By Ryan Lake

Winner of the Outstanding Graduate Paper Award at the 55th Annual Meeting of the Florida Philosophical Association

Ryan Lake, University of Miami

Is it permissible to punish people for crimes that they haven’t committed yet? Intuitively such a practice seems grossly unjust to say the least. Before committing a crime, a person is still innocent of the crime, and it is of course immoral to punish the innocent. However some, such as Christopher New (1994), have argued that this is a baseless temporal bias. Granted, epistemic limitations may prevent us from ever actually punishing people before they commit crimes. But if we could predict with a reasonable degree of certainty that a person was going to commit a crime in the future, then situations may arise, New argues, in which it is desirable to prepunish. More recently, Saul Smilanksy (2007) has tied the question of prepunishment to the free will debate. Smilansky argues that there is a principled way to resist the temptation of prepunishment, but that this strategy assumes the falsity of determinism, and hence is open only to the libertarian, and not to the compatibilist. Smilansky concludes that compatibilism thus winds up being much more radically revisionist about morality than its proponents would like, thus strengthening the case for incompatibilists (both Libertarians and Free Will Skeptics) who argue that common sense morality cannot be reconciled with determinism.

In what follows, I will offer a response on behalf of the compatibilist. First, I will lay out the case for prepunishment (from New). I will then consider Smilansky’s argument that prepunishment exhibits a lack of appropriate regard for people as persons, and argue that contra Smilansky that prepunishment is as much an issue for the libertarian as it is for the compatibilist – either Smilansky’s favored strategy is available to both, or it is available to neither. And in the final section, I will gesture at some considerations that weigh against prepunishment generally, and may provide a way to resist prepunishment that is somewhat different from what Smilansky suggests and should be open to compatibilists (as well as libertarians).

I. The Case for Prepunishment

New offers an example to illustrate when prepunishment might be acceptable, if not required. Imagine a person, Algy, who intends to and actually is going to speed tomorrow. Both Algy and the local officer Ben have this knowledge, 1and they both know that if Ben does not issue a citation for the speeding violation today, before the offense has occurred, Algy will skip the country and never be fined. So Officer Ben issues Algy a ticket the day before the crime, which Algy pays. The next day he goes on to break the speed limit just as described in the citation. Is there anything wrong with what Ben does in this case?

One natural objection that springs to mind is that it is wrong to prepunish Algy because until he commits the crime he is still innocent. But New claims that we can distinguish two versions of this basic moral intuition, one which prohibits prepunishment and one which allows it. One version is that it is wrong to punish someone for a crime which he never commits, and the other is that it is wrong to punish a person for a crime which he has not committed yet, but intends to and actually will commit. The first version is less controversial and intuitive, but it doesn’t prohibit prepunishing Algy, since he does commit the crime (in the near future). The latter version prohibits punishing Algy, but it is much less obvious that it is correct. And as New points out, there appear to be moral considerations in favor of punishing Algy before the crime in this case. After all, if we don’t prepunish him, he gets to commit the crime and get away with no punishment at all, which ought to be morally repugnant to anyone with retributivist leanings. New argues that the fundamental intuition here is that there must be some connection between actual guilt and the punishments we inflict. We must be made to pay for our offenses. Whether we happen to pay for them before the crime or after the crime is, New argues, entirely beside the point.

Another possible objection New considers is that the sort of case under discussion is not actually a case of prepunishment, but an ordinary case of post-punishment. That is, we might suppose that what Algy is being punished for is not his future crime of speeding, but his forming the intention to speed. However, this response doesn’t actually apply to the case we are considering. Whatever we may think about punishing people for their intentions, in the given case Ben writes a citation for the act of speeding that Algy will commit the next day. But still we may suppose that New’s case for prepunishment implicitly relies on the fact that Algy has already formed the intention to commit a crime. Would the case for prepunishment stand if we removed Algy’s intention? We can imagine that Algy knows that he will speed the next day without having yet formed the intention to speed. 2 Or we can modify the case further and suppose that Algy isn’t even aware that he is going to speed tomorrow. Nonetheless, Officer Ben knows with certainty that Algy is in fact going to speed tomorrow, and that unless we fine him now, he will skip the country and we will never have the chance to issue a ticket. It seems that here the same considerations in favor of prepunishment apply as in New’s original example – that Algy is going to earn the fine, that there is no special reason not to deal out the punishment before the crime rather than afterwards, and that since we can’t punish him after the fact, the only way that justice can be served is if we punish him now. So for the moment let us suppose that it doesn’t matter whether Algy already intendsto commit the crime (a point I will return to later). The considerations that New offers in favor of prepunishment seem to stand with or without the criminal already having formed the intention to break the law.

