It isn’t difficult to get people talking about free will. Just suggest they don’t have it, or that their conception of it is mistaken, and you’re in for a long, often entertaining and sometimes fractious discussion in which people’s fundamental ideas about human nature come to light. Nearly everyone (in the West at any rate) has at least a vague notion that free will is tied up with justifying moral responsibility, so the serious ramifications of the question add urgency to what otherwise might be dismissed as a mere philosophical puzzle.

The discussion might reveal that opinions on free will are divided. Some people have strong intuitions that to have it, and thus be morally responsible, we must not be fully determined in what we do, in which case free will is incompatible with determinism. Free will is thus contra-causal (or libertarian, as philosophers often call it) in that it transcends, overrides or goes against the ordinary sorts of causation we suppose obtain above the level of quantum phenomena. For these incompatibilists, if determinism is true, or likely true, then we don’t have free will; if it isn’t, then perhaps we do. For others, the compatibilists, free will is a matter of acting more or less rationally and voluntarily, even if we’re completely subject to cause and effect. Free will is therefore compatible with determinism, should it be the case; it isn’t contra-causal or libertarian. People still deserve praise and blame, reward and punishment, just insofar as they follow their inclinations, are not coerced or restrained, and are in full possession of their cognitive and impulse-control faculties. For compatibilists, should determinism prove true that wouldn’t require us to revise our ordinary ways of thinking about agency and responsibility.

Who’s right about free will, and how do we decide?  If there were a clear, easy answer to this question, it would have been settled long ago. As it stands, the free will debate continues both inside and outside the academy, with perhaps more public visibility these days as science weighs in on human nature. The empirical evidence is mounting that human beings are likely not causal exceptions to nature, so if you’re an incompatibilist (and many are), it’s getting harder to suppose we have contra-causal, libertarian free will. If we don’t, this might mean that some beliefs having to do with moral responsibility need reconsideration. The pressure from science might also be felt by compatibilists, since as the causal story of human behavior gets filled in the question about moral desert gets sharpened. Do people deeply deserve praise and blame, reward and punishment, for what they are fully determined to do? If so, why?

For those wanting to explore these questions in the context of the current debate, Four Views on Free Will is an excellent, engaging resource. Its contributors, all leading philosophers, stake out clearly defined positions that illuminate the traditional views on the free will problem and more recent permutations. The writing is crisp and collegial, if sometimes necessarily technical (see the articles on free will in the Stanford and Routledge Encyclopedias of Philosophy for background). Each contestant in the four way exchange gets two chapters, one to lay out his position and another to respond to the others. This provides a satisfying point-counter-point and helps clarify where positions differ and overlap. The reader comes away with a good grounding on the basics, plus a fairly detailed snapshot of the state of play as four experts on free will see it.

Robert Kane (University of Texas) is an incompatibilist who defends philosophical libertarianism, the view that we have free will because determinism doesn’t hold full sway in our choosing and deciding; he claims there’s a crucial indeterministic element that makes us deeply and ultimately responsible for ourselves and our actions. Derk Pereboom (Cornell), another incompatibilist, disagrees. He thinks it’s unlikely we are exceptions to determinism (and if we weren’t this wouldn’t help us be morally responsible[1]), and so ends up in what he calls “hard incompatibilism.” He forthrightly, and in my view rightly, recommends that we should rethink some of our beliefs and practices related to moral responsibility since we aren’t free in the way such beliefs and practices require.

Oppositely, John Martin Fischer (U of CA, Riverside) defends a version of compatibilism (“semi-compatibilism”) which suggests we are justified in our moral responsibility practices even if determinism turns out to be true. What matters is that human agents still have reasons-responsive control (“guidance control”) over their behavior, even if there are not alternative possible futures genuinely open to them. As he puts it “a semi-compatibilist need not give up the idea that sometimes individuals robustly deserve punishment for their behavior, whereas on other occasions they robustly deserve moral commendation and reward” (p. 81). Manuel Vargas (U of San Francisco) rounds out the debate with an argument for revisionism: since the belief we have libertarian free will is false (an empirical claim), and since most people have at least some libertarian beliefs (also an empirical claim), a “moderate” revision in commonsense concepts of free will and responsibility in a compatibilist direction is in order. However, Vargas doesn’t join Pereboom in advocating revisionism about our moral responsibility practices, which Vargas says find adequate support without libertarian free will.

