Abstract

I examine the computational functionalist account of mental states presupposed in the field of artificial intelligence , paying particular attention to their inherently subjective character. I begin by discussing the differences between computational functionalism and other forms of functionalism. In Chapter II, I critically examine several criteria proposed by AI researchers for determining when mental states can be attributed to a computer. I then defend John Searle's "Chinese Room" argument against computational functionalism by arguing that an intentional system must be able to understand a language, and genuine understanding of language requires qualitative experience. In Chapter III, I argue that computational functionalism is inadequate for qualitative states, and that because of this, it is inadequate for intentional states. My argument is based on a holistic account of intentional states, together with the notion that there are many intentional states, such as qualitative beliefs, that one can have only if one has certain qualitative states. Finally, in Chapter IV, I develop a version of the property objection to both computational functionalism and the token-identity theory, based on the phenomenon of first-person authority. This argument involves me in an extended defense of the view that de re attitudes are not problematic. I also defend the "argument from knowledge" advanced by Thomas Nagel and Frank Jackson, an argument that applies to materialism as well as to computational functionalism.