II. Smilansky’s Argument against Prepunishment

Smilansky’s objection to prepunishment can be stated simply. He argues on essentially Kantian grounds that we must respect the future criminal as a moral agent, specifically as an agent “capable of not committing the offense.” 3 Prepunishment, Smilansky argues, violates this basic principle by not giving the agent an opportunity to refrain from carrying out the criminal act in the future. For instance in New’s case, Officer Ben is treating Algy as a mere object to be dealt with, rather than as an autonomous moral agent whose autonomy must be respected. Only by giving Algy the chance to decide (perhaps at the last moment) that he should do the right thing and drive the speed limit do we fullyregard Algy a person, in particular a person with the capacity for improving his moral character and doing his duty. Prima facie this seems like a highly plausible objection to New’s defense of prepunishment. However, I won’t spend time here exploring the merits of Smilansky’s solution to the prepunishment temptation. Rather, I would like to focus on the implications that Smilansky alleges that this case holds for the free will debate.

Specifically, Smilansky argues that the objection to prepunishment that he has offered is not open to the compatibilist about free will and determinism. To see why, consider New’s example again, this time under the assumption that causal determinism is true. According to the standard compatibilist, Algy still may be fully morally responsible for his act of speeding, in the sense that he is blameworthy, and therefore deserving of whatever punishment people can deserve for such a crime. For the compatibilist it makes no difference that Algy’s behavior is causally determined. Now suppose that we do have the means to calculate with complete certainty that Algy is going to break the speed limit the next day. Should we prepunish him? According to Smilansky the compatibilist has no principled way to say no. At any rate, he seems unable to offer Smilansky’s objection – that our respect for Algy’s moral autonomy demands that we allow him the opportunity to change his mind, because it is already causally determined that Algy is not going to change his mind. Therefore the compatibilist seems committed to accepting the practice of prepunishment (in at least in the sort of case under consideration) and this seems like a substantial revision of ordinary morality. Hence the compatibilist’s standard claim that determinism makes little or no moral difference is refuted.

Before turning to the compatibilist response, a distinction is in order. We may understand Smilansky in one of two ways. We may understand him as saying that the compatibilist cannot offer the respect-for-the-agent’s-autonomy line simply because the agent in fact will not change his mind. Or, we may understand Smilansky’s claim to be that the compatibilist cannot offer this line because the agent is incapableof changing his mind.

If we read Smilansky the first way, then the problem is not just a problem for compatibilism. There at least two ways in which the libertarian may (in principle) have to deal with possible present truths about what an agent will in fact do in the future. First, the most sophisticated forms of libertarianism that are around today allow that much (if not most) of the time, a free agent’s behavior is causally determined by his character. For instance, Robert Kane argues that only a very small subset of our actions – what he calls “self-forming” actions – is in fact indeterministic. According to Kane the rest of our actions flow deterministically from our character. Nonetheless Kane regards these actions as ones that are done freely and for which we are responsible, so long as the character they flow from is one that we formed via our properly indeterministic self-forming actions. In the above case, we may suppose that being a speed demon is a deeply entrenched part of Algy’s character, and that his act of speeding tomorrow is therefore causally determined by his present character. According to a libertarian like Kane, Algy is still responsible for his action. But then since Algy’s act of speeding is causally determined, such a libertarian – like the compatibilist – cannot offer Smilansky’s defense (on this first reading of it) against prepunishment.

Furthermore, even the more radical libertarian who argues that only indeterministic actions can be done freely and responsibly isn’t off the hook. Let’s assume that determinism is false, and that Algy has this sort of libertarian free will, meaning that none of his (free and responsible) actions are causally determined by anything. So when Algy decides to speed, his act is not determined by anything that has gone before. And let’s assume further an eternalist or “block” theory of time. 4 As far as I can see, there is nothing inconsistent or incoherent about these two assumptions. 5 There are simply facts about what Algy will do in the future, though what Algy does is still entirely of his own libertarian free will, not causally necessitated by anything earlier. Now suppose there is a being of some sort (God, a psychic, a time traveler) who tells Officer Bob today that Algy is in fact going to speed tomorrow. Should Bob prepunish Algy? Giving the reading of Smilansky’s argument under consideration, the fact that Algy will in fact speed makes it pointless to respect his autonomy by giving Algy the opportunity to change his mind. Therefore once again the libertarian seems committed to prepunishing Algy. In other words, the problem of prepunishment arises not from determinism itself, but simply from there being accessible facts of the matter about what people will

do in the future. It is a problem that both compatibilists and libertarians will have to grapple with, at least in principle.

