Do advances in neuroscience threaten the idea of free will, and if so, what practical implications does this have, for instance when it comes to criminal responsibility and punishment? In a stimulating talk at the Uehiro seminar (the podcast of which is available here), Frej Klem Thomsen, assistant professor of philosophy at Roskilde University, discussed the answers that the prominent American neuroscientists Joshua Greene and Jonathan Cohen have proposed to those questions [1]. Briefly put, Greene and Cohen predict that cognitive neuroscience will make it increasingly apparent to everyone that (as some philosophers have argued centuries ago already) there is no such thing as free will as commonly understood. This, they add, will shift the approach to punishment in criminal law from the current “retributivist” one to a consequentialist one – a change they also judge desirable, on the grounds that the current approach relies on intuitions they take to be scientifically untenable.
The retributivist approach to punishment emphasizes the idea, intuitively plausible to many of us, that criminals deserve to be punished for their crimes, whereas the consequentialist approach justifies punishment by reference to its beneficial consequences for society as a whole. Thomsen summarized Greene and Cohen’s argument as follows: the commonsense view about the justification of criminal punishment is retributivist in nature, which means that it treats desert as a necessary condition of justified punishment. Desert presupposes moral responsibility, which itself presupposes free will. But as cognitive neuroscience is bringing to our attention with increasing vividness, we have good reason to think that there is no free will, because every single one of our actions is fully determined by causal chains that extend back in time beyond our individual decisions. [2] But if there is no free will, there can be no moral responsibility, and therefore no such thing as desert. Therefore, if we accept the retributivist justification of punishment, no one can ever be justifiably punished, no more than a car can deserve to be punished for refusing to start (as in this movie). From here on, we can either stick to retributivism, and conclude that punishment of any kind is never justified, or, as Greene and Cohen recommend, opt instead for a consequentialist approach to punishment that does not presuppose free will (at least as commonly understood).
One common way of responding to Greene and Cohen’s argument would be to defend a “compatibilist” view about free will. Compatibilists hold that free will is fully compatible with determinism. There are various forms of compatibilism, but for the sake of simplicity let me focus on one well-known variant of the view. On this variant, we are free to the extent that our possibilities for action are not impeded by either internal or external obstacles: e.g. it is not the case that we are paralyzed, or that someone is holding a gun to our head and ordering us to do something. Incompatibilists, by contrast, have more stringent standards for what is to count as acting freely, and as a result regard determinism and free will as incompatible. Compatibilism offers the promise of rescuing the retributivist justification of punishment against Greene and Cohen’s argument: even if, as neuroscience is said to suggest, we are never the ultimate cause of our actions, this does not challenge the reality of free will, if free will doesn’t require the falsity of determinism.
Greene and Cohen actually agree that compatibilism is the view we should prefer on pragmatic grounds, yet when it comes to finding out the truth about free will and responsibility, they think that incompatibilism is intuitively more plausible. They use the following thought experiment to support their view: imagine someone (call him Mr Puppet) whose genome, and every significant event in his life (including the way others treated him), were carefully designed in advance by a team of evil scientists. Mr Puppet ends up committing a murder, just as the scientists had intended. Greene and Cohen suggest that Mr Puppet should not be held fully responsible for the murder, since both the internal and external factors which together inexorably led him to commit it were entirely crafted by other people. But, they add, we are all in a similar situation to Mr Puppet, since our actions, too, are the result of the interaction between our genetic endowment and the environments we happen to find ourselves in. The only difference is that in our case, these things weren’t designed by a third party (unless perhaps God exists). This difference, however, changes nothing to the fact that if Mr Puppet isn’t responsible for his actions, then none of us is either. If so, contra compatibilism, determinism does undermine free will and responsibility after all.