It’s in these practices and our justifications for them that beliefs about free will and moral responsibility ultimately find expression. Further, it’s the existence of these practices that makes the question of free will matter so much. For instance, criminal punishment in the United States is widely justified by the idea that we’re giving offenders their just deserts, which in turn routinely justifies very punitive, non-rehabilitative regimes of incarceration, or even death. Consequentialist goals such as deterrence, rehabilitation and restitution may play a role, but we often punish not to achieve such ends, but simply to inflict deserved pain and deprivation on offenders. In recent debates about execution, it’s rare that anyone disputes what seems obvious, that killers deserve to die; rather the issues are largely procedural: reliability of evidence, fairness and consistency in sentencing, retaining good counsel, whether lethal injection counts as cruel and unusual, etc. It’s the machinery of death, as Supreme Court Justice Blackmun called it, not the deservedness of death, that gets most of the attention.

But according to some philosophers, including Derk Pereboom in this volume, it’s precisely the idea of retribution-entailing desert that can’t be sustained once we accept that we’re not libertarian agents, that we’re not ultimately responsible for ourselves. Pereboom writes: 

If hard incompatibilism is true, a retributivist justification for criminal punishment is unavailable, for it assumes that the criminal deserves pain or deprivation just for committing the crime, while hard incompatibilism denies this claim. And retributivism is one of the most naturally compelling ways for justifying criminal punishment.

Because retribution comes so naturally to us, Pereboom’s discounting of desert would, as he acknowledges, involve a sea change in our beliefs about moral responsibility. Libertarians such as Kane, compatibilists such as Fischer, and even self-described revisionists such as Vargas see no need or warrant to upend our settled practices and convictions about desert, so Pereboom is very much alone here in his radical prescription (he has allies elsewhere, however, such as Bruce Waller, Joshua Greene, Tamler Sommers, Will Provine and myself). The conservative, status quo defending majority in this debate gives more normative weight, perhaps, to the intuition that retribution-entailing, non-consequentialist desert is so central to our responsibility beliefs and practices that to abandon it would mean abandoning moral responsibility altogether. This is so despite the fact that they acknowledge, especially Vargas, that there are important consequentialist rationales for holding each other responsible, for instance to increase sensitivity and conformity to moral norms.

For Kane, who believes we are in some sense ultimately self-forming, the idea that we deserve punishment follows pretty directly, since we’re ultimately responsible for becoming who we are, not circumstances beyond our control. But for Fischer and Vargas, it’s a dicier proposition to square the retributive intuition with the naturalistically more plausible idea that we’re not ultimate self-originators, that we might well be fully determined in our character and behavior. Dicier because, after all, the ultimately self-made, buck-stopping self isn’t there to take the fall. So it’s interesting that Fischer spends little time defending desert per se, or reassuring those with libertarian intuitions that determinism doesn’t delegitmize their retributive instincts, as Pereboom suggests it does (p. 202). Near the end of his opening chapter Fischer says

In my view, we care deeply about being robustly free and morally responsible, and it is not straightforward to reconfigure our ideas or practices so that we eliminate residual retributive components in our attitudes to ourselves and others. Certainly, it is not easy to do so without a sense of loss.

