Compatibilism and Prepunishment

I would be very interested to see some Garden discussion of Saul Smilansky's very interesting recent article, "Determinism and prepunishment: The radical nature of compatibilism", which is in the most recent issue of Analysis.  It's a quick read, so you should really read the original piece, but here's the basics:

Suppose compatibilism is true.  Now consider a deterministic world where agents are free and responsible.  In principle, we could come to know that someone is going to commit a crime, even before the person commits the crime.  But there seems to be no relevant moral difference betweeen knowing that someone has committed a crime and knowing that someone will commit a crime.  So compatibilists cannot in principle object to prepunishment (punishing someone for a crime that he has not yet, but will, commit) on moral grounds.

I wonder in particular what responses compatibilists might give to this line of reasoning.  A first thought is that a compatibilist might draw the distinction between moral responsibility, on the one hand, and whether blame and punishment are justified, on the other.  Strictly speaking all the compatibilist is committed to is the compatibility of determinism and morally responsible agency, and perhaps something else is needed in order to justify blame and punishment, something else which would entail that prepunishment is unacceptable.  Other ideas? 

Comments

[Neil]

We have talked about this before...

http://gfp.typepad.com/the_garden_of_forking_pat/2004/12/time_and_the_co.html

[Neil]

That should be

http://gfp.typepad.com/the_garden_of_forking_pat/2004/12/time_and_the_co.html

[Neil]

Aargh! The link won't post. Someone gave me a tutorial on embedding links, obviously to no avail. I will try once more; after that you're on your own (the post dates from Dec 2004, and was called "Time and the compatibilist"; google will find it. Google is cleverer than me.

p>http://gfp.typepad.com/the_garden_of_forking_pat/2004/12/time_and_the_co.html

[Drake]

BTW, if you don't have access to Analysis, you can find the article on Prof. Smilansky's page here.

[Drake]

The page Neil wanted to link to was here.

[Tamler Sommers]

My first reaction is in line with Neal's in the first post. Bite the bullet, embrace the reductio. If complete predictability is given, the permissability prepunishment is one of a whole series of weird counterintuitive implications that might follow. I will say, although I'm not a compatibilist, that it's the complete predictability that's doing the work here more than the determinism. So maybe Smilansky is wrong to say that determinism and compatibilism are what makes prepunishment permissible. After all, some people think that divine foreknowledge and libertarian free will are compatible, right? If that's true, then it would still be permissible (for God) to punish someone before they choose of their libertarian free will to commit a crime. So again, it's not determinism that's making prepunishment permissible, it's the foreknowledge.

[Neal Tognazzini]

Neil,

Thanks for pointing me to that previous discussion. Upon looking at that thread, I had the odd sensation of confronting the opinion of one of my previous selves with which I'm not sure I agree anymore. Or, at least, I was rather surprised to find that I had already weighed in on this issue three years ago. I guess I'm not psychologically connected to that person.

[Steve Howland]

I'm seeing an inherent problem in that we are inside the equation in which we supposedly know what is going to happen.

If we can tell "beyond reasonable doubt" that someone is going to commit a crime, and then we interfere, then by the ideas of determinism wasn't it always "determined" that the person would not commit the crime? (assuming the person is prepunished and therefore not permitted to commit the crime) And if so, then on what grounds can the person be punished?

The idea is reminiscent of Minority Report in which we need to find out to what extent we are responsible for things that are "going" to happen... and if we know everything to that level of predictability is there not a way to influence the person who is about to commit the crime and somehow change his choice?

[Joe Campbell]

I think that Steve's point is an interesting one. For it and other reasons, I'm inclined to think that complete foreknowledge is not possible. In any event, I don't see that determinism entails complete foreknowledge. That is how I would get out of it.

[Mark Smeltzer]

I can think of at least one distinction that might make all the difference...

If compatibilism is true and agents deserve to be treated according to their moral characters, we know that at time T in the future agent S has (tenseless) character C and deserves (tenseless) treatment R.

The question is whether the those tenseless predicates have relevance now because the following two statements are not equal:

1. At time T in the future agent S will have (tensed) character C and will (tensed) deserve treatment R
2. Agent S has (present) character C and deserves (present) treatment R

It is not clear to see how we can conflate either the tenseless ethical statements or ethical statements about the future with ethical statements about the present. That would require some sort of argument -- one that does not readily jump to mind.

I can think of many apparent counterexamples to the operative transfer principle: the truth value of the statement "I will own a Ferrari in twenty years" does not entail the truth value of "If I will own a Ferrari in twenty years, I ought to purchase a Ferrari now".