At this point it might seem that a more plausible way to understand Smilansky is as saying we have to respect Algy’s capacity to change his mind about speeding, regardless of whether or not there are any facts about what Algy will actually do. This response still would not be open to the compatibilist (we can imagine the argument going), because if determinism is true, then Algy lacks the capacity to do otherwise than what he actually does. But if this is the right reading of Smilansky’s argument, then there is a familiar way for the compatibilist to respond. Many compatibilists have argued that given the proper understanding of “could have done otherwise” (usually in terms of some sort of conditional analysis), determinism does not imply that we are incapable of doing otherwise than we actually do. Therefore the compatibilist could agree with Smilansky that prepunishing Algy is wrong on the grounds that it violates respect for Algy’s ability or capacity to change his mind before he acts, even given the certainty that Algy will indeed break the law tomorrow. To start with the assumption that the compatibilist cannot make sense of respecting Algy’s ability to do otherwise than he will actually do is simply to beg the question against the conditional analysis of could-do-otherwise.

III. A Different Response to Prepunishment

If what I have argued so far is correct, then Smilansky has given us no grounds for thinking that compatibilism is on worse footing the libertarianism. Given that the way to cash out respect for Algy’s autonomy is in terms of giving him the chance to do otherwise, one of two results follows. Either possible future facts about what he will do undermine the objection to prepunishment (for both the libertarian and the compatibilist), or else they leave the objection untouched (for both the libertarian and the compatibilist). However, this still is an ultimately unsatisfying result. For one, there may be some independent reasons to reject the conditional analysis of possibility (which not all version of compatibilism rely on). If so, then the compatibilist’s ability to resist prepunishment is still questionable. And for another, there may be independent reasons for questioning the second reading that I offered of Smilansky’s argument. As Smilansky argues in a later reply to Stephen Kearns, there would seem to be no point in waiting to give someone a chance to do otherwise when we already know for certain that they won’t. If that’s right, then the second reading is out entirely, and we’re left only with the first which (as I’ve argued) doesn’t provide any support for resisting prepunishment at all.

With these considerations in mind, I would like to suggest a different response to the problem of prepunishment that doesn’t rely on an agent like Algy’s capacity to do otherwise. I think this can be done by pointing to those compatibilists who simply deny that the capacity to do otherwise is a condition of moral responsibility. This is the view of several notable contemporary philosophers, including Harry Frankfurt (1969), John Martin Fischer and Mark Ravizza (1998), Nomy Arpaly (2006), and many others. According to this group, our moral responsibility (and also generally our agency, autonomy, etc.) are completely independent of whether or not we actually possess the ability to do otherwise than we actually do. Such compatibilists would generally reject Smilansky’s argument that paying proper respect to Algy as a moral agent has anything to do with his ability to do otherwise than he actually does. Still they may agree with Smilansky that treating Algy as an autonomous agent and not a mere object requires not punishing him for crimes that he has not yet committed. In what follows I will very briefly sketch how one might argue for this conclusion.

One manner in which prepunishing might undermine the requirement to respect person’s autonomy is that such punishment would (in at least many sorts of cases) be unintelligible or unreasonable to the person being punished. We require that criminals have to have the capacity to understand the nature of and wrongness of their crime in order to be liable. At least in principle, the criminal should be able to understand why he is being punished. This is one reason why the law has special provisions for those who are insane or lack the mental capacity to understand the nature of their actions. Similarly, those who are being punished for actions that they haven’t committed can hardly be expected to find the punishment to be reasonable, even in principle. It’s unreasonable because from the standpoint of the future-criminal the crime hasn’t happened yet, and it’s still up to the future-criminal whether or not it will occur. And so on similar grounds, we might argue that prepunishment is unjustifiable.

