Fear of Mechanism: A Compatibilist Critique of "The Volitional Brain"

This essay appeared in the Journal of Consciousness Studies' special issue on free will, The Volitional Brain: Towards a Neuroscience of Free Will, reprinted as a book from Imprint Academic. As the title suggests, it is largely a commentary on the issue's contents, with a partisan objective. It serves somewhat to balance the influence of editor-neuroscientist Benjamin Libet, as well as some other libertarian contributors, who are bent, it seems, on discovering contra-causal free will somewhere in the brain. The libertarians and their evident fear of mechanism are a good foil to showcase a humanistic determinism which has all the necessary resources, I argue, to support our ethical intuitions and which might also soften punitive and ego-driven attitudes that permeate our culture.

As several contributions to this volume make clear, the problem of free will engages us deeply because it seems central to our conception of who we are, our place in the world, and our moral intuitions. To take a position on whether we have free will, and what sort of freedom this is, is to take positions on a host of other fundamental and necessarily interlocking issues: what we ultimately consist of as selves, the relation of mind to body, the role of consciousness in behavior, the proper methods of scientific and phenomenological inquiry, the need for foundations in ethics, and the possibility of the supernatural, among many other questions. To define the will, or volition, and argue that this definition captures the truth of the matter, is to invoke an entire world view, which must stand against its competitors.

And competition there is, since world views have consequences in policy and politics. When U.S. cigarette companies, defending themselves in class action suits, claim that smokers could have chosen to quit at any time, they are appealing to a particular notion of will and its relation to behavior. As the Truth and Reconciliation Commission completed its work in South Africa, some have questioned whether justice is served by forgoing punishment in return for the true story of one’s misdeeds. Shouldn’t retribution against the freely willing despot trump the need for social reconstruction? Or, conversely, might not the retributive impulse be softened by understanding the causal histories of political criminals? Such questions inevitably get answered using background assumptions at the heart of the free will debate.

As mentioned in the introduction, the argument over free will is in a very real sense the successor, and to some extent the companion, of the centuries old argument about God. Since the Enlightenment, Western theistic views, placing power and our ultimate fate in the hands of an almighty deity, have to some extent given way to a secularism in which our destiny is understood to lie largely within our own hands. We have usurped God’s power, or at least a good share of it, but many continue to believe the capacity to shape ourselves still supercedes the rest of nature and its physical laws. Free will, on the libertarian, incompatibilist account, in which we are ultimately responsible for ourselves and our acts (see Introduction, p. xiii), makes the self more or less a first cause, an unmoved mover: we could have willed otherwise in the radical sense that the will is not the explicable or predictable result of any set of conditions that held at the moment of choice. The question of whether we actually have such free will thus recapitulates in the domain of human metaphysics the question of the existence of God. That many believe that we stand above nature in some essential respect suggests that the Enlightenment was more successful in its glorification of the individual than in its challenge to the supernatural (I owe this last point to Keith Sutherland, publisher of the Journal of Consciousness Studies).

Libertarians tend to suppose that not only do we have such freedom, but that without it our moral intuitions and institutions are at risk. The ultimately responsible, self-originating self - the causa sui self, to use Galen Strawsons’ term (Strawson, 1998) - is thought necessary to ground ethical judgments and social justice, and a fully explanatory and inclusive science of human behavior threatens the status of such a self. Using the other (Peter) Strawson’s typology (Strawson, 1962) such libertarians are pessimists regarding the compatibility of (current) science and personal responsibility. On the other hand, compatibilists and determinists such as myself are optimists, who cheerfully accept that the will is a entirely a function of antecedent conditions, and find that no capacity for ultimate self-determination is either conceivable or necessary to found our moral intuitions.

Peter Strawson’s classification cuts across the categories represented here – neuroscience, psychology, physical science, and philosophy – in that scientists such as Libet, Schwartz, and Stapp agree with legal philosopher David Hodgson that having libertarian free will is an essential aspect of our humanity which may eventually be confirmed by science, perhaps in radical departures from current theory. On the other side, neuroscientists Ingvar, Frith and Spence, are joined by philosopher Gomes, psychologist Claxton, and polymath McCrone in taking more or less the compatibilist position: volition is no more (or less) than the brain in action; it is therefore in principle explicable within our current, more or less deterministic physical understanding (at the neural, cybernetic level, no need for quantum niceties); and that this sort of volition is sufficient for our moral purposes. This last point - the issue of moral foundations and the role of free will - is not addressed in depth in these papers, so will get a bit more attention from me in what follows (see Waller, 1998; Clark 1998).

Benjamin Libet’s piece illustrates several key difficulties faced by libertarians as they seek to find free will of the ‘ultimate’ variety within science. First, it is clear that Libet’s research, and more importantly, his interpretation of it, is driven by what we might call fear of mechanism. He worries that under standard scientific determinism, ‘we would be essentially sophisticated automatons, with our conscious feelings and intentions tacked on as epiphenomena with no causal power.’ The impression given from the start (and this concern is evident in Stapp’s and Hodgson’s pieces as well) is that there is a right answer to the free will question, one that will secure us from the bogey of mechanism and its pernicious spawn of relativism, fatalism, fascism, puppethood, abuse excuses, and the like. The implicit agenda is: since it’s unthinkable that we don’t have free will (we don’t want to be automatons, do we?) we’d damn well better come up with proof that we do.

