The Ontology of Concepts—Abstract Objects or Mental Representations? 1

objects as well, but they are still to be distinguished from senses in that functions are individuated extensionally: if function f and function g have the same extension, then f = g. 9 An anonymous referee has suggested that a different way to treat concepts as abstract objects would be to adopt Wayne Davis’s view that concepts are mental representation types (Davis 2003, 2005). Although types are certainly abstract entities, the suggestion threatens to eliminate the fundamental difference between the Psychological View and the Semantic View, viz., the disagreement about whether concepts have tokens (see note 7). On our taxonomy, Davis’s view turns out to be a version of the Psychological View, not the Semantic View. 10 A different option is to say that there really is no substantive dispute here at all, only the terminological matter of which things to call “concepts”. Perhaps we should just introduce two terms—concept1 (for mental representations) and concept2 (for senses)—and be done with it. While we can’t entirely rule out this interpretation, we do think it would be a mistake to trivialize the differences between the Psychological View and the Semantic View. As we mention in the text, some of their proponents certainly don’t see the matter as a merely terminological one; they argue quite vigorously against the opposing position. In addition, one ought to be hesitant about claiming that the dispute is merely terminological for the simple reason that any substantive dispute can be recast as a terminological one. For example, we can call consequentialist goodness good1 and deontological goodness good2, or compatibilitist freedom freewill1 and incompatibilist freedom freewill2, and then go on to pronounce that philosophers on different sides of the original disputes were just talking about different things. But such pronouncements clearly don’t resovle the original controversies. Complex issues remain concerning whether we need either or both of the theoretical kinds identified and which on balance best answers the explanatory concerns that preceded the terminological innovations. 11 These same considerations argue against Michael Dummett’s case against concepts being mental representations, which he often discusses in the context of rejecting the “code theory of language” (viz., the view that words inherit their content from the mental representations with which they are associated). “If this were the true account, communication would indeed rest on faith. Since it obviously cannot rest of faith, there must be an adequate outward manifestation of understanding. . .” (Dummett 1989, p. 187). The problem with this argument is that it isn’t obvious that communication requires anything more than reliable correspondences between language and thought. If that’s to say that communication rests on faith, then so be it. What’s more, we’d suggest that one ought to be deeply suspicious of any theory of language that does what Dummett seems to want it to do, viz., entirely preclude the possibility of miscommunication. Although it is very unlikely that people massively and systematically fail to communicate, it does seem (to us anyway) that miscommunication is a fact of life. 12 Like Frege, Peacocke seizes on the idea that every instance of a mental representation (i.e., every token) occurs in someone’s mind. While this is true, there is nonetheless, no more reason to think that every mental representation type must be instantiated than there is to suppose that every property must be instantiated. 13 For sake of argument, we are assuming that one and the same image can serve as a representation of Paris and a representation of France. But the mere fact that a single image is associated in introspection with two distinct thoughts shows very little. It might be that the