Suppose that, for every event, whether mental or physical, there is some physical event causally sufficient for it. Suppose, moreover, that physical reductionism in its various forms fails-that mental properties cannot be reduced to physical properties and mental events cannot be reduced to physical events. In this case, how could there be mental causation? More specifically, how could mental events cause other mental events, physical events, and intentional actions? The primary goal of this paper is to answer this question.(1) The explanation that emerges is based on three guiding ideas. First, a mental event (rather than a competing physical event) causes a subsequent mental event because of the special strength of certain fundamental psychological laws, namely, laws upon which acceptable nonreductive functional definitions may be based. Second, a mental event (rather than a competing physical event) causes a subsequent physical event because of (a) the strength of these psychological laws plus (b) the strength of relevant psychophysical correlations.(2) Third, when a mental event (rather than a competing physical event) causes a subsequent intentional action, we have an instance of what I call essential-constituent causation: one essential constituent of the intentional action is physical and the other mental, and the antecedent mental event causes both essential constituents (whereas no competing physical event does). A subsidiary goal of the paper is to explain, not just how a mental event can be a cause of a mental or physical effect, but also how it can be the cause-or, at least, why in a particular context it is correctly deemed to be the cause.(3) Some philosophers think that this sort of question is entirely a matter of pragmatics (interest, salience, etc.). But even if pragmatic considerations are involved, it does not follow that there are not objective criteria that, relative to a context, make it correct to identify one event as the cause of another (rather than, for example, one of two overdetermining causes, one of two joint causes, or some other alternative). A full account should make clear what these criteria are. Another subsidiary goal concerns the question of just how strong the psychophysical correlations are. Specifically, does the mental supervene on the physical as a matter of metaphysical necessity, or does it supervene in a weaker (nomic) fashion? The goal is to construct an account that remains neutral with respect to this highly controversial question, for, other things being equal, it is best to steer clear of avoidable controversies. If a successful neutral account can be given, a significant dialectical point follows: such an account will undermine an interesting theoretical argument in favor of metaphysical supervenience (hereafter, simply 'supervenience'), namely, that supervenience must be adopted as a premise in any successful account of mental causation.(4) I should note, finally, this neutral account is also compatible with various forms of naturalism (and with their denials). A final subsidiary goal is to provide the resources to answer a question receiving much attention of late, namely, how to distinguish between genuine (justification preserving) inferences and merely incidentally caused sequences of thought.(5) The answer is that genuine inferring involves a species of mental causation explained by our account-a species that is underwritten by laws of rational psychology (a key subset of the psychological laws upon which nonreductive functional definitions may be based). I will not, however, have space to develop this account of inference here. The paper is organized as follows. After developing a test for showing that a given event is the cause of a particular effect ( 1-2), I give the account of mental-to-mental causation (3). I then give the account of mental-to-physical causation and the account of intentional action ( 4-5). I close with some brief remarks on mental causation and purely reflexive behavior (6). Before beginning, I should elaborate upon my starting points. First, I will assume that mental properties are not reducible to either first-order or second-order physical properties. That is, I will assume that the identity thesis and reductive functionalism are mistaken.(6) (My reason for rejecting the ordinary identity thesis is based on multiple realizability intuitions together with a rebuttal of the scientific-essentialist (i.e., necessary a posteriori) response.(7) My reason for rejecting reductive functionalism (both "American" and "Australian") is that reductive functional definitions require the wrong sorts of things to be the contents of our self-consciousness: the contents would have to be propositions involving physical "realizer properties" (e. g., having firing C-fibers) rather than familiar mental properties themselves; therefore, given that the identity theory is mistaken, this would imply that the contents of our self-consciousness cannot involve any familiar mental properties.(8)) Some readers will of course be unwilling to abandon these reductive theses. This paper, however, should still be of interest to them, insofar as a standard objection to the non-reductionist alternative is that it is unable to account for mental causation. For, if correct, the present account will answer this objection. Second, I will assume that, even though reductive functionalism is mistaken, nonreductive functionalism is correct.(9) (Or, more cautiously, I will assume that there is a family of distinctively strong mental-to-mental ties of the sort that would be assured by nonreductive functionalism if it were correct.) By 'nonreductive functionalism' I mean that form of functionalism that identifies the standard mental properties and relations (being in pain, thinking, etc.) with the unique properties and relations that make an appropriately general psychological theory true (i.e., the sort of general psychological theory upon which reductive functionalists had hoped to base their reductive definitions). In other words, this sort of psychological theory implicitly defines the standard mental properties and relations. Third, I will assume that an adequate account of mental causation must explain the role mental properties play in mental causation, and so must go beyond token-identity coarse-grained-event accounts (e. g., Donald Davidson's) that do not explicitly provide such an explanation. If the role of mental properties can somehow be explained in a fine-grained-event framework, then, plausibly, it can be reworked into an account of mental causation constructed in a setting of coarse-grained events. Since the fine-grained framework is so easy to work with-and since I find it to be more plausible in any case-I will assume that this framework is correct. Finally, as indicated above, I will assume weak causal closure: every actual event has some physical event that is causally sufficient for it.(10) Of course, weak causal closure does not entail strong causal closure, namely, that every actual event has physical and only physical causes. As is revealed by attributions of cause in applied science, medicine, and law, causing an event is intuitively a very different matter from merely being causally sufficient for an event. It is in the logical gap between weak and strong causal closure that mental causation lives. Failure to appreciate this opening in logical space has led many philosophers to the premature conclusion that mental causation is untenable unless mental properties are somehow reducible to physical properties.