Reconfiguring basic ideas about ourselves to fit with science, what the naturalist project essentially involves and that the philosophers here endorse, has never been easy or straightforward, and we should expect to take some losses (although Pereboom, myself and some other skeptics about free will see the end of retributive desert as a significant gain). If, as Fischer suggests might be the case, “there is only a single line into the future: a single metaphysically available path that extends into the future,” it seems prima facie implausible to suppose offenders strongly deserve punishment. After all, it isn’t as if, given that they, their circumstances and their actions could only have tracked this line, they could have chosen to be and do otherwise. The compatibilist has to show why punishment is deserved under these conditions. Do offenders deserve it because they were fully determined in their actions, or in spite of the fact that they were so determined? Fischer seems to sense that retributivism is under pressure but defends it only by implication: since all our responsibility practices are vindicated by (semi-) compatibilism, so is retribution.

Vargas also skirts the issue, even though he addresses the issue of desert generally. His conception of how our responsibility system works is essentially consequentialist, although he takes pains to point out that non-consequentialist considerations can and should apply (pp. 157-160). For instance, we don’t punish the innocent even if that would somehow guarantee a safer society, since to do so would violate the Kantian deontological axiom that people be treated as ends in themselves, an axiom that reflects perhaps the central value of our liberal democracy. But the obvious consequentialist question about retribution is why punishing the guilty, irrespective of social or personal benefits (the definition of retribution), should be obligatory in a deterministic universe, as most compatibilists seem to think. In a libertarian universe in which the offender is ultimately self-created in some respect (Kane’s view), this might make sense, because non-deterministic self-creation might entail the sort of deep deservedness that in turn might justify non-consequentialist suffering. But this justification isn’t available to the compatibilist, who admits that the offender was likely fully determined to be and act as he did. So from whence comes compatibilist retributive desert? Is it that inflicting pain and deprivation on the offender also reflects a central value of liberal democracy?

In his second chapter replying to others’ views, Pereboom challenges Vargas to clarify his position on desert and harsh treatment (p 202). Vargas replies to Pereboom in his second chapter, arguing that there are types of deservedness that survive under compatibilism, such that people still fundamentally deserve praise and blame. But Vargas does not say whether he thinks retributivism is allowable on his conception of desert, that is, whether in deserving blame offenders should be punished whether or not any good consequences ensue.

It’s clear that on Vargas’ consequentialist conception of morality, praise and blame function to make people more sensitive to, and conform to, moral norms. He says that those lacking the capacities for self-control and reasons-responsiveness which praise and blame normally engage are not fully blameworthy, which leads him to take a refreshingly strong stance in favor of mitigation:  “In our era, I doubt many would want to participate in a set of moral practices that were always and everywhere ruthless in their application of blame” (214).  But the question remains: does Vargas think retribution is ruthless, or not? After all, retribution is premised on a non-functional notion of blame, so given his generally consequentialist posture he should logically renounce it. But his earlier nod to non-consequentialist concerns, plus the fact that he recommends revisions in our (libertarian) intuitions about moral responsibility, but not in our responsibility practices, suggests that if pressed he would support it.

So, to my way of thinking, in this volume it’s only Pereboom who comes directly to grips with the toughest, most pressing question about free will and moral responsibility, and who answers it forthrightly and correctly. The scandal of compatibilism is that its advocates all too often dance around the issue of retributive desert, when what’s needed is a straightforward explanation of why, for instance, some killers (e.g., “the worst of the worst”) deserve to die in a universe in which nothing and no one is self-caused in any ultimate respect. Compatibilists such as Fischer and Vargas (and law professor Stephen Morse, see here) generally agree that what makes people deserving of blame is that they have forward-looking capacities of guidance control and reasons-responsiveness that make them sensitive to the prospect of credit and blame, so that conduct is guided by the prospect of reward and punishment. On this account of moral agency, holding people responsible makes sense precisely because we endorse the consequences of doing so, namely that people behave better than if we didn’t. If so, then it seems strange, indeed impossible, that this same criterion of moral responsibility can then be used to rule out the need for consequentialist justifications of punishment, the essence of retributivism. How this works is what compatibilists need to explain.

