Cognitive Science and the Failure to Allow for Error and Misrepresentation

Abstract

The basic outlook of cognitive science is that through empirical research we can discover the functional architecture of the mind/brain. It is anti-verificationist in approach: just as we can have indirect evidence about the real nature of the world, even of unobservable particles, we can use psychological and neurophysiological evidence, to establish how the mind works. On such a view, we are required to be able to give a naturalistic account of mental content to allow it a place in the explanation of how the mind works. The going theory of content in cognitive science is a functionalist one: the meaning of a representation is a function of its role in the cognitive system. By another name this is conceptual role semantics, but it is the same basic view as what I have been calling a use theory of meaning of mental representations. The question for us is: Do we get the same rejection of error without the verificationism that leads us to it in the way described in the previous chapter? How and why would this result be obtained? To explain this, I will first sketch a general argument that applies to all forms of the realist, cognitive science side of theories of content. Then I will go into detail about how the acknowledged ‘conceptual role theories’ attempt to establish wide content, and with it at least some misrepresentation (i.e. Twin-misrepresentation). All of these theories are two-factor or dual-aspect conceptual role theories, and we will see that they all fail due to a fundamental structural flaw. Then we will look at causal, informational, and adaptational role theories of content. I call these ‘disguised use theories’, since they often proclaim not to be use theories, but they are in fact use theories, which include only a restricted class of uses to be meaning-fixing. We will look at their attempts to allow for misrepresentation, and see why they fail.