This volume is a detailed introduction to central issues in the free will debate. It is unusual in several respects; centrally, in that rather than examine the strengths and weaknesses of rival accounts of free will, it focuses on the arguments that motivate incompatibilists. These are arguments that purport to establish the incompatibility of free will and causal determinism; theories are introduced only insofar as it is necessary for further understanding of these arguments and their limitations. Since this compatibility question is so central to the free will debate as a whole, the theories introduced are not restricted to those on the incompatibilist side of the ledger, but include (semi-)compatibilist views such as those of Fischer and Frankfurt as well.

Most chapters are devoted to a single argument in favour of incompatibilism, with some interludes to explore closely related issues (whether moral responsibility requires alternative possibilities; whether responsibility is a partially historical concept); the last chapter is devoted to an argument against libertarianism (the luck objection). Many philosophers who are not actively engaged with the contemporary free will debate assume that compatibilism holds sway among thoughtful people. From within the debate, things look very different, with various incompatibilisms occupying much of the terrain. The revival of compatibilism owes much to the consequence argument, best known in Peter Van Inwagen's formulation. Roughly, the consequence argument proceeds as follows: if determinism is true, then everything we do is the consequence of states of affairs (plus the laws of nature) which precede our birth. These states of affairs and laws are not up to us; nor is it up to us that given these states of affairs and laws we act as we do. Therefore, how we act is not up to us. Given the centrality of the consequence argument to the revivification of incompatibilism, Haji makes it central to his discussion, carefully analyzing various formulations and the inference rules upon which they rest. This sets the tone for the book: one by one well-known arguments for incompatibilism are advanced and dissected; their weaknesses laid bare and more powerful formulations suggested.

 Haji believes that none of these arguments is persuasive enough to establish the superiority of incompatibilism over compatibilism. But the treatment is fair-minded and respectful of those who dissent. In many ways, the fair and detailed treatment of central topics within the free will debate makes the book an ideal teaching text. It is, not, however, suitable as an introductory text for those with little philosophy background: it is too demanding, and moves too quickly, for that. Together with a genuinely introductory texts, such as Robert Kane's A Contemporary Introduction to Free Will, it would suit an advanced undergraduate course. It will also prove of great interest to professional philosophers, including free will specialists: Haji has new and significant things to say on questions that are the subject of great contemporary interest.

I have little critical to say about the book. There is nothing like it available today, and it is certain to set the standard for some time to come. Haji's own views will prove contentious, but addressing my concerns with them would take a substantive article, not a book review. Let me simply and briefly note three points about which I had concerns.

Haji defends a qualified historicism about moral responsibility. A full-blown historicism claims that a responsible agent must have a history of a certain sort; Haji rejects this view, in favour of the claim that a responsible agent must lack a history of a certain sort. A history which undermines responsible agency brings it about that the mental states involved in action-production are not authentic. His account of authenticity, in this context, is relational:  the relevant mental states are authentic if they do not undermine responsibility for actions at some later time (the account cannot hold that the springs of action are authentic only if they do not undermine responsibility for actions right now, because Haji wants to be able to use it to distinguish the oppressive and normal socialization of children, but very young children are not capable of morally responsible action). The problem with this relational account is that it seems to commit him to the following oddity: a powerful deity could create a person with chaotic mental states, states that render that person clearly insane, but which would count as authentic on his view because these states evolve in such a manner as to allow the person to become morally responsible at a later point in time.

Second, it seemed to me that Haji wasn't sufficiently attentive to the kind of origination that at least some source incompatibilists want. This comes out in his discussion of a view like Kane's according to which indeterminism is needed if the agent is to be the source of his or her own actions. Haji argues that indeterminism fails to give the agent any kind of origination unavailable on an adequate compatibilism. He advances this claim by considering a hybrid agent-causal theory, on which actions are both agent-caused and deterministically event-caused. On this scenario, the agent is the ultimate cause of the event and therefore an originator in a strong sense. Hence, Haji concludes, it cannot be the case that we need indeterminism for origination. However, this underplays the source of the desire for origination, as I understand it. Kane believes that origination matters because only if we are the ultimate source of our actions can we be differences makers to the causal history of the world; only if I am the source of my actions is it false that everything I do was on the cards prior to my birth. I cannot have ultimate origination, in this sense, in a deterministic universe.

Third, I had concerns about Haji's discussion of the deontic argument. The deontic argument is a moralized version of the consequence argument. Its crucial premise is (1) If S is morally blameworthy for doing A, then A is morally wrong for S, plus the claim that if an action is morally wrong for an agent, then that agent can refrain from that action (a version of "ought" implies "can" – OIC). This establishes an alternative possibilities condition for blameworthiness, and therefore, like the consequence argument, is an indirect argument for incompatibilism (indirect because it presupposes the principle of alternative possibilities.

Haji believes that OIC is unassailable. But he denies that (1) is sound. He thinks that there are cases in which an agent is morally blameworthy for an action, but that action is not morally wrong for that agent. Here is his central case (in brief). Deadly believes (wrongly) that Ernie has disease D; he also believes (rightly) that injecting a substance into anyone with D will cause their death. He wants to cause Ernie's death, so he injects the substance in Ernie. But Ernie does not have D; instead he has D', which would have killed him were it not for the fact that Deadly injected Ernie with the substance. In this case, Haji maintains, Deadly might be blameworthy for injecting Ernie with the substance, but it is false that his doing so is morally wrong for Deadly.

Surely, however, the correct view is that Deadly is morally blameworthy for doing something that is morally wrong from his perspective: doing what he believes to be morally wrong (or perhaps doing something he knows would be regarded by almost everyone else as morally wrong)? Haji considers and rejects this objection. He thinks that it can be seen off if we bear in mind that blameworthiness is focused on agents, and wrongness on acts. Since wrongness is focused on the act, and not the agent, we ask whether Deadly's act was wrong by examining its actual properties, and not Deadly's beliefs about its properties. Hence his deed is not wrong.

However, while it might be appropriate to maintain the focus on the act when we assess wrongness, we cannot maintain the act-centered focus when we assess blameworthiness. We must maintain the same object of focus when we assess claims like (1); we cannot shift from one to another. I worried that the antecedent of (1) implies an agent-focused understanding of the act for which blameworthiness is in question, while the consequent shifted to an act-focused understanding. A similar worry arose with regard to Haji's claims about obligation. He defends OW: It is overall obligatory for one to do something if and only if is (overall) wrong for one to refrain from doing it. But obligation in OW seems to be agent-focused, whereas wrongness in the consequent of OW seems to be act-focused. It therefore seems to me false, contra to what Haji asserts, that when an agent does something for which they are blameworthy, but could not have done otherwise due to the presence of a counterfactual intervener, their action is not morally wrong. This seems to involves an illicit agent-focused conception of moral wrongness, illicit because the question "Is A wrong" is act-focused.

This is one point at which Incompatibilism's Allure suffers from the compression of its arguments. However, if it motivates the reader to turn to Haji's earlier books for a fuller defense of OW and related principles, this is no bad thing.

© 2009 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

Keywords: philosophy, free will