While having an agenda works for meetings, companies, and political parties, it’s out of place in scientific investigations, and in Libet’s case the need to find a shred of evidence to support libertarian free will forces him into a strained interpretation of his own data. His main finding, which very much undercuts libertarian free will, is the unsurprising one that conscious acts are invariably preceded by unconscious neural processes which prepare for action. After all, shouldn’t we expect consciousness, obviously a brain-based phenomenon, to have causal links with unconscious processes, some of which precede conscious episodes? But Libet, intent on his quest to defeat ‘mere’ mechanism, gamely tries to make a case that our will resides in the ‘conscious veto’ not to perform an act after unconscious processes have readied it for expression. The obvious question, to which Libet has only the weakest answer, is whether or not this veto itself has neural antecedents and correlates. If it did, then it would simply be yet another part of a complex physical system (the brain) responding in astonishingly intricate ways to generate appropriate behavior. He says:

I propose…that the conscious veto may not require or be the direct result of preceding unconscious processes. The conscious veto is a control function, different from simply becoming aware of the wish to act. There is no logical imperative in any mind-brain theory, even identity theory, that requires specific neural activity to precede and determine the nature of a conscious control function. And there is no experimental evidence against the possibility that the control process may appear without development by prior unconscious processes.

Libet’s case here (and in other sections of his paper) for the independence of the conscious veto from neural activity is not that there is positive evidence for it, but merely that logical and empirical considerations don’t rule it out. But even this is too strong a claim, for surely under mind-brain identity theory, any discoverable conscious control function must be neurally instantiated. And evidence acceptable to the neuroscientific community would inevitably show the connection of such a control function to other brain functions, whether conscious or unconscious. The startling fact is that Libet wants to find a non-neural, non-physical basis for free will (some sort of mental conscious control over the brain itself) and he wants to find it doing research predicated on the assumption of neural cause and effect. Such a research agenda, wedded to the a priori goal of defeating mechanism yet rooted in physicalist science, is doomed from the outset.

What exactly, one might ask, is the threat of mechanism, that it can so distort normal scientific constraints on what constitutes a plausible hypothesis and that would make proponents of the hypothesis plead that absence of disconfirmation counts as positive evidence? For Libet, as for many others, the underlying concern is that if the self reduces to the brain, and conscious control reduces to a sub-system of the brain, then the individual doesn’t have real, contra-causal free will (as Dennett 1984 puts it, the type of free will Libet thinks is ‘worth wanting’) and can’t be held responsible:

We do not hold people responsible for actions performed unconsciously, without the possibility of control. For example, actions by a person during a psychomotor epileptic seizure, or by one with Tourette’s syndrome, etc., are not regarded as actions of free will. Why then should an act unconsciously developed by a normal individual, a process over which he also has no conscious control, be regarded as an act of free will?

Similarly, Stapp warns us that

It has become now widely appreciated that assimilation by the general public of this ‘scientific’ view, according to which each human being is basically a mechanical robot, is likely to have a significant and corrosive impact on the moral fabric of society… [involving] the growing tendency of people to exonerate themselves by arguing that it is not 'I’ who is at fault, but some mechanical process within…

Libet thinks that unless conscious control (something brain-independent, remember) plays a role in either generating or vetoing normal behavior, then normal behavior is essentially no different from movements resulting from seizures or tics. But clearly we can, without resorting to obscure notions of mental agency, continue to make the distinction between voluntary, deliberate behavior in which consciousness plays a role and reflexive, involuntary acts. (Gomes this volume does an excellent job of describing the spectrum of intentional and impulsive behavior while relating them to neural function, illuminating the territory that Libet must conceal to push his agenda.) With this distinction in hand, we are at least part way to establishing a plausible, albeit compatibilist and non-libertarian, notion of personal responsibility: Individuals who can deliberate, anticipate consequences, recall relevant episodes, and who are otherwise rational are capable of change by virtue of being held responsible, in the sense that the anticipation of being praised and blamed is effective in shaping their voluntary acts (very young children and the mentally incompetent the obvious exceptions). Therefore, we should hold them responsible – treat them as moral agents – in order to encourage the sorts of behavior we want. So even though no individual on this view has anything resembling the ultimate free will or conscious control Libet hopes to find, it makes perfect sense, as well as good interpersonal, social, and legal policy, to suppose we are justified in holding most individuals responsible for their voluntary behavior (Clark, 1998a, 1998b). If this is true, then mechanism (deterministic or incorporating indeterministic elements), even if it penetrates to the core of the self, is no threat to responsibility, morality, or the social order. Having understood this, scientists such as Libet and Stapp and others exploring physical theory such as Hodgson need not beat the bushes quite so hard to discover a basis for ‘ultimate’ responsibility.

For those of the libertarian persuasion, the first person phenomenology of choice is all but indisputable evidence for free will. For Libet, the experience of agency is not to be questioned, and any science that fails to validate it is suspect:

[W]e must recognize that the almost universal experience that we can act with a free, independent choice provides a kind of prima facie evidence that conscious mental processes can causatively control some brain processes… The phenomenal fact is that most of us feel that we do have free will, at least for some of our actions and within certain limits that may be imposed by our brain’s status and by our environment. The intuitive feelings about the phenomenon of free will form a fundamental basis for views of our human nature, and great care should be taken not to believe allegedly scientific conclusions about them which actually depend on hidden ad hoc assumptions. A theory that simply interprets the phenomenon of free will as illusory and denies the validity of this phenomenal fact is less attractive than a theory that accepts or accommodates the phenomenal fact.