Does Saul address any issues similar to this one in his paper? I will have to check it out when I get a chance.

[Neil]

I don't see the problem. There is absolutely no paradox involved in asserting the claim: If I do not X, then E will occur.

[Neil]

Clarification: my post was meant as a response to Joe.

[Mark Smeltzer]

p.s. I too think that the discussion about belief/knowledge/foreknowledge trumps the discussion about determinism as far as practical ethics are concerned (i.e. ethics behind the "veil" of what we actually know and what we believe we know). My comment above is meant to be directed at the determinism side of the discussion by assuming a realist perspective of the moral facts in question and an abstract, non-personal sense of ethics beyond the "veil" (although I'm not entirely sure that picture is coherent, which is probably the motivation of the worry I raised).

[Cihan Baran]

I kind of feel that this is a knock-down argument for showing that compatibilism doesn't mesh with our ordinary moral intuitions. So much for trying to show that compatibilism is intuitive!

But maybe I am too easily won over.

Joe, you say you would get out of it by arguing that determinism doesn't entail foreknowledge - or some such thing. But suppose that in a certain deterministic world, people could tell what you were going to do in your entire life based on the data they gather when you are 4-years old. Would it be permissible to punish 4 year olds? I mean, even if our world isn't like, if it were like that, would it be okay?

And thanks for linking to the earlier post. I wasn't even aware of the free will debate back in December 2004! I'll check it out - along with Smilansky's paper.

[Jonathan Farrell]

I haven't read Smilansky's piece, but here's my response to the situation Neal describes.

First, given how the situation is set up, pre-punishment _can't_ result in our criminal not committing the crime. Given the way things are when we discover they will commit the crime (and the laws of nature), it follows that they will commit the crime. Likewise, it is determined that we either pre-punish them or don't. (So we might be in the strange situation in which our pre-punishing them causes (in part) their committing the crime.)

So Steve's problem doesn't arise: although we may be free and responsible in the world described, since we know (because it is the case) that, whatever we do, the crime will take place, we're not free to make it that our criminal avoids committing this crime.

But the issue isn't whether or not we pre-punish him (or whether he commits the crime), but whether we have moral grounds to objecting to pre-punishment. I think this is going to depend on how we justify punishing people in general.

I have my doubts about whether compatibilists (or even soft libertarians) can justify retributive punishment, so I won't consider that approach. (Perhaps by 'punishment' Neal means only to talk about retribution. But then, as I've indicated, I'm not sure if compatibilists can justify even post-punishment. Anyway, if this is what 'punishment' means here, ignore the rest of this comment!)

But we might think that pre-punishment is as justified as post-punishment when it comes to deterrence/rehabilitation. Certainly, pre-punishing the criminal won't deter him from committing _this_ crime. But it might deter him from committing similar (or even not-so-similar) crimes in the future. And it seems that 'making an example' of our criminal before the fact will be just as effective (other things being equal) at deterring others from committing crimes as doing so after the fact.

So if punishment includes privations (or luxuries - whatever gets the intended consequences) inflicted for consequentialist reasons, then, although the notion of pre-punishment might sound a bit odd, I don't even think there is a bullet here for the compatibilist to bite.

[Randy Clarke]

Suppose that it is certain now that S will make a certain promise in the future. Does S now have an obligation to perform the not-yet-promised deed? Arguably not. Can a compatibilist hold that a right to punish is similarly unaffected by certainty about the impending crime?

[John Alexander]

Do we not sometimes prepunish people regardless of our beliefs regarding free-will and determinism. What I have in mind is the parental practice of forbidding a child to do x and placing that child in a situation where they cannot do x. Being placed in that situation by the parent is a punishment to the child because she would rather be doing x and would be, or would likely be doing x if she were not being punished.

[Randy Clarke]

Lots of unwanted treatment isn't punishment. Quarantine, for example. I'm not suggesting we can rule out pre-punishment by definition. But we can't count every treatment that an agent dislikes as punishment.

[Joe Campbell]

Let me respond to some comments in an attempt to clarify what I said above.

Cihan writes: "Joe, you say you would get out of it by arguing that determinism doesn't entail foreknowledge - or some such thing. But suppose that in a certain deterministic world, people could tell what you were going to do in your entire life based on the data they gather when you are 4-years old. Would it be permissible to punish 4 year olds? I mean, even if our world isn't like, if it were like that, would it be okay?"

I can wrap my head around the truth of determinism. I can't wrap my head around the truth of complete predictability of the sort that you and Saul are talking about -- being able to predict with certainty the actions of someone based on facts about him when he was 4-years old. I do not think that predictability of this sort is possible. More to the point, I don’t see why it follows given mere determinism. That it does follow strikes me as a myth and a misunderstanding of determinism.