Thomsen considered other possible replies to Greene and Cohen’s argument, with a focus on those challenging the idea that retributivism presupposes free will (in the incompatibilist sense). He argued that these replies looked unpromising. First, one might deny that desert presupposes moral responsibility. Imagine that some country runs a state lottery where the name of Andrea, a citizen of that country, is drawn as the winner, even though she is in no way responsible for this (e.g. she didn’t choose to take part in the lottery). Isn’t it nevertheless appropriate to say that she deserves the lottery prize, since her name was drawn? While accepting that it might be, Thomsen responded that what inclines us to say that Andrea deserves the prize is the idea that she is entitled to it. But in cases involving punishment, he argued, entitlement no longer seems enough for desert. For instance, if the state in question had chosen to impose instead a ten-year prison sentence on the lottery winner, we wouldn’t want to say that Andrea deserved the prison sentence, even though she had won the lottery. A second strategy discussed by Thomsen, which he referred to as “semi-compatibilism”, challenges instead the assumption that moral responsibility presupposes free will, and takes the former to be compatible with determinism, even though free will might not be. Thomsen showed, however, that semi-compatibilism seems just as vulnerable as full-fledged compatibilism to counterexamples such as that of Mr Puppet.
Considering the failure of these responses, Thomsen appears to endorse Greene and Cohen’s conclusion, according to which we should abandon the retributivist justification of punishment while still retaining, on consequentialist grounds, the practice of punishment and the notion of criminal responsibility. Thomsen thus plausibly argued that doing so would help deter offenders, reform prisoners, and secure public support for the criminal justice system, given the importance many attach to the notion of responsibility. However, this proposal raises a question: how do we retain the notion of criminal responsibility once we have established that the notions of moral responsibility and free will on which it is founded are mere illusions – especially if neuroscience makes these conclusions apparent to more and more people? One way would be to try and cash out criminal responsibility without referring to moral responsibility, e.g. by equating it with the capacity to act rationally, but Thomsen himself agrees that in the context of criminal punishment, we are not interested in rationality for its own sake but rather as a proxy for moral responsibility. The alternative would be to follow Greene and Cohen’s suggestion to embrace compatibilism about free will (at least in the context of criminal law), in spite of the objections they raise themselves against it. Yet if – as Greene and Cohen themselves believe – most of us share incompatibilist intuitions, and if incompatibilism is in fact the correct approach, it is unclear that we can reasonably be expected to endorse compatibilism on sheer consequentialist grounds. What is proposed, after all, is to keep punishing people even though no one really deserves to be punished! It would thus seem preferable if we could defend compatibilism against purported counterexamples, so that we could sincerely embrace it.
Here is one way in which we might attempt to do this in the Mr Puppet case. It seems that part of the reason why such an example is likely to elicit incompatibilist intuitions is that it is, to some extent, ambiguous. We are told that the evil scientists have scripted everything that Mr Puppet’s relatives, friends, peers etc. should say to him, and how they should treat him. Yet does this mean that they treated him in much the same way each of us was treated by his/her own relatives and friends, or that the sort of social conditioning he experienced amounted to downright manipulation, perhaps to the point of brainwashing? It may be that when reading about his case, we tend to assume the latter. Yet on this second reading, we should not conclude that we are all in a similar situation to Mr Puppet’s. Assuming that most of the people we interact with throughout our lives are not unscrupulous manipulators, the circumstances in which we are placed do not threaten our moral responsibility in the way the conditioning process Mr Puppet is subjected to threatens his. To warrant their claim that we are indeed all like Mr Puppet, Greene and Cohen ought to flesh out the example so as to make it clear that only the first reading above is appropriate. My guess is that, were they to do so, it would weaken the force of that example as a source of support for incompatibilism.
[1] In Greene, J. and Cohen, J. 2004. For the Law, Neuroscience Changes Everything and Nothing. Phil. Trans. R. Soc. Lond. B 359, 1775-85.
[2] We can leave aside here the role of quantum indeterminacies, as they simply introduce a degree of randomness that cannot salvage the commonsense understanding of free will.
Interesting article, I’ll listen to the podcast later.
Humans are certainly creatures of will, but I find it hard to understand what adding the word “free” to that is supposed to signify. Willful decision-making is itself a deterministic process, which is what distinguishes it from dice-throwing. The wilful decision is often consciously made in accordance with various criteria that help determine its outcome. In the case of moral decisions, this may be a particular set of moral principles. But many other wilful decisions (such as the decision to consciously employ this or that set of principles) mostly reflect the inclinations of the individual as determined by factors beyond his own design or control. Most decisions may be the result of a mixture of conscious use of relevant deterministic criteria, and criteria supplied by the person’s inherent nature. None of this exercise of will can be described as “free” if one means free of deterministic constraints – and it’s hard to see what such a freedom could represent except complete randomness.
“Thomsen thus plausibly argued that doing so would help deter offenders, reform prisoners, and secure public support for the criminal justice system, given the importance many attach to the notion of responsibility. However, this proposal raises a question: how do we retain the notion of criminal responsibility once we have established that the notions of moral responsibility and free will on which it is founded are mere illusions – especially if neuroscience makes these conclusions apparent to more and more people?”
Our genetically determined nature is usually socially dynamic and responsive to different ideological influences in different ways. If it’s in the inherent nature of most people to respond to ideas like justice and responsibility in ways that help constrain their behaviour for the social good, then it’s obviously important to retain and persuasively propagate those ideas. As for punishment, it can serve as a deterrent in the case of those people (who may be a high proportion of the population) who would otherwise be more inclined to commit crimes. There may also be people who are oblivious to such deterrent effects but such people tend to represent a chronic danger to others, and can thus be justifiably incarcerated on that basis.
Abandoning simplistic notions of free will may result in more humane conditions of custody for people who are only marginally responsible for their own behaviour, and unravelling the neuro-cognitive basis of chronic anti-social behaviour may result in effective corrective therapies.
“Abandoning simplistic notions of free will may result in more humane conditions of custody for people who are only marginally responsible for their own behaviour, and unravelling the neuro-cognitive basis of chronic anti-social behaviour may result in effective corrective therapies.”
See my earlier comment below, Nicholas. I think part of the problem is that we tend to respond to each other in direct interpersonal ways, rather than in impersonal, analytic ways and when it comes to crime and punishment that tends to translate into a generic “lock ’em up and throw away the key” attitude. Perhaps “abandoning simplistic notions of free will” is part of the solution, but if so it surely has to be at most a very marginal and indirect part. Becoming more aware of r emotional reactions (generally, and on this case towards crime and its perpetrators specifically) seems more promising.
I find this an interesting piece as well, just as Nikolas Shaffer does. The intuitionist view can be very strong, especially in crimes that have a high emotional content attached to them.
What the article points to is how if our internal and external guides are indeed prompts, and can be shown to be so by close examination, then those influences must be allowed to have a bearing on how we interpret action.
This should lead us to question our intuition to punish and shift the retrobutionist grip currently dominant in many societies and cultures. America and other western societies and cultures self-report an explosion in the numbers of people held in prisons. To take steps to reduce prison numbers and scale down punitive responses, I argue, would be a positive step forward. I acknowledge a move to challenge intuition can is a hard road to adopt. Self-righteousness is a potent part of how social views of crime, prisons and punishment operate.
The article offers a hypothetical individual called Mr. Puppet. He has internal influences and external ones, both of which lead him to eventually take another person’s life. A very serious outcome indeed. One might view death as the most serious outcome, however that may be challenged in the minds of some.
I note in recent articles (The Guardian, and items on this website) material offered that suggests the pedophile has desires that are of the same kind as our Mr Puppet – that is they are not chosen prompts but operate just as desire operates in other sexual orientations (what is being said here is his desire is no different from our own). Second the social position of the pedophile I would describe as very poor, he is subject to very strong social condemnation and that social attitude is not conditional on sexual activity (in this aspect let us assume his situation is very different from our own).
I wonder, does making Mr. Puppet a pedophile change a readers thinking? I do not support the idea sexual contacts between adults and children, where force is absent, is worse than death; nor do I think the presence of force or coercion makes sexual contacts between adults and children is worse than death. Allow me to be very clear, I do not see these two as the same.
I put forward this modification in the Mr. Puppet account because in the case of pedophilia the contribution to moral judgement seems very strongly intuitionist based. Beliefs people hold resists factual details that challenge those beliefs. Public debate, as in the recent media articles in the Guardian, and the discussion in other spaces strongly suggests this.
It is, in my view, a positive step to gain a better idea of how to construct a justice system that fits what we can confirm by open questioning. It is a much weaker outcome to retain social practices which may be ‘comfortable’ to keep, but which are at odds with what our social science research and philosophical reflections demonstrate.
I think it extremely unlikely that neuroscience will ever prove that «every single one of our actions is fully determined by causal chains that extend back in time beyond our individual decisions». Even if one could, for example, conclude that (as in the nursery rhyme) the war was lost because of a horse-shoe nail, it is not equivalent to saying that the loss of the nail FULLY DETERMINED the cause of the war. There are too many other potentially determining factors at play.
A more fundamental point, and one which in my view is the best response to the incompatibilists, is the following :
If we accepted that every one of our actions were fully determined by such causal chains, it seems to me that we we would no longer be able to draw a distinction between acting because a gun was pointed at our heads and acting without this constraint – if ALL is «determined», then what reason would we have for considering that the former is different from the latter?
Perhaps there are some people who think that there is no difference between the two cases. (If they were right, then we would have to change a huge deal of our thinking and practice concerning justice and responsibility and not just our views on punishment.)
I conclude the opposite : the fact that we value the distinction implies that we do indeed have good reason to think that there is free will.
Nikolas and Anthony: thank you for your comments.
Nikolas: in response to your claim that “it’s hard to see what such a freedom could represent except complete randomness”: I guess an incompatibilist might reply that the relevant sort of freedom would consist in an agent’s initiating, out of her own free choice, new causal chains in the world (presumably starting with some events in her brain) that weren’t fully determined by antecedent causal chains; and that this wouldn’t count as randomness, to the extent that the agent would be controlling the start of the new causal chain, e.g. her decision to perform a certain action. It wouldn’t be the same as if her decisions were determined by a lottery (i.e. a process where no control is exerted over the outcome).
Anthony: you may well be right that neuroscience will never be able to prove this, in a strict sense of “proof”. Still, it might provide strong grounds for accepting that proposition. This might for instance be the case if our future observations of the workings of the brain systematically failed to reveal anything like a brain event that appeared to be a “causa sui” of the sort just described.
Regarding your two examples, an incompatibilist might accept that there is a difference between the two: namely, the first involves coercion, whereas the second one does not. However, she would add that when it comes to how free you are in both cases, there is indeed no difference, since in both cases you’re fully determined to act a certain way, and therefore not free. I agree with you that the implications of such a view at the practical level seem unappealing, hence my attempt at defending compatibilism against the “Mr Puppet” counterexample.
“I guess an incompatibilist might reply that the relevant sort of freedom would consist in an agent’s initiating, out of her own free choice, new causal chains in the world (presumably starting with some events in her brain) that weren’t fully determined by antecedent causal chains”
I’m sure that does indeed happen all the time – the human’s engagement with the world is itself a dynamic deterministic intervention that introduces all kinds of new causal threads, and these are consequences of the exercise of human will. But I’m content to use the term “will’ rather than “free will”, because human will itself is deterministically constrained by the criteria that make it willful rather than random (and “free will” is usually pronounced to mean some kind of non-deterministic process).
I’m not convinced. One of the drivers of concepts like free will and retributive justice is our tendency to respond to each other in a directly interpersonal way rather than seeing each other (or ourselves) as impersonal objects of analysis. Enhanced understanding of neuroscience will not in itself change that (except to the extent that it will lead us to reengineer ourselves to an extent that we in any case probably don’t want). What it will do is to empower those of us who prefer consequentialist approaches to refine them, and hopefully indeed facilitate humane correction and prevention. But as long as it still makes sense to regard ourselves and each other as single entities evolving through time (and this is certainly a fiction to which we seem to be quite attached), I think we will continue to hold each other morally responsible for past actions.
Comments are closed.