copernican and Ptolemaic astronomy. ECHO has also been used to model the acceptance of hypotheses in chemistry, biology, geology, and legal reasoning (Thagard 1989; in press; Thagard and Nowak 1988; 1990). The application reported in this essay is interesting both because of its historical importance and also because it is by far the largest application of ECHO to date,...

Computational approaches to scientific reasoning have traditionally focused on attempts to simulate scientific discoveries, such as Ohm’s Law (Langley et al. 1987), oxygen (Thagard 1989a), and the wave theory of sound (Thagard and Holyoak 1985), on the computer. Based on these findings, Slezak (1989) has argued that such simulations of scientific discovery refute the basic assumption of the strong programme within the sociology of scientific knowledge that scientific discoveries occur within a social context. While the computational approach to science may not refute the strong programme, it does serve to elucidate the role of cognitive processes in science. The computational...

By the late 1960s, every good materialist expected that epistemological theory would one day make explanatory contact, perhaps even a reductive contact, with a proper theory of brain function. Not even the most optimistic of us, however, expected this to happen in less than fifty years, and most would have guessed a great deal longer. And yet the time has arrived. Experimental neuroscience has revealed enough of the brain’s microphysical organization, and mathematical analysis and computer simulation have revealed enough of its functional significance, that we can now address epistemological issues directly. Indeed, we are in a position to reconstruct,...

One major task for psychologists interested in developing the psychology of science is to engage in fruitful dialogue and debate with others in the broad field of science studies (Houts 1989). For at least the last two decades, this interdisciplinary field has been dominated by philosophers, historians, and sociologists. With a few recent exceptions (Fuller 1989; Giere 1988), traditional practitioners of these latter disciplines have consistently opposed incorporation of psychological perspectives into the field of science studies. For example, philosophers of science from Carnap (1936, 1937) to Popper (1970, 1974) and from Lakatos (1970, 1973) to Laudan (1977) have sought...

Philosophers of science have made a bewildering variety of recommendations concerning how scientists ought to behave. Popper (1972), for example, tells them to falsify; Kuhn (1970) says that is good advice only during a crisis, but that the rest of the time they should do research within a paradigm or set of exemplars; Feyerabend (1975) tells them to do whatever they want. One could go on. The point is, how does one assess which (if any) of these philosophical prescriptions is right?

One answer would be to simply analyze the way scientists behave, deriving general principles of good conduct. Lakatos’s...

I argue that a radically naturalized study of knowledge would apply the methods and findings of psychology and the social sciences to itself. The results would make epistemology more robustly normative, experimental, and social than it currently is, indeed, more in the spirit of the original American naturalizers of epistemology: Peirce, Dewey, and Mead. I begin by briefly sketching how my own philosophical program, social epistemology, captures this spirit. I then proceed to show the limits to the radicalism of various naturalizers: Churchland’s eliminative materialism, mainstream experimental psychology, analytic epistemology, and even the sociological school of ethnomethodology. In the last...

Ten years ago I hoped, even expected, that the computational revolution would revive a dying enterprise, philosophy of science, and make it more intelligent and rigorous and insightful and interesting. This book and the recent work that seems to have provoked it have convinced me that my expectations were completely wrong. To judge by this sample, wherever the discipline of philosophy of science has been touched by cognitive science the result has been a zombie—philosophy of science killed dead and brought back to ghoulish, mindless, pseudolife.

“That grumpy guy, at it again,” you say. Well, pilgrim, look at what...

I think Glymour is right to be upset. The old epistemological ways are dying — cause enough for distress. Worse, their prospective replacements are still rag-tag, unpolished, and strange. Moreover, there is an unsettling uncertainty or shift of focus as to which are the truly important problems to be addressed. But even so, the objective theoretical situation holds more promise than at any time since the 1940s, and in his frustration Glymour misrepresents it rather badly. And sometimes carelessly. My aim in this note is to address and correct some of his complaints. Several of them are welcome, in that...

The overall aim ofExplaining Science(1988) is to develop a “naturalistic” approach to understanding the nature of modern science that mediates between existing rival philosophical and sociological approaches. I reject philosophical attempts to explain science as the result of applying categorical principles of rationality, whether these be explicated in terms of formal logic, probability, or historical progress. For me, the only form of rationality that exists is the instrumental use of empirically sanctioned strategies to achieve recognized goals.¹ I also reject constructivist sociological approaches that portray scientific (theoretical) claims as arising out of social processes that allow these claims...

The purpose of this note is to remedy some of Clark Glymour’s (this volume) misconceptions about explanatory coherence and ECHO. After briefly responding to Glymour’s challenges concerning the normativeness and necessity of ECHO and the nature of analogy and explanation, I shall show that his alternative approach to computing coherence is utterly inadequate.

First let us look at some of the issues Glymour raises about explanatory coherence. Is my theory descriptive or normative? This question is more complex than Glymour appreciates, since descriptions of scientific cases can have normative force. I have elsewhere discussed the general question of the relation...