Let us grant the point (although it’s eminently contestable, as Claxton and McCrone show) that most of the world’s inhabitants feel as if they are free agents in the radical, libertarian sense, able via some mental process to affect their brains without in turn being at the effect of anything else. When conducting science, the validity of such a feeling – whether or not it refers to an actual state of affairs - is exactly the question at issue, one would suppose. Do we in fact exist as something apart from the brain that could manipulate its very neurons, or sub-neural mechanisms? The truth of this proposition presumably gets established by winning scientific consensus via the preponderance of empirical support. But here Libet seems to have things precisely backwards: judge the attractiveness of a theory by how well it conforms to one’s subjective experience, which is taken as criterial. As Claxton properly observes, the phenomenology of volition, insofar as it actually feels as Libet supposes, might merely reflect the consensual Western ‘folk theory’ of free will, since ‘once such a view has been culturally adopted and become ‘second nature’, then perception itself becomes skewed and selective…and persistent interpretations self-reinforcing’[4th graph]. The feeling of freedom, therefore, hardly counts as evidence for libertarian free will, nor should the failure of science to validate this feeling reflect upon the adequacy of contrary theories. The methodological point here is the rather prosaic one that first person phenomenology, however widely shared, can at most be the basis for a hunch, which may get lucky and become a plausible hypothesis, which may yet become a full fledged theory, if the empirical cards fall in its favor. But feelings or intuitions per se never count as self-evident proof of anything.

Like Libet, Jeffery Schwartz combines (with difficulty) a practical commitment to neuroscience with a sharp ideological aversion to the notion that the will might be embodied. Unfortunately, the result is that both his science and his philosophy end up compromised. He conceives of human effort (in this instance, the efforts by obsessive-compulsive disorder (OCD) patients to overcome their symptoms) as something categorically non-physical and mental, yet emphasizes ‘the critical role of effort as a necessary component for keeping the machinery [of the brain] on track and functioning’ (p 9). This sets up the standard (and usually fatal) problem of dualism: how exactly can something non-physical influence the physical machinery of the brain? Should an interactive mechanism ever get specified, this threatens to absorb the non-physical antecedent into the physical consequent. If it doesn’t get specified, then it’s mere speculation that the non-physical has such power, and methodologically more circumspect to suppose that physical causes, as yet unknown, are doing the work.

On the one hand, Schwartz is conversant with the neurophysiology underlying the symptoms of OCD, and indeed his main findings are that there are measurable differences in brain metabolism between responders and non-responders to behavioral treatment for OCD (p 10). On the other hand, he credits the change in metabolism to an entirely obscure, theoretically disconnected (from neuroscience, and hence predictively inert) notion of ‘mental force’ that ‘generates the energy necessary to activate, strengthen, and stabilize [a responder’s] new health-giving and life-affirming circuitry’.

Schwartz’s motive is much the same as Libet’s: defend, at all costs, the intuition of mental agency, of an essentially active, willed power independent of the brain and body which can reach down and manipulate the brain itself. And again, the validity of the experience of effort and volition as referring to a categorically mental, non-neural phenomenon goes unquestioned. Rather, science must conform to experience and take the verbal reports of patients at face value (‘the directly perceived reality of the causal efficacy of volition’), otherwise we would be ‘resorting to reductionist approaches which irrationally prefer materialist as opposed to experiential perspectives’. But there is no real conflict between such perspectives, as long as we understand that phenomenology is data, just like meter readings, that needs explaining, not the benchmark of empirical truth (see Dennett, 1991 on ‘heterophenomenology’).

Schwartz’s difficulty is that he conflates passivity at the behavioral and neural levels:

Without reference to a naturally occurring mental force, the observed changes in cerebral energy use…would have to be autonomously generated by an entirely passive process – but that is plainly inconsistent with a very large amount of clinical data, most especially the verbal reports of patients who have actually undergone these treatments (original emphasis)

Presumably, patients said such things as ‘I really tried hard not to let that obsessive thought run away with me.’ But there is nothing inconsistent with the experience of trying hard and having effort itself be a ‘passive’ process in the rather strained, non-standard sense that its neural correlates are entrained without benefit of active supervision. In this sense, all brain events are passive, since no one supposes we are actually in a position to consciously pick out and activate particular neural bundles or connections. Nevertheless, effortful, deliberate, focussed action arises, distinguished from passive responding not by the additional effort supplied by ‘mental force’ (which would need explaining in turn), but by the specific sorts of neural networks involved. This distinction, of course, is not enough for Schwartz, since his concern is to buttress ‘the age old belief that human beings have in their capacity to act as genuinely self-directed agents capable of instituting real self-directed change’ (my emphasis). Mere embodied brains, apparently, can’t change on their own in response to social influences like behavioral therapy; they need extra help from the mental force. To which I am tempted to respond: may the force be with you, Dr. Schwartz; I have no need of it.

Henry Stapp starts by emphasizing the momentousness of the debate on mind/body problem. It is, he says, a ‘raging battle’ which pits our intuition that conscious thoughts guide behavior against philosophies which ‘proclaim, in the name of science, that we are mechanical systems governed entirely by impersonal laws that operate at the microscopic level of our atomic or cellular components.’ Having dealt with the fear of mechanism above, I will comment on Stapp’s characterization of mechanism, which seems simply wrong as applied to ourselves or anything much above a mollusk.

