I approached this book with trepidation. The authors — one a philosopher and one a cognitive neuroscientist — both work at the Fuller Theological Seminary, and I feared this would be an uninformed and credulous exercise in Christian apologetics. In fact the book is a lot better than that. Though there is a theological motivation, it is not allowed to blind the authors to the evidence. They aim to reconcile a robust picture of human agency with science, by way of a defense of a nonreductive physicalism. The view is physicalist in that it presupposes no laws or substances other than those that are (or ought to be) countenanced by science; it is nonreductive in that it holds that agents are themselves causes of their behavior.

Though the research strategy pursued here is a sensible one, the book is wildly over-ambitious. It seeks to develop a full account of agency, topdown causation and language, not to mention free will and mental causation, along the way defending controversial positions in metaphysics such as emergence, and taking positions on the function and nature of consciousness. Any one of these topics is fit for a book; attempting to solve them all, while laudable in one way, is rather foolhardy. Inevitably, the views defended are often sketched, rather than adequately defended, and pressing objections go unanswered.  An example, almost at random: Murphy and Brown argue that we couldn't have been zombies; ie, that phenomenal consciousness plays a functional role such that any being that lacks phenomenal consciousness could not be a functional duplicate of us. This, allegedly, is because consciousness supplies us with 'second-order knowledge' — the knowledge that we know something — and we can and do use second-order knowledge to guide our behavior. This, they argue, is why sufferers from blindsight can use visual information to guide their actions, but cannot do so unprompted: they know but do not know that they know, and therefore cannot use the knowledge spontaneously. This is an interesting idea, but it is also clearly underdeveloped and open to apparently devastating objections. First, Murphy and Brown owe us an account of second-order knowledge such that blindsight sufferers lack it, since there is obviously a sense in which blindsight sufferers do know that they possess visual information (they have been told about their success on trials). Second, the account seems to be vulnerable to counterexamples, since there are disorders of consciousness which cause a loss of conscious but not unconscious visual information and in which sufferers can apparently guide their behavior spontaneously using this visual information (this seems to be true of Milner and Goodale's famous patient). Third, the thesis Murphy and Brown are defending apparently concerns the necessity of phenomenal consciousness for agency, not its actual function (they ask whether there could be zombies functionally indistinguishable from us). So even if their story is true, why should we think that second-order knowledge must be provided by phenomenal consciousness? Murphy and Brown respond to this question, calling the supposition incoherent: zombies couldn't know that they have first-order conscious perception, since it is false that they have first-order conscious perception. But this response misses the point: the objection is not that we could have unconscious second-order knowledge of first-order conscious knowledge, but that we could have unconscious second-order knowledge of first-order unconscious knowledge. Presumably if unconscious first-order informational states are possible (and they are) then so are unconscious second-order information states.

The kinds of sketchiness of argument epitomized above seems to characterize the entire book and its major claims, as well as its minor. It is far from clear to me that at any stage Murphy and Brown's arguments, successful or unsuccessful, actually have the upshot they claim. They aim to defend nonreductive physicalism; they do so by way of attempting to show the reality of mental causation and the existence of causal powers that are supervenient on ensembles of elements. But reductivists need not deny either mental causation or the existence of these causal powers, hence the majority of the arguments simply miss their target. Consider their claims about top-down causation. So far as I can tell, their argument comes to this: wholes which are constituted of parts have causal powers — i.e., causally produce effects — which their parts could not produce on their own (Murphy and Brown have no truck with exotic causal laws; the powers they claim for wholes are explicitly held to be compatible with and explicable by the same laws of physics that apply at the level of their parts). Clearly this is true: my pressing the keys on my keyboard causes letters to appear on my screen, and were my computer to be decomposed into its constituent atoms, my pressing down in their vicinity would not cause letters to appear anywhere. Just as clearly, though, no sensible reductivist would deny this claim. Yet Murphy and Brown seem to attribute it to them. They claim, for instance, that reductivists cannot explain the behavior of an ant colony, because they are committed to holding that the ants are atoms whose 'characteristics are not affected by relations within their colony' (p. 96). No reductivist would hold this. Instead, they would claim that the ways in which the behavior of ants is altered by their relations within their colony could be captured by extremely long and unwieldy equations at the level of basic physics.

This problem gets repeated unchanged in Murphy and Brown's discussion of mental causation. They argue that mental states are not reducible to the physical state upon which they supervenes because they are embedded in a broader context, which extends beyond the brain (in other words, they side with externalism, apparently on both content and location of mental states). But this would be a good argument only if mental states were required to supervene only on brain states. The obvious move, in response to externalism, is to hold that mental states supervene on (and therefore may be reducible to) brain states plus whatever elements of its broader context are relevant. Moreover, even if it were true that the informational content of mental states could not be reduced to whatever its supervenience base turned out to be, it would not follow that that informational content did any causal work.

The purpose of the book is to show that free will and moral responsibility are possible, given the truth of physicalism. The upshot of the work on mental causation and agency is a picture of human action as flexibly responsible to reasons. This is, of course, a standard kind of compatiblism (Murphy and Brown acknowledge that their view is similar to Dennett's, though also and sensibly realist about consciousness and rationality). The picture of agency here is rich and plausible. But it is far from clear that it is adequate as a picture of moral responsibility. Murphy and Brown give us an account of moral responsibility which agents like theirs appear to satisfy, but it is clearly inadequate. For them, 'one is morally responsible when one has the ability to evaluate, in light of some concept of the good, the factors that serve to shape and modify one's action' (240). This may be a satisfactory account of moral agency, but it cannot serve as an account of morally responsible action, since it fails to exclude actions that are the result of manipulation, coercion or compulsion (for a start). Indeed, almost everything that Murphy and Brown say in their chapters about moral responsibility and free will is defensible only if they are understood as presenting an account of morally evaluable action. The problem, of course, is that no major incompatibilist denies that human beings regularly engage in morally evaluable action, nor believes that such action is threatened by determinism. The free will debate has other concerns, which Murphy and Brown fail to appreciate.

The incompatibilist concern, of course, is that if our actions are necessitated by the laws of nature and past states of affairs, we lack true control over their unfolding (Murphy and Brown do briefly advert to this concern, as expressed by Peter Van Inwagen, and argue that it depends upon a 'dichotomous option' between things being up to us or not up to so, but clearly the argument would succeed just as well if we replaced 'up to us' with 'not even partially up to us', since neither the laws of nature or past states of affairs are even partially up to us). Now, it may well be, as compatibilists have often argued, that the concern is misplaced, but in the absence of an argument to that effect, arguing that agents act for reasons simply begs the question against incompatibilism.

The picture of agency presented here is one well worth pursuing. Murphy and Brown have not presented us with a view that is defensible, both because it is far too sketchy to be properly assessed, and because many of the claims made will no doubt turn out to be false. However, the general outlines of the view are plausible, and there is a rich research agenda here. Perhaps future work will see some of the details worked out, and the gaps filled.

© 2008 Neil Levy

Neil Levy, Ph.D., Senior Research Fellow Program Manager, Ethical Issues in Biotechnology, Centre for Applied Philosophy and Public Ethics, University of Melbourne

Comment on this review