Abstract

In this work I explore the question of whether visual perception produces knowledge, or correct representations, of the external world. I argue that it does not, and that the way the world looks is rather a function of the properties of perceivers. I also argue, however, that it is not necessary for perceivers to have correct representations of the environment. The common sense view that the purpose of vision is to make acquaintance with the environment possible is mistaken. This conception of the purpose of vision can and ought to be recast, and the commitment to visual knowledge abandoned. ;The argument is set against the background of the computational theory of vision which is currently the central paradigm in vision research and a powerful tool for philosophers who wish to defend the view that vision does lead to acquaintance with the environment. I propose that even as powerful a tool as computational vision fails to support the claim that visual experience correctly represents the properties of the external world. I develop two lines of argument. First, I argue that there are no external colors and therefore that color appearances do not correctly represent the environment. I further claim that a parallel view can be defended for the property of shape. It follows, therefore, that vision systematically misrepresents the environment. The second line of argument is that whatever the environment is like, vision probably does not have the ability to represent it correctly. ;I develop an alternative to common sense and its computational extension which I call neural subjectivism. Neural subjectivism claims that visual appearances are not representations of the environment but adaptive neural responses to the regularities in the light stimulus which the sighted organism can use to guide its action. Thus, although visual experience fails to reveal the environment to the perceiver, it nonetheless can perform the crucial function of facilitating successful behavior in the environment. In a naturalistic framework, this, and not knowledge, ought to be the central concept.