V. Lot Term Introduction and the Content of Learned Terms

Robert Cummins argues that any causal theory of mental content (CT) founders on an established fact of human psychology: that theory mediates sensory detection. He concludes, " CT is false in all its forms. " 1 Cummins takes his conclusion to follow from general characteristics of human sensory systems and CT; thus, he does not discuss in detail the nature of the theories that mediate sensory detection or the nature of the mental representations that, according to CT, are tokened as the result of humans' sensory interactions with the world. I contend that the details matter: when we fully appreciate the range of ways in which mental content and the vehicles of mental content may plausibly emerge in the human cognitive system, CT appears, after all, to be consistent with theory's mediating role in human perception. I. Cummins's Argument against CT Generally speaking, CT asserts that the content of a primitive mental representation, or term in a language of thought (LOT), 2 is fixed by its causal relation to kinds or their individual members (or properties or their instantiations; assume this qualification throughout). Proponents of CT have formulated a variety of complex theories of content, each theory specifying the content-fixing causal relation in its own way. 3 Given the variety of such approaches, it is not entirely clear how to properly circumscribe causal theories, or to what extent there is a substantive philosophical division waiting to be made between true causal theories and the rest. This is, however, largely a side-issue. Cummins's critique takes aim at one specific feature shared by many theories of content, a feature Cummins places at center stage when describing CT: 2 Cummins defines CT as " the doctrine that the contents of the semantic primitives in the human scheme of mental representation are determined by their role in detection, " (535) where, in particular, the determination of content rests on the subject's ability to reliably detect members of the kind in question. Thus, whether we classify a given theory of content as a causal theory matters little for present purposes: Cummins's critique of CT simply takes as its target any theory for which reliable detection is a necessary condition of correct content assignments to human mental representations; the literature contains enough theories of this sort to merit careful and critical consideration of Cummins's argument. Cummins claims that the mediating role played by theory in human …