Cross-world luck at the time of decision is a problem for compatibilists as well

Abstract

Alfred Mele has put forward what he regards as “a serious problem luck poses for libertarians” ([Mele, Alfred. 2006. Free Will and Luck. New York: Oxford University Press.], 6): that the kind of indeterminism libertarians require for free will brings about what he calls present luck, luck about the fact that one decides and subsequently acts as one does rather than in some other way that one was, at the time, causally able to decide and act. In this paper, I argue that present luck is a problem for compatibilists as well. I grant that decisions which are undetermined in the way required by libertarians are lucky in the way Mele describes. However, I argue, this is not because such decisions are undetermined, but because they are made by an agent who is motivationally split and has certain specific dispositions to deliberate and decide in different ways in the circumstance. If a split decision of the relevant kind is performed in a deterministic world, the agent is just as lucky that she decides as she does rather than otherwise, because this contrastive fact cannot be explained by any suitable combination of mental causes of the decision. Hence, the luck problem is a problem for libertarians and compatibilists alike.

Keywords:

• free will
• libertarianism
• luck
• chanciness
• present luck
• cross-world luck

1. Mele's luck problem for standard libertarians

I will discuss Alfred Mele's version of the luck problem for standard libertarians,¹ which he offered first in a series of papers and then in his 2006 book Free Will and Luck (see especially Mele 2006, Chap. 3, and also Mele 1998, 1999a, 1999b, 2005).² Mele originally posed the problem as a challenge ³ to Kane's event-causal libertarian view, though the luck problem threatens any standard libertarian account. A lot has been written on Kane's resources to face the luck problem. This paper concerns the luck problem as it applies to any standard libertarian view.

On the account Kane offers in his 1996 book The Significance of Free Will, we perform actions which are free only if undetermined when we are motivationally split between competing courses of action and trying to resist some temptation: when we are torn between doing what we think is morally right and doing what we think will help us attain our ambitions, or between powerful present desires and long-term goals, and so on (Kane 1996, Chap. 8). In response to Mele's luck challenge, Kane (1999a, 1999b, 2007a) postulates that, in cases of conflict, agents try simultaneously to do various incompatible things (see, for instance, Kane 2007a, 26–33). This way, what is undetermined is not whether the agent's effort to resist temptation will succeed or fail, but which effort will succeed. Whatever the agent chooses, she will have succeeded, because she was trying to choose that way. Another stipulation intended to alleviate the luck problem is Kane's claim that decisions in cases of moral and prudential conflict follow metaphysically indeterminate efforts to resist temptation; since these efforts, whose outcome is undetermined, are themselves indeterminate, two efforts with different outcomes (i.e. which result in different choices) cannot be said to be exactly similar (see Kane 1996, 128–130 and 172).⁴

Crucially, Kane concedes that indeterminism diminishes our control over our behavior, because it precludes antecedent determining control, or ADC, which he defines as

 … the ability to be in, or bring about, conditions such that one can guarantee or determine which of a set of outcomes is going to occur before it occurs, whether the outcomes are one's own actions, the actions of others, or events in the world generally. (1996, 144)

But, Kane holds, this is the price to pay for the freedom to rationally and voluntarily choose and do otherwise, which is necessary to have free will (see, for instance, Kane 1996, 143–145, 2007a, 39). In contrast, I deny that indeterminism would diminish control, and claim that, even though action determination (of a suitable kind) is necessary for ADC, this kind of control is impossible in principle in split actions (which are the actions that must be undetermined to be free according to standard libertarians). Hence, I claim, the indeterminism required by standard libertarians would not undermine our control over split actions by precluding our ADC over them. I regret I don't have the space to properly discuss Kane's insightful and influential defense of libertarianism. In a nutshell, my view is that Kane need not concede as much as he does to the proponent of the luck problem, and that his account doesn't need to be sophisticated to face the luck problem. Pace Kane, I grant that the chanciness involved in split decisions won't be eliminated with metaphysical indeterminacy and dual efforts. But I believe that this chanciness won't be eliminated with determinism either.

A standard libertarian need not postulate, as Kane does, efforts to resist temptation in the antecedents of every action that is free only if undetermined, nor dual efforts, nor metaphysical indeterminacy in the antecedents of any of our decisions, so I will not discuss these commitments nor their relation to the luck problem. I am unconvinced by Kane's response to the luck problem (see Kane 1996, 171–174, 1999a, 1999b, 2007a, 2007b) and by the responses that rely on agent-causation (see O'Connor 2000; Clarke 2005). I have nothing of substance to add to these debates, so I will move on for the sake of brevity.