The worst case scenario, for Stapp and other libertarians, is if it turns out there’s nothing more to us than deterministically law-governed meat, whether it be bones, brain, muscle or gut. Now the laws governing the meat (that is, the more or less invariable or statistically reliable covariations between states of affairs we call natural laws) are not merely at the microscopic atomic or cellular level, as Stapp supposes, but also obtain at successively higher levels of biological and behavioral description, each of which incorporates whatever levels lie beneath it. There are laws, for example, of bipedal locomotion control common to robots and humans (left right, left right) which have little or nothing to do with the instantiating physical substrate. And there are, perhaps, psychological laws having to do with motivation and goal satisfaction (e.g., the conditions which generate ‘disappointment’) which would obtain for silicon Martians and vaporous Venusians as well as for us protoplasmic Earthlings. The point, elaborated by Dennett (1987) is that organisms of our complexity (naturally evolved mechanisms amenable to behavioral prediction and control) are intentional systems in which law-governed behavior is best understood, and practically speaking only understood, at the personal level of beliefs and desires, not the ‘machine’ level. But this means that on a materialist functional view of the self, there is no gulf needing to be bridged, nor any conflict, between the impersonal component level and the highest personal level of deliberate choice: true explanations can co-exist at both levels. And descriptions of the functional components of complex, conscious behavior, linked to the brain’s cybernetic control systems, are showing great promise, even in their infancy (see for instance Baars, 1997; Churchland, 1995; Damasio, 1994, Ingvar, this volume).

Stapp rejects this sort of materialism, in particular the notion that conscious experience of the sort commonly associated with our deliberative choices might be identical to some set of neural processes. But, since he believes consciousness must play an important causal role in shaping behavior (if it didn’t, we’d be robots) he must, like Libet and Hodgson (see below), find the physical, or more broadly, the natural correlates of consciousness somewhere else. This is in the interstices of quantum mechanics, of which I know little and on which I will therefore remain silent (see Churchland & Rush, 1995 for an extended critique of quantum theories of consciousness).

Nevertheless, there is a general point to be made relevant to Stapp and others engaged in the quantum quest for consciousness and free will: when you finally pin it down, will it still be what you want? Stapp wants a ‘pragmatic theory of the mind/brain that allows our thoughts to be causally efficacious yet not controlled by local-mechanistic laws combined with random chance.’ But any good scientific theory of thought will of necessity show how thoughts themselves arise within a context, not merely how they cause behavior. That is, such a theory will place thought, consciousness, the will, mental agency, what have you, within a general, law-invoking explanation in which there are predictable antecedents as well as consequents. But libertarian free will is precisely that which by definition can’t have law-like or predictable antecedents, otherwise it wouldn’t be free in the required sense. So to find it within nature is to destroy it. Since knowledge explodes the causa sui self, its defenders tend to occupy the penumbra of science and psychology, where the light of explanation has yet to penetrate.

Most of the motives and methods critiqued above can be found in Hodgson’s sally against mainstream science and philosophy, undertaken for the now familiar reason of defending ‘common-sense ideas of responsibility’. Hodgson, like Kane in his 1996 book The Significance of Free Will and a recent paper (Kane, 1999), tries to establish the possibility that a person’s motives, reasons, and character never necessitate choices, they only incline a person in particular directions. It is the sheer ‘capacity to choose’, the essential, stripped down ‘I’ or self, not anything distinctive about the person, which finally determines behavior. During the choice process, this I, or what Hodgson calls ‘volitional causation’, goes about ‘adjusting the probabilities of alternatives’. In sum,

The way the agent is, in respect of character and motivation, does not determine what the agent does: it only determines what the alternatives are and how they appeal. The agent is not responsible for having the capacity to choose between these alternatives, but it is not constrained in how that capacity is exercised, even by the way the agent then is in respect of character and motivation; and so the agent (and no-one and nothing else) is responsible for the way that capacity is exercised (original emphasis).

The final redoubt of free will is therefore nothing publicly recognizable as me, but something very much like a dispassionate ‘soul’, operating behind the scene of motivational conflict. But what interests, one might ask, does such a soul-chooser have, that would explain why it adjusts the probabilities of alternatives (the weights of reasons) in one direction rather than another? Apparently none, since the agent, in its capacity as chooser, has no character and motives, rather it operates over such mundanities, in which case the adjustments themselves are inexplicable. Nevertheless, they still get chalked up to the responsible agent. The price of ultimate responsibility it seems, is the intelligibility and explicability of freely willed choices (for a discussion of this point, see Clark, 1996).

But Hodgson’s claim is that our standard notion of explanation, that which seeks a causal context for action, perhaps leavened by various sorts of randomness, is ‘Hume’s mistake’. Hume, the arch skeptic about folk theories of causation, along with most of the philosophical and scientific community, somehow failed to see there is yet a third alternative beyond chance and deterministic causality, namely ‘volitional causation’. However, even if we accept Hodgson’s argument that such causation is a conceptual possibility (a stretch by my lights) it’s not a mistake by Hume and the rest of us compatibilists to spend little time considering it, since we have no need of such an hypothesis. That is, since we suppose we have a good account of responsibility and agency of the non-ultimate variety, that does all the work necessary to ground personal, social, and legal practice, it’s methodological wisdom not to waste time on what seems prima facie a metaphysical non-starter. Hodgson, driven by the libertarian assumption that only a radically free self can be held responsible, must play Don Quixote versus mainstream science and philosophy. (Of course that’s how compatibilists feel when taking up the lance against this same assumption as it gets expressed in Western culture. For both sides the romance of being the gadfly underdog is irresistible).