Of course, it’s not immediately clear that this works for all cases of prepunishment. Recall earlier I distinguished between cases in which the criminal intends to commit and knows about his future crime, those in which he just knows about it, and those in which he lacks even knowledge of it. The requirement that the punishment in principle be sensible (in principle) seems pretty clearly to restrict prepunishment against those who know nothing of their future crimes, but it is less clear against those who do have such knowledge, and even less obvious with those who have already formed the intent to commit their future crimes. But even from the standpoint of agents who have formed the intention to commit a future action, whether the future action is going to occur is still up to him – it’s still something he has the capacity to refrain from doing (even if we know in advance that he in fact will not). Thus when we punish such a person for his actions before they have occurred, we do indeed treat him as an object, not as a rational moral agent in control of his actions. This does not mean that respect requires that we do nothing, that we wait around for what we know will happen. If we can prevent the crime, then by all means we should. But prevention is a separate matter from punishment. The claim is only that in punishing a future criminal before he has committed the crime do we treat the person as a mere object.

In addition, there is one further point that can be made to argue that prepunishment is unacceptable, also without assuming anything about an agent’s capacity to do otherwise. The point is that by punishing people before they have committed crimes, we seem to be giving them license to commit the crimes. It is commonly said that by enduring punishment, criminals pay a debt they owe to society. But if the debt is paid beforethe crime has occurred, then it seems that society now owes them something – the right to commit the crime. We can imagine a person like Algy happily paying the fine so that he is free to speed tomorrow. Aside from the absurdity of saying that Algy now has the right to break the law, this plainly undermines one of the fundamental purposes (and justifications) of punishment – that punishment ought to deter future criminals. If anything prepunishment instead seems more likely to encourage future crimes than discourage them, and on this basis alone it seems that a compatibilist (or anyone) can deem it unacceptable.

If my above arguments hold, then we can resist prepunishment even in the case where Algy knows he will and intends to speed in the future. And we can resist it not by appealing, as Smilansky suggests, to the fact that we have to wait and give Algy a chance to do otherwise (even if we know he won’t). Rather, as I have suggested, we can resist prepunishment on the grounds that it fails to do what punishment is supposed to do (discourage future crimes), and on the grounds that it violates the basic respect we owe to people, which in part includes a requirement not to dole out punishments which we could not in principle expect the recipient to find reasonable or intelligible. But even if there are some disanalogies that weaken the argument in cases where Algy intends to and knows that he will commit his future crime, the requirement that punishments be in principle intelligible at the very least severely restricts the class of acceptable prepunishment. It would still eliminate at least the most unintuitive cases – the ones where the person has no knowledge of or intention of committing his future crime. And thus Smilansky’s claim that compatibilists has to radically oppose our ordinary moral intuitions in the case of prepunishment is still undermined.


Arpaly, Nomy. Merit, Meaning, and Human Bondage. Princeton, New Jersey: Princeton UP, 2006.

Beebee, Helen. “Smilansky’s Alleged Refutation of Compatibilism.” Analysis 68 (2008): 258-60.

Fischer, John Martin and Ravizza, Mark. Responsibility and Control: An Essay on Moral Responsibility. Cambridge: Cambridge UP, 1998.

Frankfurt, Harry. “Alternate Possibilities and Moral Responsibility.” Journal of Philosophy 66 (1969): 829-39.

Kearns, Stephen. “Compatibilism Can Resist Prepunishment: A Reply to Smilansky.” Analysis 68 (2008): 250-53.

New, Christopher. “Punishing times: reply to Smilansky.” Analysis 55 (1995): 60–62.

New, Christopher. “Time and punishment.” Analysis 52 (1992): 35-40.

Smilansky, Saul. “Determinism and Prepunishment: The Radical Nature of Compatibilism.” Analysis 67(2007): 347-49.

Smilansky, Saul. “More Prepunishment for Compatibilists: A Reply to Beebee.” Analysis 68 (2008): 260-63.

Smilansky, Saul. “Prepunishment for Compatibilists: A Reply to Kearns.” Analysis 68 (2008): 254-57. Smilansky, Saul. “The Time to Punish.” Analysis 54 (1994): 50–53.







  1. Algy knows from introspective awareness of his own intentions, of course, and we may suppose that they both have access to extremely detailed and precise neuro-physiological and psychological information about Algy.
  2. Perhaps Algy is introspective enough to realize that he will succumb to the temptation to speed once he gets out on the road the next day.
  3. Saul Smilansky, “The Time to Punish,” Analysis 54 (1994): 50–53.
  4. In short, this is the view that time is a fourth dimension analogous to the three spatial dimensions. On this view all points in time are equally real. Temporal terms like “past” and “future” are understand as indexicals that are relative to where you happen to be located in time.
  5. Smilansky responds to a related time-travel based objection from Helen Beebee, but as far as I can discern he offers no reason to think that libertarian free will is incompatible with a block theory of time.

Ryan Lake