Blog Coordinator

« Passing the Torch! | Main | Books About Agency »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Hey Marcus,

Interesting stuff. I have a question about Kant but it might be related to your libertarian compatibilism. You say Kant believes we have "libertarian-type free will," which I suppose suggests that he is a libertarian and not a compatibilist (although maybe this last move is where I make my mistake).

I always thought it was rather simple. Does Kant think that determinism is true? Yes. Does he think we have free will? Yes. Thus, he is a compatibilist. What am I missing?

Hi Joe: Thanks for your comment and questions!

I'll try to keep my answer brief, for several reasons. First, it's easy to get mired in the "weeds" with Kant, trying to figure out exactly what he did and did not think (in order to give a really detailed answer, I'd have to write a prohibitively long comment!). Second, there are probably Kant scholars out there who know Kant here far better than I, and I'd like to stay away from making unnecessary mistakes. Finally, I think the overall gist of Kant's view will suffice here.

Kant's metaphysics and epistemology are really complex--far more complex than "determinism is true, but free will is compatible with determinism."

First, Kant doesn't adopt a monistic worldview (i.e. physicalism). He distinguishes between (1) phenomena, or the world of "appearances" (roughly, the world of physical objects and properties), and (2) noumena, or "things-in-themselves" that exist in some sense but stand behind the phenomena.

Second, and far more confusingly, Kant thinks that we encounter the world from three related perspectives--(A) a theoretical perspective, (B) practical perspective, and (C) aesthetic perspective--and by way of three associated cognitive faculties (theoretical reason, practical reason, and faculty of judgment).

Anyway, long story short, Kant seems to think that:

(1) From a theoretical perspective (i.e. the realm of appearances), determinism obtains.

(2) We can deduce from the theoretical perspective (i.e. the phenomena that appear to us) that there must be real things (noumena) "behind" the deterministic phenomena.

(3) We have *no* evidence from the theoretical perspective that determinism obtains at the level of noumena, since the theoretical perspective only provides us knowledge of the nature of appearances.

(4) In contrast, from a practical perspective, we have grounds for taking ourselves to have libertarian-type free will, or the ability to conform our actions to the moral law.

Thus (5) Although the world of phenomena is deterministic, libertarian-type free will at the level of noumena is compatible with said determinism.

Finally, it's important to note here, with (5), that Kant doesn't think he can *show* that we have libertarian free will theoretically. Rather, his claim is merely that we can, in principle, have no theoretical evidence *against* libertarian freedom at the level of noumena (since we can only theoretically cognize phenomena, not noumena), and that we encounter ourselves as having libertarian freedom from a practical perspective.

As such, Kant was a compatibilist of sorts, but one of an extraordinary type. Ordinary compatibilists don't think we have libertarian free will. They think we have free will even though we don't have libertarian free will. According to them, free will in a deterministic world is something else entirely (e.g. reasons responsiveness, actual sequences, etc.).

Kant, in contrast, is a really bizarre kind of compatibilist. He thinks that *libertarian* free will (at the level of noumena) is compatible with determinism (at the level of phenomena). And of course it is here that Kant gets in all kinds of trouble.

One obvious source of trouble for Kant is that if phenomena are deterministic, and noumena are simply the things in themselves that "stand behind" phenomena, why doesn't it follow that noumena are deterministic as well? (How could we be libertarian-free as noumena when noumena are "the flip-side" of deterministic phenomena?).

Another obvious source of trouble for Kant (the one that Grenberg points to in her paper on the Phenomenological Failure of Groundwork III) is that Kant seems to think that he establishes noumenal freedom through our experience of practical reason (i.e. the manner in which we *seem* to experience ourselves as libertarian-free from a first-personal perspective). As many critics of libertarianism have long noted, this kind of argument seems like super-poor evidence for libertarianism (I would dare say it is no evidence at all). Just because we *seem* to have libertarian free will when we act, that's no reason to think that we *do* have it. After all, our seeming to have it might well just be a determinisic illusion!

Finally, I'll just add that although the view I defend--a view which I call "Libertarian Compatibilism"--shares the basic idea of Kant's (that genuine, libertarian free will in a higher reference-frame is compatible, and indeed generates, the appearance of deterministic causal closure at a physical level), I don't think my view runs into the famous problems for Kant's view that I just summarized (in part because I *don't* base my argument on the phenomena/noumena distinction). But I'll leave more on this for later... :)

Anyway, does all this make sense?