Mele raises a challenge to any libertarian account which claims that, if an agent is motivationally divided between various incompatible options, her decision is free only if she is causally able, up to (roughly) the moment of decision,⁵ to refrain from deciding as she does. This is one of the cases Mele offers (2006, 73–74):⁶ Bob agreed to toss a coin at noon to begin a football game. But Carl bet that today's starting coin will not be tossed until 12:02, and he offered Bob $50 to wait until 12:02 to toss it. Bob doesn't know what to do: he wants the money, but has moral reasons against helping Carl cheat other gamblers. After reflecting on the issue, Bob judges that, all things considered, he should toss the coin at noon. However, at noon, he decides to wait until 12:02 to toss the coin. This decision occurs at a time at which the past and the laws of nature are consistent with Bob's not deciding to wait until 12:02; in particular, in a possible world which is exactly the same as the actual world up to noon, Bob decides at noon to toss the coin straightaway, and does this. Thus, at noon, two opposing decisions are nomologically possible: the decision to help Carl cheat, and the decision to do the right thing. What Mele calls “the problem of present luck” (2006, 66) is that Bob's deciding to do the wrong thing instead of deciding to do the right thing seems to be just a matter of chance, and hence – given the decision's significance for Bob – just a matter of luck for Bob, because the causal antecedents of both decisions are exactly the same. Bob is just unlucky to decide to cheat instead of deciding to do the right thing, because after all he did (indeed, after all that happened) up to noon, each alternative decision was compatible with the past and the laws, and it was mere chance that one was made at that time instead of the other.

Mele does not think that it is just a matter of chance that Bob is significantly motivated and causally able both to toss the coin at noon and (alternatively) to wait until 12:02 to toss it. Nor does he think it is just a matter of luck that Bob decides to toss the coin at 12:02 (he makes this decision because he wants the 50 dollars). What is just a matter of luck, according to Mele, is “the difference at the time of choice or decision between the actual world and any world in which he instead decides in favor of another live option” (2006, 74).

In other cases Mele offers, the agent makes an effort to resist temptation, and whether the effort will succeed or fail is undetermined in such a way that both outcomes are nomologically possible after the effort has ended. One case of this kind is the following:

John believes that he ought to arrive on time for a meeting that begins at noon in his building, but he is tempted to arrive late, as a modest protest. Although he tries very hard and very intelligently to resist his temptation, and although he has a reasonable chance of succeeding, his ( … ) effort fails: a minute before noon (at t) he decides to go to the meeting late. At a nearby world with the same laws and a very similar past relative to t ( … ) John* (that is, John or his counterpart) believes that he ought to arrive on time for the meeting, is tempted to arrive late (for the same reason), and tries very hard and very intelligently to resist his temptation. John*'s ( … ) effort succeeds. (Mele 1998, 582–583)

Again, Mele's claim is not that John's decision is completely chancy, or merely a matter of luck for John. What is just a matter of chance is that he decides to arrive late instead of deciding to arrive on time. This difference is just a matter of luck for John, because nothing in the causal antecedents of his decision to arrive late makes it the case that he decides this instead of deciding to arrive on time.⁷ And, if John (or Bob in the example above) decides to do something prima facie blameworthy instead of something prima facie praiseworthy out of mere bad luck, it seems inappropriate, Mele says, to blame him for deciding as he does. Given that the best explanation for his lack of responsibility (if he is indeed not responsible) is his lack of freedom, the fact that it seems inappropriate to blame John for his decision indicates that his decision is not free, and hence that his ensuing action is not free either (see 2006, 8 and 60). Mele's general worry regarding all the cases he discusses is that “[i]f there is nothing about the agents' powers, capacities, states of mind, moral character, and the like that explains [the] difference in outcome, then the difference is just a matter of luck” (1998, 583, see also 2006, 9).

2. Cross-world luck challenges compatibilism

I concede that decisions which are undetermined in the way standard libertarians require are subject to present luck. However, I claim, cross-world luck at the time of decision (i.e. luck about the fact that one decides as one does rather than otherwise)⁸ is a problem for compatibilists as well. What I argue is that it's not because they are undetermined, or in virtue of the fact that they are undetermined, that these decisions are lucky; they would be just as lucky if they were determined.⁹ My claim is that a decision performed at a time at which the agent is psychologically able (and suitably skilled and placed) to refrain from deciding that way in the circumstance is subject to cross-world luck, whether the world is deterministic or indeterministic. Bob, in particular, is lucky that he decides as he does rather than otherwise because, in a nearby possible world where the salient causes and background conditions which are relevant to his deciding one way or another in the circumstance are the same (including his reasons, his character traits and even his way of deliberating), he decides otherwise instead. If we make Bob's world deterministic, we don't thereby eliminate the nearby possible worlds where Bob's counterpart does otherwise in conditions which don't significantly differ from Bob's. (Though, of course, since the worlds at issue are deterministic, either the past or the laws must differ in some way to produce different outcomes.) Let me elaborate.