Determinism entails that such facts exist. It does not entail that we can ever know those facts -- nor should it because such knowledge would lead to a slew of paradoxes, one of which is noted by Steve above -- the Minority Report Paradox.

Suppose we know that S will do X, where X is the morally blameworthy murder of another individual. We decide to prepunish S. Thus, S does not do X. For what exactly is S blameworthy? Or are we supposed to imagine that S does X even though he has been prepunished -- that what we know is that S will do X whether or not he is prepunished; that in the future we punish people and then release them, so that the punishment is justified because there is a morally blameworthy act. In this case it seems that people’s actions are compelled in a way that goes beyond mere determinism. I refuse to acknowledge that everyone who murders would murder regardless of whether they are prepunished or not, merely given the truth of determinism. That sounds absurd!

Again, determinism is not fatalism; it is not the thesis that future events will happen no matter what. To the extent that we have 'certain' foreknowledge of future events that allow us to alter those future events, we no longer have determinism! The 'foreseen' event that we prevent could not have been determined in the first place since it didn't in fact happen!

In order for Saul's objection to hold, he has to give me a reason to believe that determinism entails complete predictability (I said complete foreknowledge above but I should have said complete predictability). Since I believe the latter is impossible -- it leads to paradoxes -- yet I think that determinism is possible, I don't think that the entailment can be shown.

Neil writes: "I don't see the problem. There is absolutely no paradox involved in asserting the claim: If I do not X, then E will occur."

The paradox comes in, if I understand your point, when we put this claim into the context of the kinds of thought experiments that we are considering. That is, when we suppose that E is foreseen to be the result of S's doing X. I don't question that an event -- a murder, say -- might take place even if S does not perform the murder. But then for what, exactly, is S responsible? Suppose you have reason to think that I might steal something but confront me before I perform the act and I decide not to steal. Am I guilty of anything? I don't see how.

In sum, what I need to see are examples in which someone is both prepunished AND performs the act for which he is punished. Then I need to see a reason why I should believe that such scenarios are a consequence of mere determinism – if mere determinism were true, such scenarios would be prevalent. I don’t see either.

Lastly, it is likely that compatibilism is already unintuitive. The fact that it has an unintuitive consequences should not be surprising. As I indicated above, I don't think that it has these particular consequences. But even if it did, why should that surprise us?

[Mike Iarrobino]

I don't think Smilansky needs to show that determinism *entails* complete predictability. All he needs to show is that predictability and determinism are consistent (of course, this depends on whether complete predictability is coherent in the first place). If they are consistent then there are possible worlds where determinism and complete predictability hold; these are the worlds in which compatibilists might have to endorse pre-punishment.

[Joe Campbell]

Might have to endorse prepunishment? Why?

And this doesn't show that there is a problem for compatibilism any more than some fanciful story about the possibility of indeterminism and pre-punishment shows that there is a problem for libertarianism.

To get the pre-punishment possibility you don't NEED determinism. You could get it with eternalism and a foreknowing god who communicates with precogs. (If such a thing is possible; frankly it seems as absurd as thought experiments noted above.) There is nothing in the Minority Report, for instance, that necessitates the truth of determinism. Hence, if there are any problems they have to general problems for any theory of free will. Or so it seems to me.

[Tamler Sommers]

If that's true, then I think Smilansky also needs to show that libertarian free will is incompatible with complete predictability (even by God). Otherwise, there's nothing especially unintuitive about the (logically possible) implications of compatibilism. Incompatiblism would have the same logically possible counterintuitive result. And while I'm no believer in an omniscient God, it's equally (if not more) incredible to think that there can be omniscient humans or computers, even if determinism is true.

[Tamler Sommers]

Whoops, I was responding to Mike's post--Joe's wasn't up yet--making roughly the same point as Joe.

[Saul Smilansky]

Thanks, Neal (T.) for pointing to my paper. I don't remember reading Neil (Levy's) post of three years ago, but I might well have done so. In any case clearly his time travel into the past scenario brought up (if in a different way) issues I consider in my paper. If I re-publish my piece in the future I will refer to Neil's Garden post. [I am glad that Neal T. also forgot the post or at least he forgot that he made comments in response; this makes me feel better.]