Thanks, Marcus. That is helpful. Still, I think Kant is a compatibilist.

Very roughly, your point is something like this: Determinism is true in the phenomenal realm of appearances but this does not conflict with our having free will in the noumenal realm. In this sense, Kant is a compatibilist. Nonetheless, he thinks that freedom conflicts with determinism within the same realm. Hence, he is an incompatibilist/libertarian, as well. Is that close?

I don’t want to get caught up in a debate about Kant either because I am not a Kant scholar but I do think that you could tell the story another way, one that is more favorable to compatibilism. What I mean is that on your story, Kant thought that determinism did conflict with free will and felt the need to place them in different realms in order to “reconcile” them. But I think you could make the same point Kant makes – and that likely Kant made the point this way – without talking about realms.

I think the main point behind Kant’s compatibilism is that determinism doesn’t rule out the possibility of our being the original causes of our actions. Consider Ned Markosian’s compatibilist theory of agent causation . Each free action has two causal chains: one deterministic chain leading back to the origin of the universe; another leading back to the agent.

Of course, the addition of agent causation is not essential to the story. You could also say that every free action (in theory) has two distinct explanations. First, there is a deterministic explanation of the action in terms of prior events and the laws of nature. Second, there is a practical explanation in terms of the agent’s reasons, wants, and desires. The former grounds determinism, the latter grounds free will. Neither of these stories supposes that determinism conflicts with free will in any way.

Lastly, I have misgivings about the expression “libertarian free will.” Either we have free will or we don’t. If so, our having it is either consistent with determinism or it is not. If it is not, libertarianism is true. But I think it is wrong to believe that this kind of freedom is more robust than the kind you can get on a determinist model. That assumption begs some compatibility problem. It seems to be generally accepted that compatibilists can’t give theories of ultimate sourcehood or agent causation because most of them don’t. But it seems also that Markosian – and maybe Kant – did provide such theories.

But I don’t want to let this disagreement get in the way of your discussion since this could well be a verbal dispute between us. I can see now why you might want to call Kant a “libertarian compatibilist.” I’m sure your view is interesting, even if I might use different words to describe it! Looking forward to the discussion.

Hi Joe: Thanks for the follow-up!

I'm sympathetic with your alternative story, and indeed, develop something like it in detail in my book manuscript (reimagining Kantian ethics).

Still, I think it's important to see why (1) Kant would have a problem with your story, and (2) why your alternative story is not simply a different way of making the same point as Kant.

Kant was really explicit that he was not a compatibilist in any traditional sense. He wrote: "[Compatibilism] is a wretched subterfuge...and...a petty wordjugglery." (1788, 95–96) Why?

A crucial part of the answer has to do with moral philosophy. Kant thought that the Categorical Imperative--i.e. morality--applies to us if and only if we possess the *categorical* capacity to overcome our wants, desires, and inclinations (i.e. any and every antecedently existing motive we might have) and act on principle alone, ex nihilo.

The problem, Kant thought, is that in a deterministic world, the latter part of this biconditional is straightforwardly false. In a deterministic world, our actions are *not* categorically up to us. They partly emanate from us (i.e. Markosian agent causation), but also partly emanate from contingent laws of nature. Even though they come "from us", they are also dictated *for* us by whatever representational (i.e. beliefs) and motivational (i.e. desire) states we deterministically have.

This leaves us with two options:

(1) Deny that the Categorical Imperative is the fundamental principle of morality.

(2) Deny the above biconditional (i.e. provide a compatibilist defense of the Categorical Imperative).

Obviously, many moral philosophers choose (1). In my book, I attempt (2). But it's important to see that (2) is not easy to accomplish. How could the Categorical Imperative normatively apply to us if we lack the capacity to conform our actions to it?

Anyway, more to the point, it's crucial to see that (2) is not the "same" story as the one Kant holds. Kant thought that the Categorical Imperative requires libertarian free will, and was inconsistent with the kind of compatibilism you describe.

Finally, although I appreciate your misgivings about the expression "libertarian free will"--lots of people have them--I don't think they are accurate. As I will try to explain throughout a number of posts this month, I don't think it's true that "either we have free will or we don't." I think there are many possible types of free will--(1) Humean free will, (2) Frankfurtian free will, (3) Kantian free will--and that it's an open physical and metaphysical question which of them we have. Indeed, as I will explain in my next post, I think we may well have more than one of them. As such, I think the question, "Do we have free will?" is poorly framed. The question isn't whether we have it. The question is what *type(s)* of it we have.