The upshot under compatibilism, even as rethought by Fischer and Vargas, is that punishing the guilty somehow becomes an intrinsic good in itself, or a categorical obligation that need not answer to demands for consequentialist justification, very much as libertarians might have it. Daniel Dennett, a compatibilist who’s written two books on free will, says in a book chapter (for Psychology and Free Will, Baer, Baumeister and Kaufman, eds., forthcoming) that

We ought to admit, up front, that one of our strongest unspoken motivations for upholding something close to the traditional [contra-causal, libertarian] concept of free will is our desire to see the world’s villains “get what they deserve.” And surely they do deserve our condemnation, our criticism, and – when we have a sound system of laws in place – punishment. A world without punishment is not a world any of us would want to live in. (original emphasis)

Here Dennett acknowledges that the retributive impulse (wanting villains to “get what they deserve”) drives belief in contra-causal free will. Why?  Because the libertarian agent is the perfect target for retribution. Having contra-causal freedom means the person could have chosen not to be a villain, even given the exact conditions that on a compatibilist view fully explain why he did. He therefore deeply deserves punishment, whether or not this makes him a better person or society safer, which is what consequence-ignoring retribution wants. Dennett, who doesn’t believe we have contra-causal free will, implies in this passage that people “surely” deserve punishment in the way that (mythical) libertarian agency makes possible. So although he hasn’t come out and said so, it strongly suggests he would endorse retributivism, not just the consequentialist rationales for punishment.

That Dennett thinks no one would want to live in a world without punishment is telling.[2] Does he mean only that punishment is a (currently) necessary infliction of suffering to deter wrong-doing, or that it's a non-consequentialist good-in-itself, as retributivists generally see it? The latter seems more likely, especially since in the same chapter he (gently) chastises Daniel Wegner, Richard Dawkins and Susan Blackmore for suggesting that humane revisions in our thinking and attitudes about moral responsibility and punishment might follow in light of not being libertarian agents. But if indeed he plumps for retributivism, he owes us an explanation of why it’s good to inflict suffering that achieves no good end.[3] Other requests (reasonable, I hope) for compatibilists to explain themselves more clearly are here, here, here, and here.

Returning to Four Views, Vargas says, correctly and crucially, that

…a naturalistically plausible and normatively adequate account of free will and moral responsibility will require abandonment of important parts of our self-image. These are aspects of our self-image that are not merely the inventions of philosophers or extravagant demands detached from widely shared concepts. They are pervasive and deep aspects of our commonsense.

This is to say that we should grow up, think straight, and break off our quasi-supernatural love affair with the libertarian, contra-causal free will of the soul. The looming practical question, though, is whether abandoning the soul and its freedom might in turn require abandoning certain concepts of free will and moral responsibility and the responsibility practices they support. Pereboom thinks so, as do Sommers, Greene, Waller, Provine and undoubtedly others (come out, come out, wherever you are!). But compatibilists, represented in this volume by Fischer and Vargas, think not. That they more or less agree with Pereboom about the sorts of capacities and freedom we actually have (those compatible with determinism, p. 94), yet disagree about which responsibility practices are justifiable, is on the face of it perplexing. Surely there’s something clear and obvious that explains why they don’t follow Pereboom in tossing retributivism. But what is this? If it can’t be clearly articulated, why aren’t more disbelievers in libertarian agency non-retributivists?

I read Four Views hoping to find the answer to these questions, but did not, perhaps due to my own incompatibilist biases and blindness to certain lines of argument. At any rate, I won’t blame this terrific book, which I unreservedly recommend for those wanting to step into the thick of the free will debate. Not that these four views are the only ones on offer, but Kane, Fischer, Pereboom and Vargas do a great collaborative job of locating their positions in debate-space, explaining how they connect to the central literature on free will, and defending them against the best shots of the competition. For those gripped by the project of understanding what free will and moral responsibility could be, now that science has stripped away the soul, what could be better?

TWC, November, 2007

Notes