My point is simple. It might be easiest to explain it "historically". In 1992 Christopher New proposed (In Analysis) that we might preupunish people, when we can know beyond a reasonable doubt that the person is going to commit the crime, since "beyond a reasonable doubt" is our common standard anyway. He tells the story of a reliable guy who is about to commit an offence and offers to be punished in advance, when we know that we won't be able to punish him after he commits the crime. My reply defended the commonsense view (and yes, it was odd to find myself in that role), and said that for reasons of respect for persons we mustn't pre-punish this person, because we must allow the still innocent to change their minds, maintain their moral goodness and not be punished. Hence pre-punishment differs from post-punishment. In my recent paper I reverse the direction, and say that since this commonsense consideration is not available to the (determinist) compatibilist, compatibilism is fundamentally at odds with our moral commonsense.

A few comments in clarification and response:

1. This discussion is interesting only if we accept some measure of deontology. Pre-punishment might well be effective in certain circumstances, but the question is whether it is just (or something like that). The compatibilist need not believe that there is something good about the punishment of people in itself irrespective of the good the institution of punishment does, but she should e.g. think that it is wrong to punish the (compatibilistically) innocent, even if doing so would enhance social utility. I think that most compatibilists would think so.

2. Joe - I don't think that doubts about predictability will get you far here. The question is whether compatibilism has any philosophical resources to resist in principle pre-punishment. Taking New's example, I don't think that it does. But in any case, compatibilism can be confronted by the question even if there is never enough predictability. By analogy, if I claim that your moral theory does not give me any reason not to kill people whenever they are rich and I won't be caught, it would not be a strong reply for you to say that one can never be certain that one will not be caught.

3. Character - I don't think that this will work. If I commit a crime which is out of character, then you should still want to punish me for it, just as typically ungenerous people who act with generosity ought to be thanked, even if their act was out of character. Being out of character might even make my bad actions morally worse, or my good ones morally better. In a more general way, we punish for acts, and character comes in primarily as a side (e.g. mitigating) factor. The question, again, is whether compatibilism has the resources to resist pre-punishment, as our moral commonsense requires.

4. Randy - if the reason why we think that we are not permitted to pre-punish is that we must allow the person to change her mind, and then we learn that it is determined that she will not change her mind, then it seems to me that determinism took away the moral reason for not pre-punishing.

5. Nothing in what I have said shows that compatibilism is mistaken. Perhaps the compatibilist should bite the bullet, like radical utilitarians such as Smart, and say that the resistance to pre-punishment is just an ancient prejudice lacking principled grounding. But what the compatibilist cannot do is to take that line, and at the same time claim that determinism does not make a difference, morally, that he can affirm our commonsense beliefs, and that hence there is nothing to worry about.

6. Honest disclosure requires me to say that I like this issue in part because it helps my case for free will Illusionism.

[Tamler Sommers]

Saul, if I'm wrong to keep pressing this point, please tell me. But this issue can only help your case for free will illusionism if it shows that ONLY compatibilism has this counterintuitive implication. If the belief in libertarian free will has the same implication, then the illusion (belief in LFW) would be of no help at all.

What if God could tell us that someone was going to choose of their own libertarian free will to murder 10 children. And we were absolutely certain that God is never wrong about these matters. We can't understand HOW he knows that an undetermined event is going to happen, but there's a lot we don't understand about omniscience. Wouldn't prepunishment be permissible in that case? Maybe you think this case is too unbelievable to take seriously, but then so are cases involving complete predictibility given the truth of determinism.

It seems like you're making compatibilism suffer for an implication of a literally incredible (but logically possible) scenario, while the belief in LFW, which has the same logically possible implication gets off scot free.

[Saul Smilansky]

Tamler,

Perfectly legitimate for you to worry about an omniscient God :)

I think that the difference is this. Libertarian FW IN ITSELF has no problem defending the constraint on pre-punishment: till the person has acted, at which time it is too late of course, it is still open to him not to commit the crime. It is fully up to him, and not determined in any way, and hence punishment before the crime is a problem. By contrast, compatibilism IN ITSELF cannot say this, because it is determined that the person will commit the crime.

BTW, I don't think that it IS so difficult to think about cases where predictability will be sufficient so that the compatibilist will confront a real problem. The probability that people who have commited 47 burglaries will not commit the 48th must be very slight. The libertarian (qua libertarian, not as a believer in this or that theology) simply does not have to worry about any of it, for he does not care about probabilities and practical predictability; his stand is principled.

BTW2, If I believed in LFW, I would say that God cannot know in advance with absolute certainty LFW-free actions, as he can know determined actions. For the LFW-free actions are not there to be known in advance, they are composed right at the time of freely coming into being. And of course some theologians and philosophers have said that. It is no disrespect to God, who after all created the people with such illustrious freedom and thus (presumably freely) chose to limit himself in that way. And we can understand why, for as I have argued in detail a world with LFW (assuming that this is possible) would be much better, morally and personally, than a world without any LFW-free creatures. If one's theology forces one to say that there can be both LFW-free acts and complete predictability then that theology has a problem, but it is not our problem.