(FF) Frankfurtian free will is (a kind of) free will.

Doesn't (FF) entail compatibilism?

Hi Mark: Of course! But the issue Joe and I were discussing is Kant's view, and whether it can be construed as a simple type of compatibilism. I was denying that it can. My (separate) claim then was that, in my view, there are many different possible types of free will, of which FF is one.

I'll say more about this in my next post, but for a primer on how I'm thinking about things, you might want to take a look at Mark Balaguer's 2009 paper, "The Metaphysical Irrelevance of the Compatibilism Debate (and, More Generally, of Conceptual Analysis)" ( ).

Thanks for your reply, Marcus. This is all very interesting. Generally, I find arguments for Kant’s incompatibilism unconvincing. But your argument is compelling and I’m thankful for it. In support of your view there is this passage from CPR:

“Now if one takes the determinations of the existence of things in time to be determinations of the existence of things in themselves (which is the most ordinary way of conceiving [them]), then the necessity in the causal relation can in no way be reconciled with freedom. … For from that necessity it follows that every event, and consequently also every action that takes place at a point of time, is necessary under the condition of what was in the preceding time.” (Pereboom, Free Will, 2nd edition: 121; I don’t have Kant handy!)

Let me give my super-compatibilist interpretation of Kant another go. To be clear, I think that the story I’m telling is an acceptable interpretation of Kant. It is Kant’s story, not just another similar story. I don’t want to get bogged down with Kantian interpretation and as I confessed above I realize that your interpretation is pretty compelling, as well.

To begin, Kant did not write “compatibilism is a wretched subterfuge.” You inserted “compatibilism” within the brackets. It is reasonable to assume that the target in that famous passage was the prevailing compatibilist view of free will, where free action is identified with action that is internally, and not externally, caused. That distinction is too blunt an instrument to ground free will, according to Kant. Free action must be grounded in our power to “act on principle alone, ex nihilo,” as you put it.

Let’s switch for a moment from the compatibility problem to a debate about sourcehood. Kant thought that free will required ultimate (or original) sourcehood. According to Spinoza, only God fulfilled that requirement. Perhaps for these reasons, others (often compatibilists) adopted views of adequate sourcehood – comparable to Spinoza’s blessedness. Contrast, for instance, the views of Pereboom and Strawson versus Fischer and McKenna.

I admit that, for Kant, adequate sourcehood is not enough to ground free will. What I question is whether the debate about sourcehood maps neatly onto the compatibility problem. Markosian’s view shows that one can adopt agent causation – a kind of ultimate sourcehood – and still be a compatibilist. We both agree that, for Kant, only ultimate sourcehood will do. The issue is whether Kant also thought that determinism in principle conflicts with ultimate sourcehood.

You write: “The problem, Kant thought, is that in a deterministic world … our actions are *not* categorically up to us. They partly emanate from us (i.e. Markosian agent causation), but also partly emanate from contingent laws of nature. Even though they come ‘from us’, they are also dictated *for* us by whatever representational (i.e. beliefs) and motivational (i.e. desire) states we deterministically have.”

Again, this is not so much a criticism of compatibilism as it is a criticism of a more general view of free will. Markosian’s theory of agent causation is influenced by Randy Clarke’s libertarian theory of agent causation. Clarke questioned the Chisholm agent causation model, where free actions are mysterious and inexplicable. Why can’t a free act have causal sources in both agents and events, free actions imbedded in an overarching causal nexus? On Clarke’s view – which is a libertarian view – our acts emanate from us but contingent events play a causal role, as well. They are neither mysterious nor inexplicable.

Is your claim that an action is free only if it emanates entirely from the agent, so it is uninfluenced by both worldly events and the laws of nature? Do you reject Clarke’s view? Doesn’t that lead us back to the inexplicable, mysterious free actions of Chisholm? (I'm a Chisholm fan, so I'm not suggesting that this would be entirely bad.)

Lastly, would it be correct to classify you as a pluralist – “free will” has more than one meaning – rather than a monist?

Thanks again! Looking forward to reading more.

Hi Joe: Thanks for your reply! As you say, the Kant question comes to this: "The issue is whether Kant also thought that determinism in principle conflicts with ultimate sourcehood."

I think the answer to this is that Kant *does* think determinism conflicts with ultimate sourcehood, at least in every sense that mattered to him (viz. moral responsibility and the Categorical Imperative). Allow me to explain.