Consider again Bob's decision to cheat by waiting until 12:02 to toss the coin. Take Bob's counterpart, Bob*, who lives in a deterministic world.¹⁰ The events in Bob*'s deliberative process, his decision to cheat, and also his dispositions to deliberate and decide in various ways in the circumstance are the same as in Bob's case. Like Bob, Bob* is, up to noon, psychologically disposed to decide to do the right thing (for the same reasons which motivate Bob). Neither his deliberating as he does nor his deciding to cheat reflect an unequivocal commitment to cheating over doing the right thing, or a psychological incapacity to resist the temptation for the money, or the like. Both Bob and Bob* are, at noon, normally motivated and disposed to decide either way in the circumstance (and in possession of the skills and opportunities required to decide either way). Thus, it seems clear that the fact that Bob* is determined to decide to cheat rather than deciding to do the right thing isn't grounded in the attitudes, tendencies, skills and features of the circumstance which are relevant to what he is motivated and disposed to do in the circumstance.

Indeed, why should these salient causes and conditions render Bob* causally determined to make the wrong decision rather than the right one, instead of rendering him causally determined to make the right decision rather than wrong one? There is a deterministic world in which Bob*'s counterpart, Bob**, with the same relevant motives, skills and opportunities, decides to do the right thing instead. But the same set of causal factors cannot causally determine mutually exclusive outcomes: one set of attitudes, tendencies, skills and opportunities may causally explain either one of two mutually exclusive decisions (Bob*'s and Bob**'s), but it cannot both causally determine that Bob* will decide to cheat and causally determine that Bob** will decide to do the right thing.

The truth of determinism does not make it true that, if Bob* had not decided to cheat in the way he did, he would have decided to cheat in some other way, but would in no case have decided to do the right thing with those reasons in that circumstance. This counterfactual is false of Bob*, just as it is false of Bob (his indeterministic counterpart). If an agent is unequivocally motivated to cheat (which, by hypothesis, Bob and Bob* aren't), the nearest possible world where she voluntarily refrains from cheating is one where something significant in her relevant reasons or capacities is different. In contrast, the nearest possible world where Bob*'s counterpart voluntarily refrains from cheating (let's say it's Bob**'s world) is one where nothing significant changes in his relevant motivations and condition, and Bob** refrains for the moral reasons for which Bob* was about to refrain when he decided to cheat.

To see this, consider an action whose freedom is not in question. The first time Bob* goes golfing, on a very rainy day, and luckily holes a difficult putt, his poor golfing skills and the course conditions (together with his attempt to hole the putt and the action-enabling background conditions) are not what determines that he holes the putt. Even though Bob* is determined to hole the putt (this event is determined to happen since billions of years ago), he is really lucky to succeed in holing it, given his poor skills, his merely acceptable physical condition, the recent rain and the strong wind. These salient causes and conditions, far from determining or guaranteeing that he will succeed in sinking the putt rather than failing, make it very unlikely (in some sense which doesn't imply the falsity of determinism) that an event of this type will happen. (In nearby possible worlds where Bob* tries to sink the putt in the relevant conditions, he fails.) My point is that Bob* is determined to hole the putt only because his world is deterministic, as opposed to a skilled golfer in optimal conditions who is determined to hole an easy putt whether or not her world is deterministic.

I am assuming that, if the world is deterministic, when amateur golfers luckily sink putts, they do not exhibit a power to guarantee that they will sink a putt in those conditions, and when non-alcoholics drink beers because they desire to do so more strongly than they desire to do anything else then, they do not exhibit a compulsion for alcohol, and they are not determined to drink in virtue of, say, their desire for a beer and a lack of obstacles to their drinking a beer (even though these factors do cause their drinking and the drinking is determined). The kind of action determination that may signify great golfing skills and perfect conditions is not the action determination that affects every single action in a deterministic world, but a kind possible both in deterministic and indeterministic worlds. Whether or not the world is deterministic, some skilled golfers and alcoholics and priests are determined to sink putts and drink beers and burn papal bulls in given circumstances (in some specific way or other), in that, if they didn't do these things in one way, they would do them in another, holding fixed their relevant attitudes, tendencies, skills, opportunities and other relevant features of their circumstance. In contrast, what explains that the lucky golfer is causally determined to hole the difficult putt is the very same fact which renders him causally determined to hole the putt in the perfectly specific way in which he does – namely, a huge set of past states and events and laws of nature, and not the salient causes of the lucky success. What contrastively explains his succeeding instead of failing are things like an unlikely slip of the foot and the wind blowing at just the right moment – things which don't happen in the nearby possible worlds where he tries for the first time to sink a putt, in the rain and after a night of insomnia.

It's true that, unlike in Bob's indeterministic world, there is in Bob*'s deterministic world some fact about the past or the laws which could in principle be adduced in a contrastive explanation of Bob*'s deciding as he does rather than otherwise: the laws of nature and the whole history of Bob*'s world, up to the moment he decides to cheat, are compatible only with that decision to cheat (indeed, with that very one, down to all its microphysical details). Similarly, in the case of Bob** (the deterministic counterpart who does the right thing instead), there is a contrastive explanation of his deciding not to cheat instead of deciding otherwise (in that very circumstance and with those very reasons and dispositions, though not necessarily in that very moment): the whole history of the world up to that instant plus the deterministic laws of nature are compatible only with Bob**'s deciding not to cheat, exactly as he does. But I cannot see what good it does to Bob* and Bob**, regarding the freedom or rational control they have, that there is a contrastive explanation of their decisions, since these contrastive explanations do not feature causes which distinctively rationalize, or even saliently cause, their deciding as they do rather than otherwise.

