What Doesn't Matter To (or About) Free Will
In the vein of my meaning of life post, I want to write something on maybe the second most common question associated with philosophy – do we have free will? Like the meaning of life question, this is not an issue I care all that much about, but I care moderately more about it, and am moderately more opinionated about it than the meaning of life question, which I really had to struggle to get any take on at all.
The three classic positions on free will are: libertarianism, which says that determinism is false, and we have free will in virtue of being able to change the future; hard determinism, which says that determinism is true, and in virtue of this we can’t have free will; and compatibilism, which says that we can have free will even if determinism is true. Like most philosophers, I am most sympathetic to compatibilism out of these options, but I don’t really believe any of them. A quick summary of my opinion is that:
Determinism is probably true in both senses of the word, but is almost definitely true in the relevant one.
Determinism, in any sense, isn’t especially relevant to whether we have free will or not.
For more relevant reasons related to personal identity, we do not have any strong version of free will.
Even if we had free will, it wouldn’t ground the usual things people like to use free will to ground, like desert, so the question is relatively pointless.
I. Determinism is True
The first point requires distinguishing two different senses of “determinism”. The first is “physical determinism”, the idea that there is some set of physical laws we could use to determine anything that might happen at any point in the future, given complete knowledge of the present moment. The other sense is “logical determinism”, which roughly says that, at any given point of time, any statement we might make about the future is already either true or false.
Let’s start with logical determinism, the one that I believe is simply true. Are there any views able to challenge it? Since we have distinguished it from physical determinism, it seems like any view that conflates the two is off the table - even if we couldn’t possibly know right now whether a statement about the future is true or false, it seems like we could still say that the statement is already true or false. This disqualifies, for instance, views that suggest that there is some degree of random chance in which choice we make. That is a comment about the predictability of what will happen, not whether there is a stronger sense in which it actually could have, before happening, turned out otherwise. We flip a coin and have no idea whether it lands heads or tails, but you are in the unchangeable past of whichever outcome does come to pass. This perspective does not rely on physical determinism whatsoever – whether we could derive the past by following physical laws backwards or not, we still consider the past unchangeable. Why should we, in the past, not reflect forward on this future in the same way?
Any time when we are unsure (perhaps irreparably unsure) about the outcome at some point in time t, we are in the history of the point t, when what has happened is set in stone. If the history of any given moment is set in stone, it is trivial to imagine time t+1 second, in which both the outcome at time t, and the present moment are both set in stone, unchangeably, in the historical timeline of t+1. But perhaps there isn’t a single straight line connecting the outcome at time t with the present moment when reflecting backwards, perhaps there are multiple timelines that contain the present moment, but do not propagate forward to the specific outcome in question at t and t+1. A branching model of time indeed seems both intuitive, and implied by some plausible physical theories like the Everett interpretation of quantum mechanics. As an Everett leaner myself, I am forced to find this scenario compelling, however, I don’t think it substantially challenges logical determinism.
The most straight-forward interpretation of this branching is that both branches, in an equally real sense, occur. It may be that they don’t share a “timeline” in some sense, but there is still an inescapable correct answer before the branching occurs, which is “both of these two outcomes will equally occur at time t”. Is the fact that the two are causally separated relevant to whether this makes the outcome indeterminate, any more than if there were two separate things we could make statements about in the present which were mutually outside one anothers’ lightcones? If it can be deterministically true, if unknowable, that I bite into a sandwich and Zog of planet Grorg seven trillion light years away bites into a sandwich, can’t it be deterministically true that I bite into a sandwich in one Everett branch and also don’t bite into a sandwich in a different branch?
The obvious way to try to rescue indeterminism in this branching universe is to suggest that the branching structure is like a river, and by choosing, we steer the canoe of our futures down one path rather than the other. This captures an intuitive sense of choice, but if we consider it possible to validly choose one path rather than another, then that entirely defeats the counter from the branching structure. Whether there are any laws we could use to determine which branch we will go down in advance, there is, in fact, only one valid branch - the one that we will steer down. We can trace our path through time back from the history of t+1 through time t and to the present once again, and find the only timeline we will propagate forward along from the present. If we actually do steer down only one path, then there must be a fact in the world that determines the valid branch, even if this fact is unknowable. Again, knowability is about physical determinism, logical determinism only requires the fact to exist.