[Randy Clarke]

We might think we aren't permitted to pre-punish because the person hasn't yet committed the crime, just as we might think we can't pre-require a to-be-promised deed because the person hasn't yet made the promise. One might ask why it matters that the crime hasn't yet been committed (given that we know it will be); but equally one might ask why it matters that the promise hasn't yet been made (if we know that it will be).

[Mark Smeltzer]

Saul,

Your point about character misses my meaning. When I talk about character I consistently intend to be talking about that which provides content to moral statements about agents. I often equate character to an agent's moral properties.

So, when I talk about character I am not talking about perceived habits or tendencies that a person may have. In the sense of character I invoke, it is never appropriate to punish/reward an agent for behavior that is "out of character" because that is the same thing as saying that the agent in question "didn't do it".

[Tamler Sommers]

Saul,

Case One: Determinism is true. A burglar has committed 47 burglaries. We know that there's a 98% chance he will choose (be determined to choose) to rob again. But we can't be 100% certain, because of epistemic limitations.

Case Two: People have libertarian free will. A burglar has committed 47 burglaries. Our best inductive theories tell us that there is a 98% chance that this burglar will choose of his own free will to rob again. But we can't be 100% certain because the burglar has libertarian free will.

Why is there more of a principled reason to resist prepunishment in case two? Either it is certain that the person will commit the crime or it isn't. If it's not completely certain that the crime will occur then there doesn't seem to be more or less of a reason to prepunish in either case. And if it certain, then again, I dcn't see how principle favors the libertarian.

And yes, I'm extremely worried about the existence of an omniscient God. If there is, my only hope is that He's a merciful God as well.

[Randy Clarke]

According to the Space-Time Theory, the future is just as real as the present. How is that inconsistent with future events' being undetermined (where determined means just what it does on the standard definition offered in discussions of free will)?

[Cihan Baran]

Joe,

That's pretty convincing. Let me think about it some more though.

Cihan

[Neal]

Randy,

About the promise analogy. When someone makes a promise, the content of the promise is about the future with respect to the time that the promise is made. (You say, 'I promise that I will X', not 'I promise that I have Xed'.) I would think that the reason we can't force people to perform to-be-promised actions ahead of time is that anything they do ahead of time won't count as a fulfillment of the promise. Of course, perhaps we could warn them ahead of time that, since they will promise to do X, they better perform when the time comes. But that's something different.

[Randy Clarke]

Neal,

Good point. Though it seems to leave it open that, in virtue of a promise one will make tomorrow to A the next day, one now has an obligation to A day after tomorrow.

[Alan]

I have scanned this thread--not word for word--and if someone has said something about complexity and chaos, I've missed it. But deterministic systems of any complexity exhibit chaotic characteristics, and they are not inherently predictable except within the scope of general patterns--and I doubt if that will include or entail specific predictions of behavior that are susceptible to condemnation. This is an empirical/mathematical point--but it relates at least to an actual deterministic world like ours rather than ideally predictable world like Laplace's. I want to echo Joe's point: determinism is not fatalism, not empirically, not conceptually (though I have independent arguments for this latter point).

[Kip Werking]

Firstly, I agree with Cihan and Randy: prepunishment is pretty embarrassing to compatibilism but not all unwanted treatment is punishment.

It might be best to give unwanted treatment to people we know will commit a crime in the future. But, at that point, let's stop saying "you freely chose, or will freely choose, to commit this crime and so you deserve punishment."

Instead, let's say "you are sick, or something like sick, and we need to treat you to protect society. Perhaps we will quarantine you. Perhaps we will give you an unwanted vaccine or brain surgery. But we have to protect ourselves from you, just as we have to protect ourselves from Ebola. It's not your fault, except in a sense of fault that is *shallow* [to use Smilansky's beautiful terminology]."

In other words, the question of prepunishment just reinforces my tentative belief in the correctness, and inevitability, of a medicalized society that regards the ideas of free will and punishment as obsolete, if not ridiculous.

Secondly, I don't understand Joe's resistance to the idea of predictability. It might be true that determinism doesn't entail predictability. Entailment is pretty strong. But determinism might leave open the possibility of predictability---and this is what has worried incompatibilists over the ages. Take the powerful and intelligent being of your choice: God, aliens, robots... whatever it is, determinism spells out how they could predict what would happen before it actually does:

1. just gather all of the information on the closed deterministic system in question
2. gather all of the rules governing how this closed system proceeds from one state of the system to the next
3. program these into a computer or calculator that operates faster than the closed system itself

[Note that this can happen by just simulating, or recreating, the system in question. But it can also happen by finding shortcuts and heuristics that don't require calculating every detail in order to reach the conclusion.]