Given these modal facts concerning Bob's and Bob*'s decisions, Bob* is as unlucky as Bob about the fact that he decides to cheat instead of deciding to refrain from cheating. Therefore, Bob's decision is chancy or lucky not because it is undetermined, but because Bob is psychologically disposed to decide differently without deliberating in a substantially different way, and holding fixed his relevant attitudes, tendencies, capacities and circumstance. Even though Bob* is not subject to what Mele calls present luck (which requires indeterminism by definition), he is subject to cross-world luck at the time of decision, since his deciding to cheat instead of deciding not to is chancy, and the fact that he makes the decision he makes is significant – in this case bad – for him.

Consider a different case, which will be useful to illustrate some of the claims I will make in the following section. Imagine that you are trying to decide whether to tell your friend now (in the following minutes) that she will probably be fired soon, or to wait until after the road trip that you just started with her. You think that you should do whatever is better for her, and that, probably, the sooner she knows the better for her, but you are undecided because you don't want to spend four days talking about her possibly moving to another city or country to find another job, and four nights checking the Internet and making plans. You feel guilty about the fact that you are thinking of hiding it from her out of self-interest, but you are less than certain that getting the news now would be better for her than getting the news after the trip. (You think that it may be good for her to have four carefree days before a potentially long period of anxiety. You also think, reasonably, that if she had a real interest in knowing about the next dismissals in the company, she would have found out already.) You think that, probably, what would be best for you to do all things considered is whatever is best for your friend, but you are not sure what's best for her, and you really don't feel like giving her the bad news now.

We can imagine a tree representing the relevant nomologically possible alternative paths that involve your action. At a certain point, your friend smiles at you happily and starts talking about possible trips you could take together in the Summer. This makes you feel guilty because in the Summer she may not be in a position to go on a trip, and you feel a strong urge to interrupt her and give her the bad news now, at t1. But she's difficult to interrupt, and by the time she stops talking, t2, your urge to tell her is not that strong. Moreover, she has raised an issue that she almost never raises and you want to take advantage of this, so you don't tell her at t2 either, though you are about to. Then, as you are talking about more delicate issues, she asks you something that forces you to either tell her that she may be fired soon, or lie to her and pretend you know nothing about the company's next dismissals. This happens at t3, the time on which I will focus.

Imagine that, before the first relevant branching (t1), two alternative actions of yours were nomologically possible at t1:¹¹ interrupting her with a fast I have to tell you something, or remaining silent (as you actually did). If you had interrupted her at t1, two different actions would have been nomologically possible afterwards: telling her the truth, and inventing some lie about what you had to tell her. If you had invented any of the lies you could have invented, you would have had more opportunities to tell her the truth in the following minutes. In some branches that depart from your lie, you feel guilty that you lied to her and tell her the truth, without stopping to form a better judgment; in others, fear at being discovered in your lie makes you, without forming a better judgment, decide not to ever tell her yourself, and to arrange things so that someone else tells her immediately after the trip. In other branches that depart from your lie, you make an effort to form a better judgment; in some of these branches you judge that you should tell her, and in most of these you consequently tell her, while in others you judge the opposite and decide to tell her after the trip. There are branches where you judge you should tell her but then reconsider and judge the opposite, and branches where you judge you shouldn't tell her yet but then reconsider and judge you should. It is not the case that, in every branch which ends in your telling her, you judge it best to do so, nor that, in every branch which ends in your decision not to tell her during the trip, you judge this is what's best all things considered. And some branches which end in opposed decisions aren't significantly different from each other.

But you remained silent at t1. After this, and up to t2 (when she stopped talking), you had various urges to interrupt her, and if you had done this at any point, both telling her the truth and inventing a lie would have been causally open to you. However, you didn't interrupt her between t1 and t2, and at t2 (when she stopped talking), you decided to continue the conversation. We're now just after t3, when she asked: Did you hear if they're going to fire someone else soon?, taking you completely by surprise.