For Kant, the Categorical Imperative applies to us absolutely unconditionally, or necessarily. Kant recognizes that this raises a profound puzzle. How can *anything* apply to us completely unconditionally? Everything in nature, after all, is contingent, including our beliefs, desires, the laws of nature, etc.

It is for this simple reason that Kant thought free will--in the sense of our being morally responsible for our actions--cannot in *any* sense be located within a deterministic world. For Kant, the kind of freedom necessary for moral responsibility (viz. the Categorical Imperative) is, fundamentally, freedom to act unconditionally, i.e. independently of the laws of nature.

This is why Kant thought he had to locate free will in the noumenal realm. Kant thought:

(1) Everything in the phenomenal realm is *conditional* upon the laws of nature.

(2) Morality is absolutely unconditional.

Thus (3) The normative source of morality (i.e. free will) must exist outside of the deterministic phenomenal realm, as a kind of unconditional, libertarian capacity to will actions from *nothing* aside from pure principle.

Now, I actually think there's a way around this argument--and I pursue it in my book manuscript. But, as far as Kant is concerned, I don't think there's any way to rescue him (within his way of thinking) from this (incompatibilist) result.

Next, you ask: "Is your claim that an action is free only if it emanates entirely from the agent, so it is uninfluenced by both worldly events and the laws of nature? Do you reject Clarke’s view? Doesn’t that lead us back to the inexplicable, mysterious free actions of Chisholm? (I'm a Chisholm fan, so I'm not suggesting that this would be entirely bad.)"

My reply: Yes, I think it does lead us back to inexplicable, mysterious free actions of Chisholm--and I think there is nothing wrong with this. As I will explain in some future posts, I think *physics* invokes inexplicable, mysterious actions (at the level of particles, forces, etc.), and so adding one more inexplicable, mysterious force--brute, inexplicable libertarian free will--is no less coherent (and, as I will argue, a part of a full metaphysical picture that explains stuff that extant forms of naturalism don't!). In other words, I'm right with you. I'm a Chisholm fan here, and think we all should be. :)

Finally, yes, I think it is right to classify me as a pluralist. I think there are many permissible construals of "free will" (I think the term is systematically ambiguous), and that there are many different types of free will, of which libertarianism is the coolest and most desirable type. :)

Well this exchange has me primed!

If pluralism about FW is true, then something other than pure reference to objective states of affairs is involved. Of all relative factors, I've suggested that values might be the likely culprit based on my Veatch-death-definition examples. (Other than FW, what else is more important than death? Though I would point out that pluralism about death is probably easier to advance than pluralism about the beginning of the moral status of human life for obvious reasons. . .) But then isn't the issue what value is most basic to FW, or why (as Veatch argues for death) we should tolerate a pluralism of values as a matter of public policy? Though I doubt as a matter of public policy or philosophical debate many would advocate a pluralism of FWs as significant of how to deal with MR.

V. Alan White: Thanks for your comment, and I'm glad to hear that you're primed! :)

Anyway, yes, I think that'a right. If pluralism is right, then something other than "pure reference to objective states of affairs" is involved. As I explain in my new post (just posted now!), I think our concepts of free will and moral responsibility are deeply influenced by context. Very roughly, I think we have incompatibilist concepts that involve ultimate metaphysical concerns, but also compatibilist concepts that are triggered by ordinary-everyday concerns. Both concepts, as such, are not "purely objective." They are triggered by different *concerns* of ours.

I am also right with you on your point about public policy. Although I don't mention it explicitly in my new post, I think our compatibilist concepts are most directly relevant to ordinary life and public policy, and our incompatibilist concepts bear only a very distant (but not entirely non-existent) relation to public policy.

To get a flavor as to why I think this, I think the "ultimate" question of whether we have libertarian free will is almost (almost!) entirely irrelevant to public policy. Public policy should aim at the public good (at protecting people from crime, etc.). Since libertarian free will is (mostly) irrelevant to this, it follows that whether we have free will at an ultimate metaphysical level--or whether libertarian free will is incompatible with determinism--is almost entirely irrelevant to public policy. (I say "almost" because, if incompatibilism is true at an ultimate metaphysical level, perhaps this disables retributivist arguments for punishment).

What *is* relevant to public policy, on the other hand, are compatibilist questions: whether a person's actions are reasons-responsive, for instance. We don't punish the mentally insane, but rather provide them treatment, because we recognize that their actions did not satisfy certain criteria for legal accountability--criteria that make sense relative to ordinary-everyday life.

Thanks again for your comment!

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan