Free Will Matters

Free will is a topic of immense practical importance. This, in contrast to some less momentous philosophical topics, such as whether statements about story-book characters can be true (e.g., “Peter Pan can fly”). We care about whether someone should be held responsible for an action such as child molestation, both because it matters overwhelmingly – to the victim and to the family members– but also because of implications for the level of community-wide trust in the institutions of criminal justice (Seabright 2005). That matters to the well-being of all, because when the justice system is deemed deeply flawed, people are apt to take the law into their own hands. Vigilantes, posses and rough justice can be a shockingly ugly business. Moreover, although levels of trust in the system of justice are closely tied to fair-minded concerns for public safety and civil stability, they are also linked to prevailing beliefs about when someone is in control of his actions, or when control is diminished. Granting that such beliefs tend to be fairly stable over time, they are not, however, absolutely fixed, but are sensitive to revision as new discoveries are made; e.g. about the tendency of brain tumors or strokes in the orbitofrontal area to compromise impulse-control. So long as the discovery is well-explained, the effect on public understanding of how to view responsibility in such cases can be substantial.

One problem that continues to complicate virtually every discussion of free will concerns semantics: What exactly does the expression mean? One useful observation from experimental psychology that is yet to have its due impact concerns the cognitive architecture of everyday categories. Thus George Lakoff and Mark Johnson (1999) have pointed out that categories, including freely chosen or voluntary, are radial, as are most other familiar categories such as vegetable, house, mountain, friend. That is, they have highly recognized prototypes at the center, with diminishing degrees of similarity as examples resemble less the prototypes in the central area. Additionally, the category boundaries are fuzzy, not sharply defined. By contrast, many scientific categories, such as amino acid or methylation are more precisely defined. In the case of free choice, a prototype of voluntary behavior is that of a fully awake human who walks out the door and picks up the morning paper, or the financial ‘wizard’ who sets up a Ponzi scheme and rakes in ill-gotten gains over many years. By contrast, someone suffering an epileptic seizure does not freely choose his actions. Less close to the central freely chosen prototype is the behavior of a very hungry and nervous graduate student who scarfs down a plate of appetizers before the guests arrive; at even greater distance is that of an exhausted new parent who falls asleep during one of my most exciting lectures; or the Touretter who barks out “crap, crap, crap” during that same lecture. Populating the fuzzy boundary area are the problematic cases that are argued before the law; the severely depressed woman who drowned her babies; the ten year old who shoots his abusive father; the manic-depressive who buys two Jaguars in a single afternoon; the patient in the early stages of Alzheimer's disease who makes a decision regarding treatment (see the discussion by MacKenzie and Watts 2011).

Lakoff and Johnson's (1999) observation is useful not only because it fits well with data on how ordinary human understanding is organized, but also because it accounts for common inferences; e.g. “since running a Ponzi scheme was a voluntary undertaking, then the person knew what he was doing was wrong.” Moreover, it explains the lack of agreement on the problematic and contentious cases, relative to the prototypical cases. This perspective also implies that while we can hew out rough definitions, precise definitions (necessary and sufficient conditions) that sort out in a satisfying way, all the messy cases in the boundary area may continue to elude us. This is true for river and friend as well as for freely chosen. Such cases populate the fuzzy area exactly because no clear and established criteria suffice to sort them.

It is a boon when science can shed light on particular kinds of cases in the fuzzy boundary, by helping us understand their neural basis, even if only sketchily. Thus Caspi and colleagues (2002) reported that in their longitudinal study in New Zealand, a small subset of males—those who had a genetic variant for the enzyme monoamine-oxidase-A (MAOA)—were highly likely to show self-destructive and resilient aggression if they were also subject to abuse as children. Some feature in the “self-control” networks of these males is different from that of healthy lads in the cohort, and it makes them volatile and aggressive. Although a similar result was obtained in a Swedish study, it remains unclear whether the finding can be replicated in less homogeneous groups (Aslund et al. 2011). Assuming the results are widely replicated, this particular case is worth dwelling on because the phenotype is impulsively hostile and randomly so—a matter of nontrivial social importance.

Nevertheless, it should be emphasized that knowing the general explanation for the unprovoked aggression by an abused MAOA variant should not be taken to require us to let the person go free. And that logical point needs emphasis because occasionally scientists in an “upset-the-apple-cart” frame of mind will gleefully proclaim that an explanation is tantamount to an excuse. More daringly, it is sometimes proposed that no one should really be held responsible for anything, be it child molestation or running a Ponzi scheme. This is unrealistic. One unwavering look at the behavior of an MAOA variant who suffered child abuse should be enough to usher in reconsideration of that radical idea. Certainly the ordinary folk that I chat with on dog beach are troubled by the prospect of emptying the prisons of all their psychopaths, serial killers and child molesters, even while agreeing that we incarcerate too many people for too long. On the other hand, as Felsen and Reiner (2011) argue, if the conventional wisdom regarding free will turns out to be clearly at odds with neuroscientific data, this should spawn thoughtful reconsideration of what that expression ought to mean (see also Ramos, 2011).

Our current concept of free will does, it seems, entwine separable strands: one strand emphasizes the “could-have-done-otherwise” element, and another, emphasizes the social cost of that kind of action, which is frequently expressed in terms of convictions about responsibility for that kind of action. These convictions then shape the judgment about voluntariness (crudely: we must protect ourselves from such a person, so his action must have been voluntary). Thus a strand that is essentially factual (was the person in control?) and a strand that largely social (the person should be removed from open society) become interlocked in a way that can take us around in confusing circles instead of down a pathway to clarity. It may be that we do best to separate the neurobiological issues of control from the social issues concerning responsibility and the need for public safety (Dalley et al. 2011). For this reason, I prefer to avoid the expression, “free will” in favor of “control” (see a related point concerning “autonomy” in Felsen and Reiner, 2011). Thus, in the case of the abused MAOA variant who strangled his neighbor, it may be that he lacks the capacity to control his aggression, and that we currently have no way of treating that condition to restore the capacity. Nevertheless, given what is known, we can predict that the behavior is likely to recur and hence that he cannot be left at large to strangle other neighbors. The three basic questions of criminal justice (Did he do it? Can he be treated? and Will he do it again?) can more straightforwardly shape the judgments about what to do in a particular case once the factual and the social are separated.

Shifts in our social evaluation of how best to respond to criminal behavior can occur when effective treatments become available (see Glannon, 2011). A case in point concerns addiction. When judges have discretion to offer heroin addicts a choice between prison and a drug that will block heroin's hedonic effects (an option known informally as “beads in the butt”), then a better outcome for all may result. Despite this optimistic scenario, an effective treatment for psychopathy, for example, does not seem to be within reach, and in general, good treatments for nervous system disorders are very hard to come by. Additionally, the issue of intervening in the nervous system raises further questions concerning safeguards against “treating” the merely eccentric or obstreperous.