Daniel Gilman

Of the many controversial claims in Thomas Kuhn’s The Structure of Scientific Revolutions perhaps none are more troublesome than those made in his account of the role of a paradigm in perception. For if we take it that “a paradigm is prerequisite to perception itself (Kuhn 1970, p. 113) and that “two groups of scientists see different things when they look from the same point in the same direction” (Kuhn 1970, p. 150) then we seem to be burdened with all the familiar problems about whether, and how, experimental evidence could possibly serve to promote the acceptance of a paradigm. And, indeed, these problems have been amply discussed in the last several decades. Dudley Shapere, for instance, has questioned Kuhn’s account of how science comes to move between incommensurable paradigms. (Shapere 1964) Still, Kuhn’s discussion of perception has an empirical foundation in psychological research which gives it a certain resilience in the face of surface problems with the consequences of his account.

David Marr's theory of vision has been a rich source of inspiration, fascination and confusion. I will suggest that some of this confusion can be traced to discrepancies between the way Marr developed his theory in practice and the way he suggested such a theory ought to be developed in his explicit metatheoretical remarks. I will address claims that Marr's theory may be seen as an optimizing theory, along with the attendant suggestion that optimizing assumptions may be inappropriate for cognitive mechanisms just as anti-adaptationists have argued they are inappropriate for other physiological mechanisms. I will discuss the nature of optimizing assumptions and theories. Considering various difficulties in identifying and assessing optimizing assumptions, I will suggest that Marr's theory is not purely an optimizing theory and that reaction to Marr on this issue prompts interesting considerations for the development of inter-disciplinary constraints in the cognitive and brain sciences.

The dissertation sketches a solution to the problem of pictorial representation. By appealing to the visual system as an information processing system, we understand how it is that certain sorts of pictures are seen as representing their subjects. ;The first chapter introduces the problem and discusses existing philosophical treatment of pictorial representation. Conventionalist arguments against the possibility of a naturalist account are refuted, thus clearing the way for a naturalist, realist, "resemblance" view of pictorial representation. ;The second chapter discusses the notion of the modularity of mind. Aspects of perceptual holism are criticized. I support the modularist claim that one can make a useful distinction between observation and inference without suffering familiar positivist problems. I suggest that much of vision can fruitfully be studied in isolation from other mental processes. ;The third and fourth chapters provide a contemporary account of human vision. A model of the neural circuitry of vision is provided. Similar treatment is given to mathematical modeling of spatial vision. Also, some attention is paid to lower level biological details and to higher level functional attributes of the system. Research methodologies are discussed, as are the connections between the various levels of investigation. ;The concluding chapter begins with a theoretical and methodological comparison of experimental work cited as empirical foundation for various forms of conventionalism with that supporting modularity. The questions of Chapter I are then reconsidered in light of the vision science of Chapters III and IV. It is shown that current understanding of vision elucidates many of the problems of pictorial representation. Especially relevant is the observation that human spatial vision functions by abstracting certain features from the environment, features that are preserved in certain sorts of pictures. It turns out that many of the conventionalist claims about mind, vision and pictorial representation are unsupportable, given current understanding of mind and vision.

Various claims for theory-laden perception have involved empirical as well as conceptual considerations. Thomas Kuhn cites New Look psychological research in discussing the role of a paradigm in perception (1970) and Paul Churchland (1988) appeals to biological evidence, as well as New Look sources similar to Kuhn's. This paper offers a critical examination of the empirical evidence cited by Kuhn and Churchland, including a look at the underlying experimental work. It also offers a comment on the application of such evidence in a naturalized epistemology.

The dissertation sketches a solution to the problem of pictorial representation. By appealing to the visual system as an information processing system, we understand how it is that certain sorts of pictures are seen as representing their subjects. ;The first chapter introduces the problem and discusses existing philosophical treatment of pictorial representation. Conventionalist arguments against the possibility of a naturalist account are refuted, thus clearing the way for a naturalist, realist, "resemblance" view of pictorial representation. ;The second chapter discusses the notion of the modularity of mind. Aspects of perceptual holism are criticized. I support the modularist claim that one can make a useful distinction between observation and inference without suffering familiar positivist problems. I suggest that much of vision can fruitfully be studied in isolation from other mental processes. ;The third and fourth chapters provide a contemporary account of human vision. A model of the neural circuitry of vision is provided. Similar treatment is given to mathematical modeling of spatial vision . Also, some attention is paid to lower level biological details and to higher level functional attributes of the system. Research methodologies are discussed, as are the connections between the various levels of investigation. ;The concluding chapter begins with a theoretical and methodological comparison of experimental work cited as empirical foundation for various forms of conventionalism with that supporting modularity . The questions of Chapter I are then reconsidered in light of the vision science of Chapters III and IV. It is shown that current understanding of vision elucidates many of the problems of pictorial representation. Especially relevant is the observation that human spatial vision functions by abstracting certain features from the environment, features that are preserved in certain sorts of pictures. It turns out that many of the conventionalist claims about mind, vision and pictorial representation are unsupportable, given current understanding of mind and vision

A large body of research in computational vision science stems from the pioneering work of David Marr. Recently, Patricia Kitcher and others have criticized this work as depending upon optimizing assumptions, assumptions which are held to be inappropriate for evolved cognitive mechanisms just as anti-adaptationists (e.g., Lewontin and Gould) have argued they are inappropriate for other evolved physiological mechanisms. The paper discusses the criticism and suggests that it is, in part, misdirected. It is further suggested that the criticism leads to interesting questions about how one formulates constraints--across "levels of organization" and disciplinary boundaries--on one's models of complex systems, such as human vision.

Criticism of the observation/theory distinction generally supposes it to be an empirical fact that even the most basic human perception is heavily theory-laden. I offer critical examination of experimental evidence cited by Thomas Kuhn and Paul Churchland on behalf of this supposition. I argue that the empirical evidence cited is inadequate support for the claims in question. I further argue that we have empirical grounds for claiming that the Kuhnian discussion of perception is developed within an inadequate conceptual framework and that a version of the observation/theory distinction is indeed tenable. The connection between cognitive science and epistemology is also discussed.

Paul Churchland has recently argued that empirical evidence strongly suggests that perception is penetrable to the beliefs or theories held by individual perceivers (1988). While there has been much discussion of the sorts of psychological cases he presents, little has been said about his arguments from neurology. I offer a critical examination of his claim that certain efferents in the brain are evidence against perceptual encapsulation. I argue that his neurological evidence is inadequate to his philosophical goals, both by itself and taken in concert with his psychological evidence