Mohrhoff, in his well-written essay, takes the problem of freedom to be an offshoot of the so-called hard problem of consciousness (see Sheer, 1998). If, as some contend, consciousness is causally epiphenomenal (the conjoined claims that experience qua experience serves no function above and beyond neural functions, and that it is not identical with neural functions; see Clark, 1995; Papineau, 1998) and free will is a property or an attribute of consciousness, it looks as though free will might be illusory. Mohrhoff’s exposition concentrates on showing that mind-body interactionism of the sort which might confer causal priority on the mental - the power to influence without being influenced in turn - is not inconsistent with physical theory. Since I’m not competent to assess Mohrhoff’s analyses of physical claims, I will simply comment that his conclusions do not give much comfort to dualists. Mohrhoff admits that

Although there are no compelling theoretical or experimental reasons why mental events should not be capable of causing departures from physical laws, it may remain difficult for interactionists and proponents of free will, at least for some time, to disabuse the contemporary physicist, biologist, biologist, or philosopher of science of the doctrine of physicalism…

The reason for the obduracy of contemporary scientists is simply that although Mohrhoff may have adduced arguments supporting the consistency of some vague notion of mental causation with physical theory (e.g., ‘modifications, by the conscious self, of the electromagnetic interactions between particles’) neither he nor anyone else to date has given a convincing positive account of mental causation backed up by scientific findings. Until an experimentally testable model of how a ‘self’ can modify electromagnetic fields is forthcoming, few scientists are likely to jump on the interactionist bandwagon. However, if research on ‘psi’ powers or directed prayer (e.g., see Dossey, 1997) starts to show a robust, replicable effect and a testable hypothesis connecting such an effect to the rest of science is forthcoming, then interactionism might get rolling. Until such time, we all would be well advised not to disabuse ourselves of physicalism as the methodologically most parsimonious and productive first guess about phenomena.

Wilson usefully discusses the biological mechanisms which a non-physical mind might employ to influence the brain, and concludes (agreeing with Mohrhoff) that any such mechanism must involve violation of basic physical laws such as the conservation of energy, and further that it must involve energy above the quantum mechanical level. Wilson actually gets his feet wet in the biological mind/brain interactive territory where Libet, Schwartz, and Stapp fear to tread, which of course is essential if we are to discover a positive account of dualism. But in Wilson’s view no such account is forthcoming any time soon, and he summarizes nicely what might be called the naturalistic challenge to libertarians:

For those who believe that free will must be totally independent and free of physical causes, the idea that physical laws must be violated should not be taken as a negative but almost as an expectation, especially to the extent that physical laws appear to specify a universe that is either determined or randomly probabilistic. If a non-physical mind exists, the research project for the next century should be to explore the impact of such non-physical influences – where in the brain does such influence occur and what laws are broken?

The challenge, which I think is insuperable, is to stay within naturalism (i.e., no ad hoc, mysterious, immeasurable, or theoretically disconnected forces or entities such as your typical incubus or angel) and yet find something causally privileged, something that operates as a first cause, unmoved mover, or is self-constructing from the ground up. [footnote: Or from the sky down - see Dennett’s Darwin’s Dangerous Idea (1995) for an extended discussion of ‘skyhooks’: miraculous devices that are brought in to do the heavy explanatory lifting when reductionist explanations are deemed too unpalatable. Libertarian free will seems the paradigm example of a skyhook.] This indeed is a research project that we should defer until the millenium (starting 2001, remember) if not the following century, or until such time it is shown not to be completely misconceived.

Meanwhile, research on the mind, conceived as more or less the brain in action, proceeds apace. Here David Ingvar surveys the literature on brain mapping studies of volition, while Wolfram Schultz reports on the activation of single neurons during self-initiated acts. Although the physiological levels investigated could hardly differ more between these two papers (large brain structures vs. individual cells) the implicit assumption common to both is that will, intention, and voluntary control of behavior are physically instantiated. The will, as quantifiable and locatable, is what the brain does in translating desire into action, and it arises in the neuronal context of the entire nervous system, from perceptual inputs to motor outputs, which in turn is securely embedded in the physical and social environment. Of course, none of this research has the least relevance for libertarian free will, since it’s all about how internal cybernetic mechanisms respond, more or less deterministically, to environmental contingencies. (Although its of interest that Robert Kane thinks that some sort of indeterminism, generated by conflicting neural processes, may turn out to be the basis for a free will incompatible with determinism; see Kane 1999.) But for compatibilists, for whom the will need not be independent of mechanism, such studies offer the beginnings of evidence that the brain indeed has the neural resources to accomplish what, until recently, only a controlling mental agent was thought capable of.

Although they pretty much take for granted a compatibilist definition of free will, neuroscientists Spence and Frith are refreshingly non-ideological in their description of the brain systems subserving voluntary behavior and intention. Without a philosophical axe to grind, they are free simply to describe biology on the one side, and behavior, including verbal reports, on the other. No mental supervisor is needed, on their account, to help the brain sort through the pressing demands of acting on time-limited opportunities. Instead, attention is focussed, priorities are set, and choices made by various brain sub-systems themselves, interacting with the environment. In particular, Spence and Frith explore the feeling of freedom, which is intimately connected with the sense of ‘ownership’ of one’s body and the covariation of intention and action, both of which they show to be matters of neurobiology, not mental agency. The sense of being a particular, freely choosing self is plausibly tied to how our brains and bodies produce unique, character-specific behaviors, rather than to being something like Hodgson’s stripped down soul.