This is what worries incompatibilists. It shows how one's entire life story, one's entire destiny, might be in a real sense "written" before one is even born. If Joe can't/won't entertain the possibility of this happening, then no wonder he's a compatibilist!

Indeed, it's important to remember that, for all any of us know, there is such a God/alien/robot, and it does know everything about the rules and initial conditions of the universe, and our universe is deterministic, and the God/alien/robot has already calculated ahead to spell out our entire life story. There is no way to know this isn't happening. To such a God/alien/robot, watching our universe unfold is like watching a movie it has already seen, and we're just characters playing our parts, written eons before we were even born.

[Neil]

Joe, I don't see that the force of the argument depends upon it being the case that the agent commits the crime anyway. If the agent would have committed the crime, then aren't they as deserving of punishment as if they have? Why wait? Compare attempts: what makes an agent deserving of punishment for attempts is that they tried to commit the crime. Well, why not interrupt the trying before it goes to completion?

In any case, I don't think that cases which have the features you want are hard to construct. All we need is to imagine a scenario in which you possess the relevant knowledge, can punishment the agent before the crime, but cannot prevent it (perhaps they are too powerful).

[Cihan Baran]

Kip,

You write:

just gather all of the information on the closed deterministic system in question

I don't know what you mean by "closed" but if it's anything like I am interpreting, then the system in question is not "closed".

If you predict someone's behavior and interact with her on the basis of this prediction, the system is no longer *closed*. You have expanded the system to include the person in question as you modeled in your prediction AND whatever effects/changes your interaction based on predictions will have on the system. So your predictions no longer are valid and you must recalculate everything.

In such a situation, in order to make correct predictions, your interaction with the system must be self-reinforcing. For instance, one day while I am trying to defeat my addiction to online chess, you say to me "Your efforts will never prevail." Losing my confidence all the more so due to your remark, I spend yet another night playing chess on the internet.

So, if we are thinking of self-reinforcing pre-punishment, then the agents in question can hardly be thought to be morally responsible or engaging in morally impermissible acts. Take a case like following. Somebody pre-punishes Jones for something Jones'll do in the future and then Jones goes ahead and does that thing because somebody pre-punished Jones and Jones figures "I was already punished for this act - so why not go ahead and do it?" I am not exactly sure but this kind of case is hardly troubling for the compatibilist.

Maybe in rare cases, prediction will really have no effect on the system. That is to say, I will succumb to my addiction for online chess no matter what anyone says, dismissing anyone's remarks about my situation and yet not having quite the will-power. It's hard to tell but even in such a case, one's predictions may not be valid, since one's changing the system.

However, it's hard to argue that pre-punishment won't have any effect on the individual. After all, then why are you doing it if it has no effect?

One might argue that "well I can take into account both the effect my predictions will have on the system and the system to arrive at true predictions." Well then there are certain cases which cannot be predicted in principle. Suppose that you are trying to predict whether I am going to eat ice cream or Oreo cookies and I am committed to pick whichever one you don't predict me as eating. Well, I don't see how you could predict that one. (Credit to Hilary Bok for this point.)

One might think about cases where pre-punishment deters the individual in question from the harmful act. This seem justified to me but these can hardly be called "punishment" - if you quarantine someone so that he doesn't spread the illness, that hardly counts as punishment.

However, as I argued earlier, and I thought this was the gist of Joe's point, there may be cases where this is impossible - (i.e. when I am committed to eat whatever you don't predict me as eating, i.e. when I am committed to do whatever you pre-punish me for...)

[Cihan Baran]

If the agent would have committed the crime, then aren't they as deserving of punishment as if they have? Why wait?

I don't know about punishment but can anyone tell me if the agent's are morally responsible according to Fischer's criterion in such a case? Suppose you predict how I'll behave based on the laws of nature and pre-punish me to prevent that from happening.

If I am remembering Fischer's elaborate conditions correctly, one has to look at the mechanism that produces the agent's decision and the world must be appropriately sensitive to agent's bodily actions -here is the gimmick- HOLDING FIXED the triggers. Since your pre-punishment is a trigger, and positing the appropriate reasons and bodily sensitivity, you are morally responsible even if you don't do the act.

Or am I missing something?

[Joe Campbell]

What a wonderful thread!

I think Alan is right - chaos theory suggests that complete predictability is impossible, or explains why it might be.

