Against Moral Responsibility and Retributive Justice
Free will compatibilism fails to justify belief in moral responsibility, so we should learn to do without it.
The most fundamental and active debate in the philosophy of free will pits free will compatibilists against incompatibilists. The debate focuses on the issue of moral responsibility. Compatibilists believe that people can be held morally responsible for their actions even if human behavior is deterministic, whereas incompatibilists believe that determinism requires abandoning moral responsibility. Examples of compatibilist thinkers are Daniel Dennett, Owen Flanagan, and John Martin Fischer. Incompatibilists include Derk Pereboom, Gregg D. Caruso, and Robert Sapolsky.1
To hold someone morally responsible means to deem them deserving of praise or blame because of the moral value of their actions. If we believe that someone deserves blame for a murder because it is immoral to commit murders, then we are holding them morally responsible.
Moral responsibility requires free will. When we hold someone morally responsible, we are saying that they deserve praise or blame because they freely chose to act morally or immorally.
The notion of free will emerged from religious ideas that are no longer deemed scientifically defensible. Christian philosophers like Immanuel Kant conceived of the soul as an immaterial entity that was self-causing, rather than being caused by anything else. This idea of free will underlies Kant’s theory of moral responsibility.
Both compatibilists and incompatibilists have abandoned the religious theory of the soul and free will. Neither side believes in an immaterial soul that is not subject to physical law. Rather, both sides of the debate accept a deterministic theory of human behavior, according to which human actions are the result of physical law. I have discussed and defended the deterministic theory of human behavior here.
Compatibilists believe that the sort of free will necessary to ground moral responsibility is compatible with determinism. Incompatibilists, who are often called “hard incompatibilists” or “hard determinists,” believe that this sort of free will is not compatible with determinism, and that moral responsibility must, therefore, be rejected.
If the incompatibilists are right, then major revisions to common human practices are warranted. The notion of moral responsibility is central to human moral evaluation: as we shall see, judgments of moral responsibility are an essential component of many of our emotions. Also, moral responsibility is one of the major justifications of criminal punishment. Moral responsibility underlies the retributive theory of punishment, which holds that people deserve punishment when they commit the type of serious moral wrongs that we call crimes, and that they deserve to be punished in proportion to the severity of their crimes.
I will argue here that the incompatibilists are right and that moral responsibility must, therefore, be rejected, along with our common practices of moral evaluation and criminal punishment. I will go on to discuss how these common practices can be reframed and revised so that they are philosophically defensible.
Compatibilism
In an engaging essay, the popular philosophy/neuroscience Substacker Tommy Blanchard defines and defends compatibilism. While much more detailed and closely argued defenses of compatibilism exist, Blanchard’s discussion is sufficient to illustrate the nature and weaknesses of this theory, as long as one makes an additional tweak.
Like all compatibilists, Blanchard bases his theory on the ideas of intention and reasons-responsiveness. An intentional action is one that you consciously will. Reasons-responsiveness designates the capacity to be influenced by rational considerations, including moral ones. Compatibilists believe that if you are a reasons-responsive person who intentionally commits an immoral action, then you have chosen that action of your own free will and are morally responsible for it.
Blanchard’s theory explains much about our intuitive judgments of moral responsibility. Cross-cultural research has shown that large majorities of people believe that humans have free will and deserve to be held morally responsible for their actions. Moreover, compatibilism captures our intuitive notion that there is a moral difference between harming someone intentionally and unintentionally. If you kill someone by pushing them in front of an oncoming train, you are a criminal. But if someone behind you pushes you so that you can’t help but push someone else onto the train tracks, you would probably not be judged morally or legally responsible for any harm that resulted.
Similarly, it is intuitive that only a reasons-responsive person is morally responsible for their actions. If someone gives you a drug that so thoroughly disorders your thoughts that you cannot tell right from wrong, you may be acquitted by a jury for any harm that you cause in this state. Reasons-responsiveness is the basis of insanity and diminished responsibility legal defenses. (Legal responsibility is often a form of moral responsibility, and I will use it in this sense throughout this essay.)
However, Blanchard does not discuss an additional component of our intuitive sense of moral responsibility: the effect of factors beyond our control. Consider the following case. Susan is a mentally normal, reasons-responsive adult who works as a bank teller. One day she gets a call from an abusive ex-boyfriend who tells her that he has kidnapped Susan’s child. He will release the child only if Susan steals money from the bank. So she steals the money, gives it to the boyfriend, and gets her child back.
In this case, a reasons-responsive adult intentionally commits a crime, but no one would hold her morally or legally responsible for the crime because she was acting under what the legal system calls “duress.” Rather, we would assign responsibility for the action to the boyfriend. We would say that Susan’s action was determined by factors beyond her control. So a more complete version of compatibilism would define a freely willed action as an intentional action performed by a reasons-responsive person free from the pressure of factors beyond their control.
Incompatibilism
The idea of factors beyond our control is at the heart of the incompatibilist case against moral responsibility. Incompatibilists recognize the moral significance of intention and reasons-responsiveness, just as compatibilists do. However, incompatibilists hold that our actions are ultimately explained by factors beyond our control so that we cannot be held morally responsible for what we do.
Compatibilists are right that the conscious, willing self is the origin of much that people do. However, incompatibilists believe that we are not responsible for the self that acts. The philosopher Galen Strawson argues that we do not choose who we are because the self is formed by the interaction of genes with environment. Of course, none of us chooses our genes. We also do not choose the childhood environments that do so much to shape our characters. As adults, people normally can choose their environments, but any choices that they make are causally determined by the selves that they have become due to the interaction of their genes with their childhood environments. Because factors beyond our control are the ultimate cause of our actions, we can no more be held morally responsible for our actions than Susan can for the bank robbery.
Incompatibilism is a radical challenge to many aspects of the intuitive moral outlook. However, it does make sense of other aspects of moral judgments that compatibilists downplay or neglect entirely. By now, most of us accept the scientific evidence that genetic endowment combined with childhood environment forms our adult character. A person with a low IQ who grows up in an abusive family and a high-crime neighborhood is many times more likely to become a criminal than one who has the opposite background. The incompatibilist case is that no one chooses to have a low IQ or grow up in an abusive household or any of the genetic and environmental factors that make us who we are. Therefore, no one can ever be ultimately morally responsible for their actions.
Two thought experiments
I will make my case for incompatibilism through two thought experiments. The first is an abbreviated and slightly modified paraphrase of Derk Pereboom’s manipulation argument and the second is original to me. You can read Pereboom’s original manipulation argument here.
The manipulation argument begins with a case of wrongdoing that everyone would acknowledge is not blameworthy because it is caused by factors beyond the agent’s control. In subsequent examples, the cause of the wrongdoing is changed so that the action comes progressively closer to meeting what compatibilists believe are the criteria for blameworthiness. The point of the thought experiment is that there is no relevant moral difference between the first and last examples.
In all the examples below, Andy is a normal human capable of forming intentions and responding to reasons.
Example A: Neuroscientists have implanted a chip in Andy’s brain, and they can activate the chip by radio technology. When the chip is activated, Andy will be transformed into a violent and heartless egotist who will inevitably and deterministically kill someone for money. While he is capable of responding to moral arguments, none of them will dissuade him from murder. The neuroscientists activate the chip, and Andy kills someone for money.
Is Andy responsible for the murder? Even though he is a reasons-responsive adult who intentionally kills someone, I think most people and juries would not hold him morally responsible, as his action was determined by a factor beyond his control.
Example B: Neuroscientists using novel surgical techniques program Andy’s brain at birth so that he will become a violent and heartless egotist as an adult who will be inclined to kill people. As an adult, he kills someone for money.
Again, I think it would be wrong to hold Andy responsible for the murder. If Andy had not been programmed by the neuroscientists, he would likely never have become such a monster.