In reaching such conclusions, Spence and Frith’s paper combines with Gomes’, Claxton’s, and McCrone’s contributions to create a formidable rebuttal to mentalistic and dualist theories of free will. Gomes digs very deeply and forcefully into Libet’s data, recognizing that ‘the dualist thesis [Libet’s] is of course not a solution to the dilemma’ of how to reconcile naturalism with our sense of voluntary agency. Having cut the libertarian knot, Gomes sets out the compatibilist alternative in its methodological context:

[F]rom the third person perspective – which is anyway the proper perspective for explanation – there is not so much difficulty in considering choice, decision and action as part of the natural world…All we need to suppose is that there is, in human beings, a decision system that can represent actions and action sequences before their performance, that can select among them, and the output of which is not fully determined by its input, but also by its internal state, by its representations of aims to be achieved, by internal criteria that affect its activity (moral and other personal values), and also by a certain degree of randomness (which gives the arbitrary character that our choices often have).

Given this brain-based decision making system, the next compatibilist move is to identify with it. As Gomes goes on to say:

If the ‘I’ is such a system as we have roughly described above, we can consider this intuition [that I act freely] to be essentially correct. It all depends on the concept we have of the self, of the ‘I’. When we see our actions as determined by ourselves, this [intuition of freedom] can be considered to be right. It is when we consider our self to be pure spontaneity – a being that is not subject to causality – that we are in illusion.

Such compatibilism gives Gomes a clear advantage over Libet in making sense of Libet’s own data on how readiness potentials precede voluntary, conscious choices. Gomes doesn’t have to side-step the obvious interpretation, that consciousness and deliberate acts are, in toto, brain-based phenomena. The guiding assumption – that the mind just is the brain in action – helps ground Gomes’ somewhat tortuous, but ultimately rewarding analysis of intention, consciousness, voluntariness, and deliberation and how these are linked in various ways. He combines careful attention to the phenomenology of choice with a nuanced typology of choice-making acts, concluding that we act with the most (compatibilist) free will when we deliberate – that is, consciously engage our intentions – concerning the real alternatives facing us.

Although Claxton certainly agrees with Gomes that we don’t have contra-causal, incompatibilist freedom, he might contest the notion that we are most free when carefully deliberating. His aim (largely achieved, I think, but of course I’m very much in his camp) is nothing less than to demolish the myth he sees at the heart of the Western conception of free will: that intentions, generated by an ‘instigatory self’, cause ensuing behavior. He makes a plausible case that the virtues of conscious control are overrated, and its influence over voluntary behavior vastly inflated. Claxton elegantly dissects the phenomenology and metaphysics of choice, exposing the assumption of a supervisory ‘I’ as a fraud and, occasionally, a positive hindrance to effective behavior.

What’s most striking about Claxton’s approach is his painstaking insistence that we call into question our culturally given assumptions about the whole issue of free will. We must, he says quite rightly, be extra vigilant not to imagine that we already know the truth of the matter, since our natural tendency is to confabulate a role for the freely willing self given the least opportunity. If we really pay attention, Claxton says, we will find that thoughts, intentions, and feelings arise on their own (for an extended exploration of the personal and social implications of this see Breer, 1989). These arisings are appropriated by the phenomenological impression of being an instigatory self, an impression which helps maintain the self/other distinction and the sense of personal continuity - both useful, perhaps, as day-to-day working hypotheses.

But isn’t this a dangerous, even unpatriotic, admission: that we’re not really in control, that we can’t be trusted not to ‘run amok’? Yes, if we continue to identify with the now disempowered conscious controller, but emphatically not if instead we take Claxton’s recommendation (and Gomes’) to identify with the cognitive processes embodied in our brains. (Which are, after all, so idiosyncratically us: no two brains are alike; disinterested souls are like peas in a pod.) Brains, suitably connected to bodies, and given the right sorts of upbringing, control themselves quite nicely, it turns out; no nanny necessary. The paralyzing effect of excess conscious control might lift, Claxton suggests, should we accept this reality. The only problem, of course, is that brains, unlike souls, don’t go to heaven to live the life hereafter, which is perhaps one reason we’re not so keen to identify with them. (For materialist reassurances on the fear of death, see Clark, 1994.)

But are we just our brains and bodies? In the only classically philosophical essay of this set, Lowe argues that the self or person, although physically embodied, cannot be identified with the body or any part of it, such as the brain. Furthermore, since Lowe believes the self’s mental intentions are categorically not physical states, but nonetheless have physical consequences (actions), it must be the case that non-physical states can cause physical events. Although I and many others have sought to undermine the intuition that the mental just can’t be the physical (see Clark 1995; Papineau, 1998), let us grant Lowe his major conclusion: that under his ‘naturalistic dualism’ non-physical beliefs and desires are essential in causing and explaining human behavior, and that therefore thoroughgoing physicalism is false. What follows from this regarding free will?

Perhaps libertarians (such as Schwartz, above) will take heart from the fact that the mental does, after all, control the physical. Since it’s the sheer physical inexorability of impersonal mechanism that seems so alien to common intuitions of freedom, Lowe’s location of causal power within a self’s non-physical intentions may buy some sort of psychological relief or cognitive consistency, especially for those who strongly interpret the phenomenology of choice as the mental ‘I’ controlling the physical body. Well and good, but what about the issue of ultimate freedom? Are intentions any less conditioned by the context in which they arise simply by virtue of their non-physical status? Does the self and its attributes, even if not identical with the brain and body, have a self-constructing freedom independent of its surroundings? It is not clear that it would, short of demonstrating, which Lowe does not, that the non-physical creates itself ‘out of context’, so to speak. There seems to be no obvious way in which splitting the natural world into two domains, the physical and non-physical, helps to ground libertarian free will.

