In this paper, I describe and defend a theory of practical and theoretical reasoning which I call the 'Endorsement Theory'. Some philosophers, Gilbert Harman in particular, have held that reasoning consists of transitions from one set of dispositional states to another. Inferring that q from p, for example, consists in being led to believe that q from believing that p. Practical reasoning differs from theoretical reasoning in that it consists in transitions to states of intending and wanting, as well as believing. I argue that this theory, which I call the 'Dispositional-Transition Theory', faces several difficulties, most notably an inability to account for cases of re-inferring what one already believes or intends. The Endorsement Theory, which, in accordance with common intuitions about reasoning, takes reasoning to essentially involve occurrences of thinking, avoids these difficulties.
Reasoning -theoretical and practical -can be looked at in two ways. When a person makes an inference, there is something that he infers. And what he infers, he infers from something else. What the agent infers and what he infers this from make up the contents of his inference. The contents of his inference are not mental states, attitudes, or entities of any sort, but rather abstract entities, e.g., propositions. If S infers that his car is in the garage from something else he believes, say, that he saw his wife drive it in there, he comes to believe that his car is in the garage, but he does not infer his believing that his car is there. He infers that his car is in the garage, which is a proposition. And he does not infer this from his believing that he saw his wife drive it in there. He infers this from what he believes, viz., that he saw his wife drive it in the garage.
We can describe acts of inference abstractly in terms of their contents. We might e.g. represent the above inference as *I (= S) saw my wife drive my car in the garage; therefore, my car is in the garage*. The word 'therefore' indicates that what follows is what-is-inferred and what precedes is what-was-