Example C: Andy’s parents are both violent and heartless egotists who frequently murder people. They train Andy as a child to believe that murdering someone for money is morally justified and even admirable. They also shield him from other viewpoints. He grows up to be a violent and heartless egotist who murders someone for money.
In this case, Andy is capable of judging right from wrong, but, because of his upbringing, his notion of what constitutes right and wrong is very different from most other people’s. Because Andy is the victim of hideous brainwashing, it would be wrong to hold him responsible for the murder.
Example D: Andy grows up in a universe where people are molded by their genes interacting with their environments according to deterministic processes. Because of these factors, Andy becomes a violent and heartless egotist who kills someone for money.
Because Andy is a normal adult who intentionally kills someone, compatibilists would normally hold him responsible for the murder. But what has really changed? Andy’s behavior is no less determined by factors beyond his control in this case than in the others. If you believe that Andy is not morally responsible for the murder in Example A, there would seem to be no justification for a different conclusion in Example D.
One problem with this thought experiment is that people are accustomed to attributing moral responsibility to human beings, so the use of one in the examples perhaps might bring out people’s prejudices. I have come up with another thought experiment that will perhaps be even more convincing because it does not elicit this prejudice.
Example A: ACME Inc., an industrial conglomerate, builds many types of machines, including cranes. One day, one of the cranes malfunctions and drops a girder that kills someone. ACME is held legally responsible for the death.
Example B: ACME builds self-driving cars with very sophisticated functioning that mimics many human abilities, like visual cognition and route-planning. One day one of the cars malfunctions and kills someone because of a flaw in its software. ACME is held legally responsible for the death.
Example C: ACME builds a robot named Jake who is indistinguishable from a human being. Jake has a human-like ability to form intentions and respond to reasons. Jake is programmed to work hard, take great pride in his work, and feel human emotions like anger. One day someone unfairly mocks Jake for doing low quality work. Jake throws a paperweight at his tormentor and kills him. A subsequent examination of the data reveals that Jake was capable of distinguishing right from wrong when he did this. Who is legally responsible for this death, Jake or ACME?
Example C is more ambiguous than the previous examples. It is clearly absurd to hold a crane or a car morally or legally responsible for killing people, but, because Jake so closely resembles a human being, it is more plausible to hold him (it?) morally responsible. However, I do not think that a prosecutor would have much difficulty in convincing a reasonable jury that ACME is at fault here. After all, why would ACME design Jake without including a “never kill people” rule?
The upshot is that, just as ACME builds machines, people are built by the interaction of genes with environment. People are, therefore, never morally responsible for their actions, just as the ACME machines are not.
Incompatibilism and criminal justice
Even though moral responsibility cannot be philosophically justified, it still plays a crucial role in our moral judgments and criminal justice system. If incompatibilists do not propose plausible alternative justifications for these practices, critics can claim that this philosophy may be justified in theory but is impractical.
We saw above that belief in moral responsibility is the basis of retributive justice, or the theory that a criminal punishment is justified because of desert. To atone for their crimes, criminals deserve to suffer in proportion to the severity of the crimes. Retributivism is fundamental to current day and historical theories of justice. The lex talionis or the “eye for an eye” practice of punishment is an ancient form of retributivism.
Retributivism is a backward-looking philosophy of punishment in that criminals are punished based on what happened in the past. Retributivists also believe that punishment is intrinsically good rather than good for its positive consequences.
Incompatibilists believe that no one deserves to suffer for their actions because they are not ultimately responsible for them. Incompatibilists commonly justify criminal incapacitation through the right to self-defense rather than desert. Just as we are justified in harming someone to prevent them from harming us, so we are sometimes justified in incapacitating criminals to keep the public from harm. Incompatibilists also believe that criminal incapacitation is not intrinsically good, but good because it produces positive effects. For this reason, their theory of criminal justice is forward rather than backward-looking.
