Singular Thought and the Cartesian Theory of Mind


Kirk Ludwig




In this paper, I defend some elements of a traditional picture of the fundamental nature of representational states, which I call the ‘Cartesian Theory of Mind’.  The theory consists of the following five propositions:

(1)        Content properties are non-relational, that is, having a content property does not entail the existence of any contingent object not identical to the thinker or a part of the thinker.

(2)        We have non-inferential knowledge of our conscious thoughts, that is, for any conscious thought of the form ‘T(p)’ where ‘T’ represents its mode and ‘p’ its content, we have non-inferential knowledge that we have a thought in that mode with that content.

(3)        The content of a thought determines its truth conditions.

(4)        We (can) refer and know that we (can) refer to contingent particulars.

(5)        Content properties are causally relevant to our behavior, non-intentionally and non-relationally described.

I think all of these propositions strike us as plausible when we first begin to think about the nature of thought and content properties.  But difficulties arise when we conjoin these propositions with a number of others that also seem very plausible.  In the following, I first develop three problems for the Cartesian view by adding to (1)-(5) a number of plausible semantic, epistemic, and metaphysical theses (in particular, the now widely accepted thesis that proper names, indexicals, and demonstratives are directly referring terms), from which together with (3) we can derive the negations of (1), (2) and (5).  Among the consequences of these theses will be that there are singular thoughts, i.e., thoughts whose contents cannot be characterized completely independently of reference to some particular object.  Independently of the above difficulties, admitting singular thoughts raises a difficulty about how (2) could be true, because it is puzzling what knowledge, particularly non-inferential knowledge, of the content of a thought which involves a particular object could come to.  I argue, however, that we cannot avoid commitment to singular thought, given (4), and another extremely plausible epistemic thesis, but sketch a conservative account of singular thought that allows us to retain (1)-(5), and show how to resolve (almost) the three contradictions we obtain by conjoining (1)-(5) with our other assumptions with a minimal alteration to our inconsistent set of propositions.  The resulting picture, I argue, is not only consistent with the semantic and epistemic theses which seem to undermine it, but can be deployed usefully in defense of them.  In addition, I defuse the sense of mystery surrounding knowledge of singular thought and show how this helps as well to shed light on some puzzles about self-locating thoughts.