What is it to have a concept? What is it to make an inference? What is it to be rational? On the basis of recent developments in semantics, a number of authors have embraced answers to these questions that have radically counterintuitive consequences, for example:
One can rationally accept self-contradictory propositions (e.g.
Smith is a composer and Smith is not a composer). Psychological states are causally inert: beliefs and desires do nothing.
The mind cannot be understood in terms of folk-psychological concepts (e.g. belief, desire, intention).
One can have a single concept without having any others: an otherwise conceptless creature could grasp the concept of justice or of the number seven.
Thoughts are sentence-tokens, and thought-processes are driven by the syntactic, not the semantic, properties of those tokens.

In the first half of Conceptual Atomism and the Computational Theory of Mind, John-Michael Kuczynski argues that these implausible but widely held views are direct consequences of a popular doctrine known as content-externalism, this being the view that the contents of ones mental states are constitutively dependent on facts about the external world. Kuczynski shows that content-externalism involves a failure to distinguish between, on the one hand, what is literally meant by linguistic expressions and, on the other hand, the information that one must work through to compute the literal meanings of such expressions.
The second half of the present work concerns the Computational Theory of Mind (CTM). Underlying CTM is an acceptance of conceptual atomism the view that a creature can have a single concept without having any others and also an acceptance of the view that concepts are not descriptive (i.e. that one can have a concept of a thing without knowing of any description that is satisfied by that thing). Kuczynski shows that both views are false, one reason being that they presuppose the truth of content-externalism, another being that they are incompatible with the epistemological anti-foundationalism proven correct by Wilfred Sellars and Laurence Bonjour. Kuczynski also shows that CTM involves a misunderstanding of terms such as computation, syntax, algorithm and formal truth; and he provides novel analyses of the concepts expressed by these terms. (Series A)