Imagine that three alternative reactions to her question are nomologically possible up to t3. Your reaction is not something you control and, given the little time you have to say one thing or another, and your inability to remain cold and consider the issue further before answering, what action you perform is determined by how you react when she asks the question. First, her question can make you think that she clearly would prefer to know now rather than in four days, and this, your guilt and your hurry to answer her will lead you to decide to tell her. Second, you can think: How can I tell her now? I've been hiding it for hours, she'll be really mad that I was not telling her because I wanted to relax, decide to pretend you never got news about the next dismissals, and answer that you heard nothing. And third, you can think that you really need more time to figure out what's best to do, decide to distract her to avoid answering her question, and do this by pretending that the highway exit you just passed was the one you had to take and asking her to check the map fast. Imagine that you react in the first way and decide to tell her. Each alternative reaction to her question is a difference-maker, but not, I take it, one that is suitable to contrastively explain each decision with respect to the other two in the way that is satisfactory for someone who is worried that cross-world luck at the time of decision undermines rational control. Thus, your decision is subject to present luck.

It seems clear to me that, if your world had been deterministic and, say, only the first one of your psychologically possible reactions had been nomologically possible at that moment in particular (but the actual events in the sequence and your dispositions to deliberate and decide differently at each point in the process had been the same), your being causally determined to answer her question with the truth would not have enhanced your rational control over your action with respect to the indeterministic scenario just described, because the causal determination at issue would not have been grounded in your capacities, attitudes and tendencies (in particular, this causal determination wouldn't have amounted to ADC). Why would the relevant capacities, attitudes and tendencies render you causally determined to answer with the truth, instead of causally determined to lie, or to postpone the issue? During the span in which you could have acted, you were psychologically disposed to continue hiding the news from her. Neither your reacting like that to her question nor your telling her the truth reflected an unequivocal commitment to telling the truth over lying or postponing the issue, or a stably strongest motivation to tell the truth, or a psychological inability to lie. At various points, you were psychologically disposed to continue deliberating in a way that would have led you to continue hiding the news, not necessarily after doing anything significantly different from what you actually did – in particular, not necessarily after judging you should hide the news until after the trip. Up to a moment before your decision, you were psychologically disposed to decide otherwise instead, without first doing anything substantially different in deliberation, such as judging it would be best to hide the news, deliberating more carelessly or self-deceptively or self-indulgently, or the like.

Thus, determinism would not eliminate your luck regarding the fact that you told her the truth then instead of lying then, and analogous conclusions are true of Bob's case and of the rest of the split decisions which must be undetermined to be free according to standard libertarians. Therefore, the indeterminism required by standard libertarians in no way brings about cross-world luck at the time of decision, and hence this kind of luck is not only a problem for libertarians; it is a problem for compatibilists as well.

Consider this possible objection to my argument: Even though Bob*'s decision is subject to deterministic cross-world luck, the indeterministic cross-world luck which affects Bob is worse for control, because only Bob is, given all the laws and past up to the time of his decision, able to decide otherwise. Bob*, in contrast, is able to decide otherwise only with at least slightly different past or laws. Only Bob is, at the relevant time, causally able to choose either option. Or, as some would put it, only in Bob's case are the different alternatives really available at the time of decision; unlike Bob, Bob* is not really, or perhaps not himself, able to decide otherwise (though he counts as able on compatibilists analyses of the ability to do otherwise), and hence he is not as lucky as Bob to decide as he does rather than otherwise.

I myself have, sometimes, the intuition that indeterministic cross-world luck is more real than deterministic cross-world luck (I still have incompatibilist intuitions). But note that there is a clear tension between choosing this line of resistance to my argument and holding, with most contemporary compatibilists, that whether the world is deterministic or indeterministic is not relevant to whether we have free will. Compatibilists often insist that the difference between drinking another beer compulsively and non-compulsively has to do with certain dispositions of the agent and is independent from whether the world is deterministic or indeterministic; being determined to voluntarily drink another beer is not sufficient for being compelled to do so (as almost all incompatibilists agree). According to most compatibilists today, it does not matter for free will whether you are causally able to do otherwise, or able to do otherwise (if “able” is the right word here, which not all compatibilists claim) only in some sense which is implied by a lack of compulsions, phobias, physical incapacities, and the like. I cannot see how holding this is compatible with claiming that deterministic cross-world luck at the time of decision is less worrisome than indeterministic cross-world luck at the time of decision.

On the other hand, it is true that libertarians and compatibilists are not exactly in the same position regarding cross-world luck at the time of decision. Libertarians claim that an adult has free will only if he performed undetermined directly free actions (i.e. actions whose freedom does not derive from the freedom of actions in their causal antecedents); and these undetermined actions, on standard libertarian accounts, are subject to cross-world luck at the time of decision. Hence, if a standard libertarian grants that cross-world luck at the time of decision precludes free action, she is giving up on free action entirely. In contrast, a compatibilist need not require directly free actions which are subject to cross-world luck. Thus, a compatibilist who grants that cross-world luck at the time of decision precludes free action still has plenty of room for free action, unlike a libertarian who grants this.¹²

However, I don't think this dialectical difference is that important. Actions which are likely to be subject to cross-world luck at the time of decision are very common, and many of them are prima facie morally significant. Saying that we cannot exercise free will in performing these actions would be admitting that we lack sufficient grounds to morally praise and condemn many of the actions that we, in everyday life, would want to praise and condemn. Moreover, many of the decisions and overt actions through which we acquire and maintain our tendencies and attitudes are, plausibly, subject to cross-world luck, and most compatibilists think that an adult acts freely only if the tendencies and attitudes on which she acts were acquired and maintained freely. It seems to me that their commitments to historical conditions on free action should lead these compatibilists to accept that, if cross-world luck precludes free action, most of our adult actions, including actions which are not subject to cross-world luck, are not free.