In his account of free will, McCrone, very much like Claxton, seeks to show that our internalized sense of agency is a socially-induced artifact, which can vary tremendously across cultures. The most recent Western version, McCrone suggests, derives from the Romantic rebellion against the Enlightenment, a rebellion which reified the will as a ‘pure motive force’, something which could counter the constraints of the clockwork universe described by 18th century science. At the close of the millenium, we have the curious paradox that radical individualism, expressed in the ideology of libertarian free will, is itself a culturally reinforced norm. And further, McCrone notes, following B.F Skinner’s crucial insight from Beyond Freedom and Dignity, the norm is actually a means of control:

It is central to the script of modern Western society that moral agency lies within and that this self is essentially free. Even our legal, political and educational systems depend on the assumption that it is fair to treat individuals as point-like moral agents, fully in charge of what comes out their own minds…[T]he clear goal is to keep pushing individuals until they can live up to the model of autonomous behavior set by Western culture…[T]he message is to take charge so society can judge you hard for the quality of your self-regulation and decision-making.

Here’s the irony: the Western myth of the causa sui self (Skinner’s ‘autonomous man’) licenses the culture to control behavior by exacting stiff after-the-fact sanctions against the individual, since, after all, the individual (self-originating first cause, unmoved mover) is alone responsible for authoring acts. By fostering the illusion of free will, the conditions which create individuals and shape choices are largely ignored as factors in behavior, setting the person up to take the fall, or take all the credit. Now, without disputing the fact that controls are necessary, the question arises as to whether this often punitive, and sometimes disproportionately rewarding system is really the best. Is feeding the fiction of libertarian freedom, and its attendant blindness to causality, any way to run a culture? Although McCrone doesn’t take up this question, his (and Claxton’s) analysis of the social construction of free will should prompt its reexamination, nearly 30 years after Skinner so impolitely raised it.

Resistance to any reconsideration of free will runs deep, as testified by the libertarian contributions in this volume. Not many, perhaps, are ready to challenge the ‘consensual Cartesian trance’ (Claxton) and take the heat for daring to suggest that 1) we aren’t unmoved movers, and 2) we don’t need to be in order to hold people responsible, ground moral judgments, and maintain a just, democratic social order. As mentioned at the start of this commentary, the notion of free will has ramifications for our most fundamental conceptions of self, agency, and responsibility, and to these we are deeply attached. We do want credit, we do enjoy imposing just deserts, we can’t, realistically, give up entirely those ‘reactive attitudes’ (Strawson, 1962) such as gratefulness and resentment which seem to point to the ‘instigatory self’. As much as free will exposes us to the threat of unlimited retaliation for wrongdoing, it nevertheless compensates us by making us the lords of our little domains, the micro-gods of our minds. Contra-causal freedom pays us the ultimate compliment, even if sometimes it exacts the ultimate price, by making us finally responsible for ourselves.

There is the truth of the matter about free will, which must be approached with all due humility (again, Claxton eloquently says why), and then there is the practical issue of whether the truth, in this particular case, is something we are ready to face. Some here have argued that freedom is directly perceived in the irrefutable experience of choice, but others, including myself, take the truth about free will to be a matter of where a scientific understanding of ourselves leads, and experience is simply data added to the mix. To decide between these two approaches is to decide between intuitionism and empiricism, or between personal modes of knowing versus collective and experimental modes. I suspect there is no final arbitration of this issue, except to point out that one can pitch intuitionist arguments successfully to those, like much of the public, that are prepared to buy them, but to convince scientists and (most) philosophers, you’d better have a theory linked to some institutional, peer-reviewed wisdom (see Clark, 1992 and 1993). So if libertarians want to get funded for research, they’ll have to come up with a plausible naturalistic model for an unmoved mover (not likely) and if compatibilists like myself want to change the public’s conception of free will, we first have to change attitudes about what counts as evidence (not likely).

That libertarians search science to confirm their interpretation of the experience of volition indicates that they too feel the empiricist pull. They, just as much as compatibilists, want the satisfying unification of intuition and observation, the personal and the impersonal perspectives. My claim is, however, that as much as libertarians want validation by science, it will never be forthcoming since the very notion of the ultimately responsible self is inherently opposed to scientific objectives of explanation and prediction. In short, there’s a conceptual conflict at the deepest level which blocks this sort of cognitive unification for libertarians. And for that reason, I pity them.

They will reject my pity, and reply that I am dangerously confused in supposing that our commonsense concept of personal liberty should evolve to become compatible with determinism. Whatever the truth about free will may be, too much is at stake, in terms of our social and legal institutions, our self-esteem, and our personal power and creativity to relinquish this most central of assumptions: that we alone choose ourselves and our futures. Better to finesse the science indefinitely by clinging to the straw that just because there’s no good evidence for free will, that doesn’t constitute proof that it doesn’t exist. Since science can’t ever pronounce the definitive death of ultimate freedom, why not persist in our libertarian convictions?