The incompatibilist philosopher Gregg D. Caruso proposes a public health model of criminal justice. This approach posits that criminals are no more deserving of punishment than people who suffer from a contagious disease. Most of us would regard it as grotesque to punish people for having involuntarily contracted a disease, but we would regard it as defensible for the state to forcibly quarantine disease carriers to prevent contagion and facilitate the treatment of the disease. Caruso believes that the detention of criminals and other forms of incapacitation are justified on a similar principle: detention prevents them from doing further harm and facilitates efforts to rehabilitate them. If a criminal is incapable of reform, then indefinite detention is justified. However, detention cannot be justified once a criminal has established that they are no longer a threat to the public. The public health model also requires the state to do everything that it can to eliminate the sources of crime, just as the state takes steps to prevent disease. Thus, the public health model demands that the state should work to eradicate risk factors for crime, like homelessness, mental illness, and poverty.
Pereboom argues that some deterrence justifications of criminal punishment are also consistent with the right to self-defense. If the threat of punishment effectively deters crime, it can be justified as long as it does not inflict more suffering on the criminal than is necessary for deterrence. Pereboom gives the example of financial crimes. He thinks that if these crimes can be deterred effectively by threatening potential criminals with the suspension of their professional licenses, then this punishment is to be preferred over a prison sentence, since it causes criminals less suffering.2
Incompatibilists usually favor criminal justice systems like that of Norway that minimize the sufferings of criminals and emphasize rehabilitation. Norway’s prisons are famously comfortable and are sometimes compared to two-star hotels. Norway also makes executions and life-sentences illegal. Some have argued that such a criminal justice system will not have a sufficient deterrent effect, but in fact recidivism rates are lower in Norway than they are in the USA, where the criminal justice system is much crueler.
Incompatibilism and the reactive emotions
In the classic 1962 article “Freedom and Resentment,” compatibilist philosopher P.F. Strawson (not his son Galen Strawson, cited earlier) argued that incompatibilism is unworkable for humans because judgments of moral responsibility are intrinsic to what he calls the reactive emotions, like resentment, indignation, guilt, and gratitude.
If someone treads on my hand accidentally, while trying to help me, the pain may be no less acute than if he treads on it in contemptuous disregard of my existence or with a malevolent wish to injure me. But I shall generally feel in the second case a kind and degree of resentment that I shall not feel in the first. If someone's actions help me to some benefit I desire, then I am benefited in any case; but if he intended them so to benefit me because of his general goodwill towards me, I shall reasonably feel a gratitude which I should not feel at all if the benefit was an incidental consequence, unintended or even regretted by him, of some plan of action with a different aim. (49)
You resent someone not only because they cause you harm, but because you judge them morally responsible for that harm. You feel guilty not only because you did harm, but also because you feel morally responsible for it. Strawson believed that incompatibilism must be rejected because it invalidates our natural emotions and results in a cold and unemotional “objective attitude” towards ourselves and others when we deal with moral problems.
Pereboom elaborates various defenses against this objection. The first is that there are many emotional responses available to us that do not imply moral responsibility. Instead of resenting wrongdoing, we can learn to greet it with shock, disappointment, sadness, and protest. Just as in the case of criminal punishment, incompatibilists focus on exerting a positive effect on immoral people rather than blaming them for something that happened in the past. Incompatibilists believe that we should react to moral wrongdoing as parents react to their children’s misbehavior. Parents normally administer punishment to their children not to express resentment, but to improve their future behavior.
If our reactive emotions are too powerful to alter in this way, we can allow ourselves to feel them while at the same time recognizing that they are irrational. People often feel emotions that they know are irrational: someone may know intellectually that flying is safe but still be emotionally afraid of flying. Incompatibilists can learn to develop this kind of complex attitude towards reactive emotions.