3. Does cross-world luck at the time of decision preclude free action?

For all I have said, decisions (and subsequent actions) which are subject to cross-world luck are unfree because of this. But why think that cross-world luck makes a decision unfree? I cannot find any principled reason why a decision's freedom should depend on whether or not the agent was psychologically able to decide otherwise in the circumstance only after deliberating in a significantly different way.

The fact that a given agent would have decided otherwise only if she had deliberated in a significantly different way need not reflect anything significant about the agent's motivations, values and character traits, but may be due to unpredictable events which indirectly affect deliberation, or due to psychological facts which do not reflect anything important about the agent, or due to other coincidences. Take a given agent in a given circumstance, who is motivationally divided between various options. Holding fixed her relevant attitudes, tendencies and skills and the aspects of her circumstance which are relevant to what reasons and capacities she has at the time, and changing only non-significant events in her deliberation or in her environment, we can generate cases where she is subject to cross-world luck at the time of decision and cases where she is not, in such a way that the very same significant factors contribute to accounting for the different conditionals which are true of each case. Take, for instance, the case discussed above where you are trying to decide whether to give the bad news to your friend, holding fixed your beliefs, desires, aims, preferences, values, habits, character traits, capacities, and so on, and arrange along the branches considerations that may occur to you, moods you may be in, and events which may distract you or affect the direction of your thoughts, in such a way that no branches which end in different decisions contain the same significant deliberative actions. It sounds to me rather arbitrary and hence unfair to say that, in the original version of the case, cross-world luck at the time of decision precludes free will, but in this new version you may act freely in the way required for moral responsibility, since you are not subject to cross-world luck at the time of decision.

Relatedly, take a given agent with a given set of attitudes, tendencies and skills, and place her in two different scenarios, one where she has more than one live option and is vulnerable to cross-world luck at the time of decision, and another where she has only one live option and hence isn't subject to cross-world luck. Also in this case (though perhaps less than in the previous one), it sounds rather arbitrary to me to say that only in the second circumstance can the agent act freely; she has the same set of attitudes, tendencies and skills as in the first scenario.

On the other hand, we often act as we do rather than otherwise quite arbitrarily, due to hastiness, absent-mindedness or confusion, or in such a way that we regret what we do immediately after doing it. If acting like this is not a good reason to be exempted from blame, it seems that being subject to cross-world luck at the time of decision shouldn't be either.

Relatedly, it seems to me that philosophers with a broadly Strawsonian view of moral responsibility have reason to accept that cross-world luck at the time of decision does not preclude free action. Bear in mind that we don't blame or praise someone any less if we know they were torn when they acted. It would be inappropriate, and very strange, to exempt someone from praise or blame upon learning that they acted against reasons which pushed them to do otherwise in the circumstance. In particular, it would be inappropriate to exempt someone from blame upon learning that they acted akratically (i.e. that they knowingly acted against their better judgment, free from compulsion and force). We often praise people for resisting temptations and blame them for failing to resist them. These points are not controversial among participants in the free will debate. But, if this is granted, and given that we cannot know all the counterfactuals which are true of any given deliberative process nor whether any given decision is subject to cross-world luck (the most we can aspire to reasonably guess are the significant aspects of agents' motivations, capacities and circumstances), it seems that a Strawsonian has no way to claim that cross-world luck precludes free action without generating a tension in her view.

Fischer and Ravizza (1998), for instance, adopt a broadly Strawsonian view of moral responsibility (5–8), and their sufficient conditions for free and morally responsible action do not rule out cross-world luck at the time of decision or overt action.¹³ On the account Fischer and Ravizza develop, an agent exhibits the kind of control over behavior which is required for moral responsibility if and only if the mechanism that issues in the action is both the agent's own and moderately reasons-responsive. The mechanism leading to an action is simply the process that leads to the action, or the way the action comes about (38). A mechanism is the agent's own (as opposed to, say, a result of brainwashing or hypnosis) if, thanks to normal moral education, the agent has taken responsibility for behavior which issues from the mechanism. Taking responsibility, in the technical sense in which Fischer and Ravizza use the expression, is not a voluntary action; it consists, roughly, in coming to view oneself – based on suitable evidence gathered during one's moral training – as a source of actions which have consequences in the world and as a fair target of reactive attitudes, punishments and rewards in virtue of the way one acts in certain contexts (see especially 208–214). A mechanism is moderately reasons-responsive if and only if it is the case that, in the possible scenarios where the agent acts on the mechanism, she recognizes sufficient reasons for doing various things – some of which are moral reasons (76–81) – in a way that is “minimally grounded in reality” (73), coherent and understandable by a third party who knows the agent's preferences (71–73), and there is at least one sufficient reason to do other than she actually does for which the agent acts in some possible scenario where she acts on the mechanism (73–75). Compulsive and phobic deliberations, and the deliberations of certain psychopaths and mentally insane agents, for instance, are not moderately reasons-responsive. Note that one may act on a moderately reasons-responsive mechanism without being able to do otherwise (on any reading of this expression) in the actual circumstance; the possible scenarios which are relevant to whether a mechanism is moderately reasons-responsive do not represent possible continuations of the mechanism's actual operation.¹⁴ Hence, even in cases where one is motivationally divided, meeting Fischer and Ravizza's sufficient conditions for acting freely does not require being subject to cross-world luck at the time of decision or overt action. But meeting these conditions is compatible with being subject to cross-world luck at the time of decision or overt action.¹⁵ For instance, Mele's cases and the case I offered above satisfy Fischer and Ravizza's sufficient conditions for free action.