A final way to try to deny logical determinism is to take issue with the idea that there can be a “fact” about what will happen in the future in the needed sense at all. One thought, for instance, is that distinguishing logical determinism from physical determinism requires saying that there is a fact about a future event at a point in time when nothing in the world grounds that fact. The fact is eventually grounded in a future event, but why should we say that it is true now? This claim seems to leave us in a weird spot. As I’ve just discussed, we normally consider facts about the past to be unchangeably true. If physical determinism is untrue however, we don’t just have reason to believe in the possibility of future events that can’t be predicted by extrapolating facts about the present forward, but also we seem to have some reason to expect there might be facts about the past that we can’t derive by reflecting backwards on evidence in the present. You could coherently deny this by saying that not every cause has a single unique possible effect, but that every effect has a single unique cause, and that every event has both a cause and effect (or some similar view of causality). This seems strange to me however, and is just a special half-baked form of physical determinism. A question remains of whether, if this form of physical determinism turns out not to be true either, facts about the past are nonetheless fixed?
But maybe you can bite the bullet that some facts about the past aren’t determined in the present. Even this might be too generous to this view however. Why just ask about the existence of facts relative to time, why not space? Must it be possible to gain enough information at any given point in space for it to be true, here, than an event elsewhere occurred? If not, then why must it be necessary to have enough information now to predict what will happen in the future, in order for it already to be a fact that it will occur? Take this view of epistemology far enough, and you essentially convert a fact into an object – a fact is not justified by, but in some sense identical with, a certain set of evidence in the world. Where this evidence is not, the fact is not. At the very least is seems as though, if this is the view of epistemology you need to retreat to in order to avoid logical determinism, you probably aren’t getting the version of indeterminism that will be satisfying to someone worrying about free will. “your choice is only determined in the same sense as causally inert choices you made in the past”.
All of this points to logical determinism as the only plausible conclusion about reality. By comparison, physical determinism seems rather weak. People usually appeal to quantum mechanics when discussing it. If physical determinism means we can predict all real futures with enough information, the Copenhagen interpretation denies physical determinism, and the Everett interpretation affirms it. I lean towards the Everett interpretation, and so, in the second sense of physical determinism at least, I am a physical determinist. In a previous draft of this, I wrote off alternative interpretations like DeBroglie-Bohm because I was under the impression that Bell’s Inequality disproved it. That said, I am a mediocre physics student at this point, especially about more advanced subjects, and since then I heard an interview with my extremely smart professor in which he argued that Bell’s Inequality does not disprove DeBroglie-Bohm. I lack the skills to judge this for myself, so I also want to mention this possibility, which would, I think, rescue a single branch version of physical determinism. I think all the evidence so far suggests that we can understand all facts about the future of this reality, in principle, with enough information about the present moment and the laws of physics, but this at least is in question (both because something like the Copenhagen interpretation might be right, and because there might be unknown physics that is indeterministic).
II. Indeterminacy Does not Give us Free Will
I don’t, however, think the difference matters much to how we think about free will. Two of the most classic boardgames of the 20th century are Candyland, and Trouble. Candyland determines outcomes through cards that are distributed at the beginning of the game, whereas Trouble determines them through a dice-rolling contraption people press each turn. I like to think of the difference for free will, between logical and physical determinism, as that between playing Candyland and playing Trouble. If we care about physical determinism, the problem for free will is that we could, in principle, look at the cards, and figure out who will win in advance. If we are playing trouble, maybe there is a fact about who would win, but there’s nowhere on the board to look where you could find this outcome. And yet the difference neither depends on skill, nor is legible to the players. Does it really matter to the free will issue whether we are playing Candyland or Trouble? Both games seem importantly different from, say, Chess, but you won’t find that difference anywhere in the determinism debate.