Finally, Pereboom points out that overcoming reactive emotions may bring about improvements in our emotional lives and behavior. Reactive emotions like resentment and indignation are inherently angry, and anger is often counter-productive when dealing with moral problems. Psychological research has established that angry people are prone to make mistakes in judgment, like overlooking mitigating circumstances and relying on stereotypes. Directing anger at others also often makes them less likely to do as we would wish them to. There are many circumstances, then, in which we are better off not responding to wrongdoing with reactive anger.3
Against Moral Responsibility and Retributive Justice
Although judgments of moral responsibility may be natural to us, it seems clear that they are philosophically unfounded and do more harm than good. Retributivism is especially influential on criminal punishment in the United States, and it is becoming increasingly accepted that the American criminal justice system is cruel, ineffective, and excessively expensive. The widely followed blogger Scott Alexander did a deep dive into the scholarship on the relationship of incarceration to crime and found that long prison sentences are an expensive and relatively ineffective way of reducing crime. Since incompatibilism would normally result in shorter prison sentences and milder punishments, adopting this philosophy might save taxpayers money and result in less cruel treatment of prisoners.
Belief in moral responsibility requires that legal scholars, judges, and juries spend a lot of time thinking about what criminals do and don’t deserve. Such calculations of desert are also a common feature of moral reflections: we wonder what we should do to get back at someone for sinning against us. Incompatibilism suggests that all of these cogitations are likely a waste of time, as the whole notion that people deserve punishment is invalid. Rather, we would be better off thinking about how to improve people’s future behavior rather than making them atone for the past.
In the essay, I have tried to link to sources that are available online. However, my views were mainly formed by books. The books that I have found most useful in understanding incompatibilism and the free will debate are:
Derk Pereboom, Free Will (New York: Cambridge UP, 2022). This is a great short introduction to the free will debate by an important philosopher.
Derk Pereboom, Wrongdoing and the Moral Emotions (Oxford: Oxford UP, 2021).
Gregg D. Caruso, Rejecting Retributivism: Free Will, Punishment, and Criminal Justice (New York: Cambridge UP, 2021). This is a massively researched book that thoroughly addresses a multitude of philosophical issues related to retributive justice from an incompatibilist perspective.
Robert M. Sapolsky, Determined: The Science of Life without Free Will (Penguin, 2023). Sapolsky builds the case for incompatibilism with a wealth of scientific detail, then switches to philosophical argument in the last part of the book.
John Martin Fischer and Mark Ravizza, Responsibility and Control: A Theory of Moral Responsibility (New York: Cambridge UP, 1998). This is one of the most sophisticated defenses of the compatibilism.
Pereboom, Wrongdoing, pp. 83-96.
Ibid., pp. 26-52.
Thanks for this article! This is really good, I think you've significantly raised the bar for hard determinist takes on Substack.
Overall, to be clear, we agree more than we disagree. Most importantly, we both reject the spooky libertarian kind of free will that breaks the laws of physics. To some extent we're probably splitting hairs on things like what language to use around moral actions and judgments, and just choosing to emphasize different things.
But to get to the heart of our disagreement: I'm not convinced by the manipulation argument. I've always found it (like many slippery-slope arguments) a bit like saying black and white can't be different because there are shades of grey in between. To me it seems clear that at each step along the argument we're putting in a few more pieces for standard moral responsibility.
But probably more importantly, I don't see how we can just put moral responsibility aside. Our Strawsonian reactive attitudes, values like justice, and even our need for forward-looking/consequentialist deterence and praise all need some concept of accountability. We need to separate intentional and unintentional actions, and fully capable adults who should be held accountable from children who should not. All of these things seem to cohere pretty well with a framework like reasons-responsiveness (or something close to it). Maybe the claim is that the manipulation argument should temper our responsive attitudes and the value we place on justice. Maybe it does for some people, that's an interesting empirical question. I doubt it's possible to completely eliminate them, but at this point we're just talking about speculations around human psychology. None of this changes our need to hold each other accountable to interact socially and cooperate in a larger society, and as long as there is that, we have moral responsibility.
Anyways, seriously, this was a great article, you summarized the hard determinist case and the consequences of it really well. Best article I've read on Substack on free will (excluding my own, of course 😉)
Really enjoyed this. Determined, by Robert Sapolsky, was gifted to me over the holiday season. I'll likely get to that in the next few months, between my mathematics studies and other readings. I'm not yet well enough read in this area to provide any real comments, but I'll get there hopefully. This article helped. Thanks.