In any case, it seems clear that, at least in many cases, the things we would usually take to ground blameworthiness and praiseworthiness are compatible with there not being a significant motivational difference-maker for every pair of alternative decisions the agent is suitably disposed to make with those reasons in that circumstance. Most people who are paradigmatically prima facie responsible are motivationally divided very, very often, with countless significant and insignificant consequences for what they decide and do, and many prima facie blameworthy and praiseworthy decisions and overt actions are performed in a divided state of mind. Plausibly, many of these decisions and actions are subject to cross-world luck. I find no principled reason to believe that, because of this, these decisions and actions are not after all free and responsible.

4. Conclusion

It's often pointed out that, while determinism is a prima facie threat to free will (primarily because it seems to preclude the ability to do otherwise), the indeterminism required by libertarians is as well, because this indeterminism seems to bring about luck or chanciness in how we act. Hobbes (1651), Hume (1748), James (1884), Hobart (1934), Nowell-Smith (1948), Ayer (1954), Smart (1961), Van Inwagen (1983), Nagel (1986), Waller (1988), Double (1991), Strawson (1994), Haji (1999), Smilansky (2000), Almeida and Bernstein (2003), Cohen (2006), Mele (2006) and Fischer (2012) are among those who accept that indeterminism at the time of action would bring about luck or chanciness in how we act. The argument in this paper bears directly on this traditional assumption. I argued that cross-world luck at the time of decision is inherent to certain decisions, and independent from whether the world is deterministic or indeterministic. The decisions which, according to standard libertarians, must be undetermined to be free, have deterministic counterparts which are just as subject to cross-world luck at the time of decision. In conclusion, cross-world luck at the time of decision challenges compatibilism as well. If this luck decreases control, this is a problem for any account, compatibilist or incompatibilist, which wants to leave room for the possibility of free decision in cases where agents are suitably disposed to decide differently without deliberating in a substantially different way.¹⁶

On the other hand, I motivated the view that free action is compatible with cross-world luck at the time of decision, especially because there needn't be significant differences between an agent who is subject to cross-world luck at the time of decision and an agent who is not. It seems to me that a view which requires a lack of chanciness in whether a prima facie free and responsible agent voluntarily decides one thing or another is a view which imposes too demanding conditions on free action.

Acknowledgments

I am grateful to Al Mele for very helpful comments on a previous draft, to Stephen Kearns for interesting discussions on the main argument in this paper, and to two anonymous referees for Philosophical Explorations, whose comments significantly improved this paper. Research leading to this paper was supported by the Catalan Government's Agència de Gestió d'Ajuts Universitaris i de Recerca and by the Spanish Government (research project FFI2010–15717).

Notes on contributor

Mirja Pérez de Calleja obtained her Ph.D. in Philosophy from the University of Girona (Spain) in 2013, with a dissertation titled The Problem of Free Will and Determinism: An Incompatibilist-Friendly View. Her research focuses on free will, philosophy of action more generally, and ethics.

Notes

1 By “standard libertarians”, in this paper, I mean libertarians who place the required indeterminism at (or instants before) the time of decision, intention-formation or action, and not, for instance, in prior events which happen during deliberation, such as the coming to mind of certain considerations, the wandering of attention at a crucial point, and the like. It is worth noting that no libertarian has defended a non-standard libertarian view.

2 Mele notes that similar luck-related challenges to libertarianism were offered in Van Inwagen (1983, 140–141), Nagel (1986, 113–114), Waller (1988), Almeida and Bernstein (2003), Cohen (2006), Haji (1999) and Strawson (1994).