If we are to say that this isn’t fair, because while physical indeterminism is insufficient for free will, it is necessary, then it seems like we must do this by both saying that we can personally cause a particular outcome in an important way, and that which outcome we personally cause is a matter of random chance. This is a bit of a cliché point among compatibilists at this point, but it seems as though the “free” part of free will confuses many people out of any substantial focus on the “will” part. If both our will and random chance determine a choice, does this make it more a product of our free will than if only our will did? Or is it our will itself that is random? The key question is at what point in the causal path of our decision ought randomness to enter such that this makes us more free?
If we are randomly assigned a will at birth, does the fact that this will play out in a single way feel more acceptable for its random origins? If we are deliberating on a choice, does some random fluctuation in our thought process make the final choice more our own? It seems like what we want isn’t to be more Trouble than Candyland at any point in the process, but rather to be playing Chess in a way that only involves our own personalized coherent stream of intention. Strangely we just also want this to wind up in several possible plays. That there is no one logically certain winner. Mere physical determinism hardly seems like it’s enough for this, indeed it might take away some of our freedom by having chance play a role in the outcome somewhere our will otherwise would.
III. What Really Gives us (or Fails to Give us) Free Will
Insofar as the aspect of determinism worth caring about in free will is logical determinism, we do not have that aspect of “free will”. However, as I have pointed out, it simply doesn’t seem like this is the most important aspect of free will anyway. It isn’t even clear that it is a coherent one. What matters is whether a decision was really yours, and much less whether there is some metaphysical sense in which your choice could have really gone multiple ways in theory. Indeed, the two seem mutually undermining. Indeterminism perhaps matters in the more colloquial way, in that you can find the “possible choices” somewhere in the world where the decision is yours, but not in worlds where it wasn’t. Aspects of your reasoning process that wound up going one way or another due to missing causal connection with other contemplated options, but missing these connections in your head rather than outside of it.
Something like a choice is yours because the reason it didn’t go some other way is that the causal chain in your head “missed” the other options, rather than because something outside your head blocked your mind’s ability to causally connect a will to act in some way to the actual action. This seems relevantly like what we mean by “could have done otherwise” on close inspection, and it just doesn’t have much to do with either physical or logical determinism. It’s a matter of folk psychology rather than metaphysics.
We have free will in that a choice relates to our psychological states in the right way and not the wrong way, and specifically in that we can distinguish causes “inside” and “outside” of us. Free will is not, in the end, about determinism, it is about personal identity, whether a choice is “yours”. If my jailer locks me up, we might think my being locked up is an act of free will, but it doesn’t relate to my free will, I do not have the free choice to be or not be locked up. Insofar as the act is free at all, it the jailor’s free will that has been exercised. For many aspects of the mind, like thoughts and feelings, we can speak of them without needing to speak of a thinker or a feeler, even if we often do speak of them in those terms. Free will is different, in that the crucial thing the category itself rests on is an actor. No actor, no free will. In order for free will to exist at all, that is, we need to be able to say whose free choice something is, there can be no free floating free will. An event that happened to you is only the result of free will if you were the one with the choice. Given this, a reductionist view of personal identity seems to give trouble to free will.
As a reductionist concerning personal identity myself, this puts me in the position of being a reductionist concerning free will as well. Either I can say that I don’t believe in free will, because I don’t believe in “agents”, or I can say that I believe in something that is kind of like free will, and we can think of it as free will by degrees, but its freedom doesn’t come as some further fact in the world. Whenever I say someone’s choice is “free”, this is a shorthand. I’m not sure whether this should make me say that I don’t believe in free will, or to say that I believe in a rather deflated version of free will in which some events are useful to treat as “yours” in certain systematic ways that coincide with and construct our idea of these events as being freely willed.
IV. None of Us is Deserving
Most people are interested in free will in large part because they think it could make the difference between the possibility and impossibility of guilt, and indeed many people specifically don’t believe in desert because they don’t believe we have free will. In a final departure I take from the standard beats of the free will debate, I don’t think that, even if we have free will, it would justify desert. I simply don’t believe in desert either way. Roughly speaking, desert assigns the moral value of welfare improvements or harms, in part, based on whether someone has acted wrongly or rightly in some way.
It might involve any of a wide range of views about how exactly to portion this value out, from, at the extreme end, the view that the suffering of those who have done some wrong is good up to a point, to the more moderate view that some suffering of those who have done something sufficiently bad is good, and to the quite modest liability view that the suffering itself is never good, but that it counts for less in the guilty than the innocent, or could at least ground certain types of cost-benefit that would be an unjustifiable rights violation if the bearers of the costs were innocent.
I feel no draw to the versions of the view that take the suffering of the guilty to sometimes be an intrinsic good, but will admit the liability version produces results I find intuitive. All this said, the principled side of this judgment looks very dodgy to me, and I tend to think the simplest explanation for why something is good or bad and how good or bad it is is always how good or bad it is for the moral patient in question. Equal suffering in the guilty and the innocent is equally bad for each, and from here it follows for me that we should be very suspicious of any judgement that there is some other sense in which one is better or worse from the perspective of morality itself. This doesn’t appeal to free will or guilt or innocence anywhere, but simply to the basic moral instruction to look for the value of a feeling precisely where that feeling is found.
The deontological flavor of the liability view is able to accommodate this principle to an extent. This is the version of desert on which the suffering and happiness of the innocent and the guilty count the same amount, but the innocent have rights that the guilty don’t. It is, therefore, permissible to actively or intentionally harm the guilty for the greater good in some instances where it would be impermissible to do the same to the innocent. The main trouble with this for me is that, well, I’m not a deontologist. As I have discussed in the past, my problem isn’t so much with the suggestion that we treat the innocent in a rights-like way, but rather that, on close inspection, doing this is usually better interpreted by saying that there is something worse about harms to the innocent than those to the guilty, as an outcome. Splitting the ethics up makes it look like you aren’t making a strong claim about the relative badness of harms to the innocent and the guilty, but I’m suspicious that this is just a way of hiding the real judgement, that the suffering of the guilty and innocent count a different amount, or at least in significantly different ways. If so, we return to the problem of not valuing an interest to the degree and in the place it is found in the world.
The other way to attempt to rescue the desert judgement is to suggest that treating punishment as some form of good in itself is not treating the innocent and the guilty differently, but rather that the difference in judgement depends on a difference in some component of welfare besides raw suffering. An intuitive way to do this is to say that the suffering counts the same to each sufferer, but it is good for the victim or victims of the guilty person’s crime that the guilty person suffers, and so there is an additional, non-impersonal, non-discriminating value. This matches much of the rhetoric of people who call for harsh sentences or punishments of criminals, and has the appealing feature for someone like me that it rules out the possibility of applying desert to victimless crimes. To deserve punishment, you must have a victim, because otherwise there is no one that punishment is good for.
I think much of the appeal of this approach is based on the fact that, as a matter of fact, punishment of the guilty often is good, in more conventional ways, for victims. Namely, the victims often want the guilty to be punished, and feel happy or relieved when they are, there is a preference and hedonistic good that often coincides with punishment. While this is true, a direct appeal to these features is unpromising. For one thing, the guilty person will almost certainly have a much stronger preference/hedonistic opposition to punishment. Will the victim feel as differently about a 50 year versus a 49 year sentence as a prisoner will feel about the full extra year of imprisonment? The guilty might also feel vengeful towards the victim for this punishment, should we think in such cases that, for the same reasons that punishment of the guilty is good for the victim, that a similar punishment of the victim for the good of the guilty is justifiable?
Finally, our judgement about the justifiable level of punishment is perhaps somewhat sensitive to the good of the victim, but not strongly. If we really are justifying the punishment of the guilty through the good of the victim, then should we issue a much harsher punishment to a pickpocketer whose victim feels as vengeful about it as a typical rape victim? Or should we issue a weak punishment to a rapist whose victim only feels as vengeful about it as the typical pickpocketing victim? At the very least we usually don’t think unusual vengefulness on the part of a victim justifies disproportionately harsh punishment, we don’t go around executing those who happened to pickpocket extremely angry and sadistic people. On the other hand when it comes to leniency, we do seem somewhat interested in the victim relative standard, we place some value on forgiveness and restorative justice, some crimes aren’t even prosecuted unless the victim presses charges. If a victim doesn’t want an offender punished harshly, that matters to people to at least some extent, but again, it isn’t going to cause us to treat a rapist like a pickpocketer.