A few notes to Saul. I'm glad whenever Saul writes because I worry. I won't say why but I do worry. Also, some of the sweetest folks I've ever met are free will skeptics: Saul, Derk, Ted Honderich, Tamler, Kip -- maybe even Peter van Inwagen! Anyway, it is good to hear from you, Saul!

Back to the arguments.

You question that compatibilism has the "philosophical resources to resist in principle pre-punishment." But I think everyone has those resources. Pre-punishment requires that "we can know beyond a reasonable doubt that the person is going to commit the crime," and for various reasons I think that such knowledge is impossible. Thus, I think that commonsense responses -- like "we mustn't pre-punish this person" are as available to the compatibilist as they are to anyone else.

I don't mind biting the bullet -- admitting that pre-punishment is justified -- but I don't feel a need to do so. And part of this has to do with the fact that I don't think that determinism "make(s) a difference."

Furthermore, I don't see why libertarianism limits predictability. For one thing, libertarianism is consistent with one's developing a fully formed character, such that one's actions inevitably flow from it. Given the character, predictability of future actions should be near certain – as certain as any fully determined actions. Yet that the act traces back to some non-determined event might be relevant. Here you have an act that is free, in a libertarian sense, yet is fully predictable. Predictability is a non-issue, as I see it.

Thus, your problem -- which I admit might be a problem -- is not a problem that is particular to compatibilism. I acknowledge that it might, in the end, support illusionism. But only if you can give me a reason to think that complete -- or sufficient -- foreknowledge (the kind needed to support and justify pre-punishment) is possible.

Lastly, thanks to Randy for his comments -- which are spot on -- and to Neil, with whom I disagree. I think, "If the agent would have committed the crime" but did not commit the crime, then they aren't deserving of punishment. For the same reason I think that if you had decided to do something sinful, yet a friend happened to stop you with wise words, you would not be sinful. Who among us can say that we haven’t been prevented from doing something bad because of advice given by a friend, or even a stranger? Are we sinful? No, for we didn’t do the deed. Perhaps the wanting to do so is sinful, too, but not to the same degree, or not for the same reasons.

There might be a bit of luck here, too, but the fastest among us do not always win the race.

[Neil]

There's a big difference, Joe, from being disposed to commit a crime under all conditions except one in which one is is prepunished, and being disposed to commit a crime under conditions except those in which one is advised not to.

The link between the 'sin' and the desert might seem exiguous in the prepunishment w/out crime case, but it seems pretty exiguous in many post-punishment cases - those in which the crime was in the distant past/out of character, for instance.

[Saul Smilansky]

Thanks, Joe.

Tamler - I just don't see why the libertarian needs to worry about pre-punishment nearly as much as the compatibilist (if at all). Not sure I have anything very new to add here, but I'll try to make my point from a slightly different direction. The issue is not (as I see it) about the statistics of probablility, but concerns a principled moral requirement: the libertarian can make a moral demand, grounded in respect for persons, that until the last minute we allow the agent to change his or her mind. This demand makes no sense when coming from the mouth of a compatibilist, since there is no scope for such change. That's why determinism matters a lot here. Let's assume that we cannot predict what the person is going to do: the compatibilist should be sorry about this (if pre-punishment would be advantageous), because this prevents pre-punishment, which so far we haven't seen any reason why he shouldn't favor, while the libertarian would rule out pre-punishment in principle.

[Saul Smilansky]

Randy,

The promise analogy is intriguing. But I wonder how much work is being done here by our assumption that promises are things one can make but also not make (e.g. marry Zelda), and so are a bit like a gift. Hence asking to get the promise delivered earlier invites a negative response: you should be grateful to get it when you get it (according to the promise); don't push your luck. By contrast, you get punished for doing something wrong, and hence it is not clear why you may not get punished earlier (if, say, that's the only way to punish you). At least there isn't the idea that getting punished is supererogatory, which is typical for promises (of course once a promise is made it is prima facie obligatory, but typically making the promise in the first place is not).

This seems like a good place to introduce the notion of desert. You don't deserve to get today what I promised to give you in a month. You will deserve it in a month (or deserve today to get it in a month). But given determinism, the act for which you deserve to be punished is already "there", and if there is good reason to pre-punish you for it then perhaps you cannot deny that you deserve it (or pre-deserve it). Of course this sounds odd, we are not used to the idea of pre-punishment, but the question is whether the compatibilist can resist pre-punishment if it is otherwise tempting. My point here is that because deserving punishment means that you are bad, you cannot complain about pre-punishment (after all, you will be doing wrong!); you can expect to get punished (pre-punished) for doing wrong, in a way that you cannot expect to be pre-rewarded with what was promised.

[Kip Werking]

Cihan,

There are cases where it is logically impossible for the predictor to intervene. However, even in these cases, the predictor can still predict, and it seems to be the prediction, and not just the prepunishment, that worries Joe.

However, in a closed system case, then one can make accurate conditional-predictions, and also intervene. What do I mean by this? Well, consider a virtual world running on a laptop in the year 2100. This virtual world might be sufficiently detailed to contain, informationally, agents about as sophisticated as we are.

This system is closed in the sense that you don't have to know the state of *our* entire universe, and laws of physics, in order to know what the future states of the world in this virtual world will be. You could just get the information about that world (not unlike the information about a game like Halo today), program it or its essentials into a faster computer, and see where the program goes.

Then you could obtain conditional information: if I don't intervene, then character X will murder character Y in 12 days. Of course, you can intervene. You can, for example, pull the plug on the laptop. But you can still make meaningful, conditional predictions.

I hope that helps explain my point.

[Randy Clarke]

Consider the idea that being punished is paying a debt to society. Then pre-punishment is pre-payment--paying the debt before it is incurred. But now, the predictability of my borrowing money from you tomorrow wouldn't seem to support the view that I owe you the money today. (Not even if the act of borrowing the money is 'already there' in the future.) Perhaps pre-punishment is resistable on grounds of this sort. (Though I don't know what to think about the idea of punishment as paying a debt.)

[Clayton Littlejohn]

My response is somewhat similar to Clarke's insofar as I think the compatibilist can tell a plausible story about how there are either duties not to pre-punish or how pre-punishment cannot fulfill the aims of punishment even in the deterministic world.

First, we ought to assume that we are not working with a consequentialist approach to punishment. It is hard to see how there is any deep problem with pre-punishment on such a view. So, let us work with a non-consequentialist approach. On some such views, there is either (a) a prima facie duty to make it intelligible to the person being punished what the grounds of the punishment are or (b) one of the functions of punishment is to communicate condemnation to the person being punished. Suppose that the authorities have foreknowledge of the crime but that the wouldbe perpetrator does not. It seems that inflicting punishment under these conditions would either violate the duty described in (a) (just imagine the horrors of being subject to what you would not unreasonably regard as unjustified punishment) or would fail to serve the purpose outlined in (b). Now, if we insist that the future crime is known to both the authorities _and_ to the perpetrator, my intuitions are that you do not violate the duty described in (a), you do not thwart the intentions you have if you wish to communicate condemnation, and there is nothing wrongful with pre-punishment.

[Justin Capes]

Randy,

Suppose we grant that punishment is a sort of debt payment. Why does the temporal order of debt to payment matter in principle? If it's true that S will certainly commit some crime in the future, the scales of justice will, so to speak, have to be balanced out in the end. What’s wrong with S paying the debt now? Indeed we might think it better to do it sooner rather than latter—just get it over with. In any event, the end result is the same: in the end S will have squared things away with us. So why does the order in which we do the squaring matter?

[Tamler Sommers]

Joe,

Did you just call Peter van Inwagen a free will skeptic? When did that happen? Did I miss the memo?

[Mark Smeltzer]

Kip,

I'm afraid that your explanation only serves to show that you must have refrained from following up on those resources I sent you in Niel's previous thread about chaos theory. Suffice it to say that there is a gross gap between what you assume is possible and what is actually possible regarding knowing future states of sufficiently complex deterministic systems (and yes our world and all relevant nearby worlds would meet the criteria for being sufficiently complex).

[Mark Smeltzer]

Tamler,

I was more surprised that he suggested that PVI is a sweet person. I can't imagine any possible worlds where that is a true statement!

[Joe Campbell]

Van Inwagen is not really a free will skeptic -- but he is a sweet guy!

He belongs in the same broad camp, though, for he holds what Kane calls a 'successor view.'

Here is an interesting quote from "Free Will Remains a Mystery."

"The problem of free will in its broadest outlines is this. Free will seems to be incompatible with both determinism and indeterminism. Free will seems, therefore, to be impossible. But free will also seems to exist." (2000, 11)

Certainly he has given some of the best, most interesting arguments for 'incompatibilism' between indeterminism and free will, which coupled with the Consequence Argument, yeilds the conclusion that free will does not exist. He thinks it does exist but he agrees with the free will skeptic that he has no response to any of the arguments noted above. It is a mystery! He has also argued that even if we have free will genuinely free actions are rare! Recall, too, that at the Action, etc. INPC he has some very nice things to say about Saul's paper.

He is not a free will skeptic -- but he is pretty dang close!