3 Several authors have relied on luck-related considerations to argue that libertarianism is false because the indeterminism required by standard libertarians undermines rational control, since it brings about present luck. But Mele himself does not argue for this. He thinks that present luck poses a challenge to any standard libertarian account (see, for instance, 9, 66, 67 and 70), but, far from arguing that this challenge cannot be answered, he offers a daring soft libertarian solution to the problem of present luck (113–133). Alternatively, he offers a modest libertarian view, a non-standard (see note 1) libertarian account which, he thinks, leaves present luck out of the picture (10–14). See Levy (2011, Chap. 4) for an argument against the claim that a view like modest libertarianism avoids the problem of present luck.

4 For Kane's hypothesis on the relation between metaphysical indeterminacy and causal indeterminism in undetermined free actions, see Kane (1996, 128–130, 1999a, 108 and 112–113, 1999b, 224–226).

5 The libertarian can grant that perhaps what happens just before a certain agent in a certain circumstance decides to A cannot happen immediately before that agent in that circumstance decides any other thing or postpones decision. But this is neither a big concession nor help against the luck problem, because non-significant difference-makers in deliberation, which do not give the agent stronger motivation to decide as she does rather than otherwise, do not render actions contrastively explained in the way that is satisfactory for the proponent of the luck problem.

6 See also Mele (2006, 7–9 and 58–61).

7 Correspondingly, there's nothing in John's character, motives and way of deliberating (nor in these plus relevant action-enabling background conditions) which can be adduced to explain why John decides to arrive late rather than deciding to arrive on time. For instance, it's not that, in the world where John decides to arrive on time, his commitment to doing whatever he judges best, or his fear of confrontation, is stronger than his corresponding commitment or fear in the world where he arrives late.

8 On Mele's construal, present luck requires indeterminism by definition (see, for instance, Mele 2006, 7–9). In this paper, I use “cross-world luck” (which is often used as a synonym of “present luck”) to encompass both present luck and what I regard as its deterministic counterpart.

9 Whenever I say in this paper that a decision is lucky, I mean that the decision is subject to cross-world luck (see note 8), so that the agent is lucky about the fact that she decides as she does rather than otherwise. I take all my claims to be consistent with the standard view that people are lucky or unlucky about events only in contrast to other events that could have happened instead: one is lucky that some thing happens rather than something else (cf. Pritchard 2004; Coffman 2007; Driver, 2013). Importantly, the argument in this paper assumes what Driver (2013) calls the modal view of luck, namely the view that “luck is not simply epistemic but instead corresponds to flukes – occurrences of this world that fail to be occurrences in the relevant set of nearby possible worlds” (12). (The main alternative to this view is what Driver calls epistemic reductionism about luck, the view that “luck simply reflects a state of ignorance on the part of either the luck attributor or the ‘lucky’ individual” (12). On this view, in a deterministic world, nothing is really chancy, unlikely or lucky (everything is inevitable), and it is because we are ignorant of relevant facts that some events appear chancy, unlikely and lucky (12–14).) I would like to thank an anonymous referee for pointing my attention to Driver (2013). I am also grateful to E.J. Coffman for directing me to Pritchard (2004) and Coffman (2007). Interestingly, Coffman (2007, 389–390) argues that there could be chancy events and lucky people in deterministic worlds, and he mentions that an account of luck such as the one he offers could be used to defend libertarianism against the luck objection (note 16, 394).

10 The libertarian who thinks that not only free action, but action itself is incompatible with determinism will get off the boat here. (Thanks to an anonymous referee for pointing this out.) For a very interesting and insightful defense of agency incompatibilism, see Steward (2012).

11 Or, more plausibly, roughly after t1, and not necessarily at the same moment (see note 5 above).

12 Thanks to Al Mele for bringing this to my attention.

13 Thanks to Al Mele for making me aware of this.

14 In this connection, see especially 37–38 and 73–74, and note 4 on 244:

When the mechanism that issues in action is appropriately reasons-responsive, it does not follow that the agent could have responded differently to the actual reasons. Rather, when the mechanism is reasons-responsive, it has the general capacity to respond differently to the actual reasons for doing otherwise (emphasis in the original).

15 In fact, Fischer (2012, Chap. 6) has recently argued that cross-world luck at the time of decision does not preclude free and morally responsible action.

16 I do, however, grant that, even though indeterminism does not decrease control (at least not by bringing about cross-world luck), cross-world luck does pose a problem for libertarians that it does not pose for compatibilists. Libertarians claim that indeterminism is required for free will, and hence that indeterminism either makes rational control possible or enhances rational control significantly enough to ground moral responsibility. But the actions which, according to libertarians, must be undetermined for agents to have free will are precisely those which are subject to present luck. And, if all the abilities that you have only if libertarianism is true are abilities to chancily choose B over A, it's not clear how having these abilities enhances your control over your choices with respect to the control you can have in a deterministic world. (For a similar point, see Clarke 2003, Chap. 6.) So there is, after all, a luck problem for libertarians only, since libertarians' task does not end, like compatibilists', with arguing that cross-world luck at the time of decision is compatible with free action. But this is not Mele's luck problem for libertarians, which was the topic of this paper.