There are many reasons why not, but mere absence of evidence is perhaps the least persuasive reason for those convinced that the causa sui self simply must exist. First, the fear of mechanism must be diffused, by showing that a causally embedded self can be a moral agent, responsive to the value-reinforcing effects of social sanctions and rewards (indeed, only such a self is fully responsive; a causa sui self is, by definition, not). Responsibility doesn’t have to be ultimate to justify praise and blame, which means a causal understanding of voluntary acts isn’t tantamount to excusing them, as is often supposed. Nor does the fair application of sanctions need a basis beyond the fact that behavior is indeed modified by its consequences. Punishing others for my sins is unjust because they are not the ones needing correction; their suffering is undeserved - non-functional and needless - because the punishment fits no crime of theirs that needs extinguishing, in the Skinnerian sense.

Beyond the moral issues, compatibilists must show that determinism is no threat to human efficacy and creativity: our personal power derives not from having some mysterious causal priority over circumstances, but from exploiting the causal context of which we are so inextricably a part. Being ultimate self-choosers would merely tie us in knots, as we tried to discover some basis for taking that first step towards self-definition. Being proximate self-choosers, on the other hand, is all we need to fulfil the desires nature and nurture bequeath us in such variety (see Dennett, 1984; Clark, 1999).

And finally, there may well be benefits flowing from a thoroughly naturalistic conception of the self and its choices. The retributive impulse, cut off from its metaphysical justification in free will, might soften, leading to a less punitive culture. More attention might be paid to improving the social conditions shaping individuals, and no longer will policy makers so blithely blame the victim (remember the multitudes in the U.S. who ‘chose’ to be homeless during the Reagan era?). On the personal level, dethroning the supervisory ‘I’ might help us become less self-conscious, more playful, and less likely to wallow in excessive self-blame, pride, envy, or resentment (see Breer 1989).

Might we become less ambitious, once we see that we don’t ultimately choose ourselves or our projects, and that our successes (and failures) result from thousands of combining circumstances? Perhaps, but this might be all to the good, given that the unfettered accumulation of wealth seems likely to compromise the long-term sustainability of resources, or at least concentrate them in a very few hands. And after all, we need not worry that putting the self in its natural, causal context will extinguish desire, any more than we need worry that it will undermine our rights and liberties. Our selves, physically embodied, are virtually constituted by desire, and real freedom lies in having the opportunity to pursue our motives as we discover them arising in us. Seeing that the self neither has, nor needs, ultimate responsibility for itself may well lead to the more responsible use of such freedom.

©  Thomas W. Clark, 1999


Baars, B. (1997), ‘In the theatre of consciousness’, Journal of Consciousness Studies, 4 (4), pp. 292-309.

Breer, P. (1989), The Spontaneous Self: Viable Alternatives to Free Will (Cambridge, MA: Institute for Naturalistic Philosophy).

Churchland, P.M. (1995), The Engine of Reason, the Seat of the Soul (Cambridge MA: MIT Press)

Churchland, P.S., Grush, R. (1995) ‘Gaps in Penrose's toilings’, Journal of Consciousness Studies 2 (1), pp. 10-29.

Clark, T. (1990), ‘Free choice and naturalism’, The Humanist, 50 (3), pp. 18-24.

Clark, T. (1992), ‘Relativism and the limits of rationality’, The Humanist, 52 (1), pp. 25-32.

Clark, T. (1993), ‘Faith, science and the soul: on the pragmatic virtues of naturalism’, The Humanist, 53 (3), pp. 7-12).

Clark, T. (1994), ‘Death, nothingness, and subjectivity’, The Humanist, 54, (6), pp. 15-20.

Clark, T. (1995), ‘Function and phenomenology: closing the explanatory gap’, Journal of Consciousness Studies 2 (3), pp. 241-55. Reprinted in Sheer, 1998.

Clark, T. (1996), ‘The freedom of Susan Smith’, The Humanist, 56 (2), pp. 8-12.

Clark, T. (1998a), ‘Materialism and morality: the problem with Pinker’, The Humanist, 58 (6), pp. 20-25.

Clark, T. (1998b), ‘To help addicts, look beyond the fiction of free will’, The Scientist, 12 (16), p. 9.

Clark, T. (1999), ‘The flaw of fatalism’, on the World Wide Web at the Fatalism page.

Damasio, A. (1994), Descartes’ Error (New York: Putnam).

Dennett, D. (1984), Elbow Room (Cambridge, MA: MIT Press).

Dennett, D. (1987), The Intentional Stance (Cambridge, MA: MIT Press).

Dennett, D. (1991), Consciousness Explained (Boston: Little, Brown).

Dennett, D. (1995), Darwin’s Dangerous Idea (New York: Simon and Schuster).

Dossey, L. (1997) Healing Words : The Power of Prayer and the Practice of Medicine (New York: Harper Collins).

Kane, R. (1996), The Significance of Free Will (New York: Oxford University Press).

Kane, R. (1999), ‘Responsibility, luck, and chance: reflections on free will and determinism’, Journal of Philosophy, XCVI (5), pp. 217-40.

Papineau, D. (1998) ‘Mind the gap’, in Philosophical Perspectives, 12: Language, Mind and Ontology, James E. Tombelin, J.E., ed., pp. 373-388 (Oxford: Blackwell Publishers).

Sheer, J. ed. (1998) Explaining Consciousness: The Hard Problem (Cambridge, MA: MIT Press)

Strawson, G. (1998), ‘Luck swallows everything’, Times Literary Supplement, 26 June, pp. 8-10.

Strawson, P. (1962), ‘Freedom and resentment’, Proceedings of the British Academy, XLVIII, p. 1-25.

Waller, B. (1998), The Natural Selection of Autonomy (Albany: State University of New York Press).