This latter fact is at least easier to explain without referencing anything new, even those who don’t believe in desert typically believe a certain level of punishment can be justified for deterrence reasons. The former is not so easy to write off. It might make sense to say that there are two components to deciding the just punishment (well really three once you count deterrence), victim’s good and appropriate upper limit to punishment, but if it does, this is because it captures the relevant intuitions about all of these things mattering. It entirely gives up the justificatory role of victim’s good (after all, a victim independent desert term has been entirely reproduced within the framework), while the victim’s good is relegated to the minor role of determining how much mercy to provide between the limits of just punishment and deterrence, where such a gap exists.
The alternative way of understanding desert as being rooted in the good of the victim is to propose an alternate version of “the good” of the victim, one not based on preferences or happiness, but just whether “justice is served” against the perpetrator. This essentially takes the impersonal good version of desert, and personalizes it by tacking it onto a person. It honestly looks quite ad-hoc to me, this is just the same sort of objective threshold for punishment an impersonal view would judge, but tacked onto the welfare function of a person who might not want the punishment, might never know about it, might even be repulsed by the idea that someone has been harmed for their sake. In many or most cases this is not how the victim will feel, as discussed earlier, but that hardly changes the implausibility of separating this type of good from more conventional versions of a victim’s good.
The final way to try to draw up a more personalized account of the good in desert, is to locate the benefit with the punished. This sounds extremely strange on its face, but I can imagine a certain way of trying to justify this. It is relatively intuitive to many people that autonomy, or being respected as agents, is part of living a good life for moral agents. There is appeal to this in the positive case of having one’s will respected, even when one chooses in a way that is otherwise bad for oneself – a common objection to paternalism. We could imagine that a flip side of this is the idea of responsibility, that the good life for an agent involves others respecting them as agents, and punishment is a way of respecting this agency, because it involves recognizing that they have made a choice, freely, in the wrong way. Even if you say that someone does not exactly benefit from punishment when they are punished, you could argue that they benefit from the threat of punishment in general, because it recognizes them as being in each moment fully capable of making either the right or wrong choice in a way that matters.
This still is a bit of a difficult headspace for me to get into. Even if autonomy is in someone’s interest, and part of this interest involves accepting the consequences of making the “wrong” choice in order to recognize their agency, it doesn’t automatically follow that we should make punishment one of those consequences where we otherwise wouldn’t have a practical reason to (barring this the reason would just be whatever this independent reason is, like the practical deterrence and prevention considerations). We could imagine that agents generally want to live in a society where this is a consequence of wrongdoing by agents, and so punishment respects not just the will to make a decision with these consequences, but it respects this will to provide punishment as a consequence in general. But we will need some other explanation for why they want to live in such a society – is it because they would be happier in case they were victims? Because society would be safer and happier overall? Because it seems like some impersonal social good? Then, again, these explanations and not the personal autonomy one does the primary justificatory work.
I think once again, the strongest appeal I can imagine from an account like this is rooted in an independent fact about experience that tends to be true, but seems importantly different. Guilt is the emotion that people who are mentally well-off feel if they act wrongly, and punishment is a way of respecting guilt. Therefore punishment is a good for the best version of the punished person. Again, this hardly seems to translate in a clean way into the idea that punishment is good for the person who deserves it.
What all of this leaves me with, once again, is the view that desert doesn’t come from anyone’s good, it is a value quite aside from whether it is valuable to anyone, and requires us to treat some equal interests unequally. As I reject theories with this feature, I reject desert. Notably, nowhere in this argument does free will come up. It is possible that, on my account, free will would be necessary for desert. I make no comment on this matter because I don’t find any account of desert compelling regardless, though it does seem like many people do wish to ground these concepts in some way on free will 1. I only wish to suggest that, on my account, free will is not sufficient for desert. So at the end of it, all of these related debates in philosophy feel like they are talking past me.
If the standard pop-philosophical question is something like, “if everything we do is predetermined, then how can we be punished or rewarded for our actions?” then I depart from pretty much every assumed premise in this statement.
If , help us write more by donating to our Patreon.