In an essay titled "How to Think about the Problem of Free Will", Peter van Inwagen writes:

‘free will’, ‘incompatibilist free will’, ‘compatibilist free will’, and ‘libertarian free will’ are four names for one and the same thing. If this thing is a property, they are four names for the property is on some occasions able to do otherwise. If this thing is a power or ability, they are four names for the power or ability to do otherwise than one in fact does. All the compatibilists I know of believe in free will. Many incompatibilists (just exactly the libertarians: that’s how ‘libertarian’ is defined) believe in free will. And it’s one and the same thing they believe in. Compatibilists say that the existence of this thing (whose conceptual identity is determined by the meaning of the English word ‘able’, or of some more or less equivalent word or phrase in some other language) is compatible with determinism; incompatibilists say that the existence of this thing is incompatible with determinism.

Reading this part surprised me at first. Is what Inwagen claiming here true and uncontroversial? Do 'compatibilist free will' and 'libertarian free will' by default mean one and the same thing? Don't incompatibilists routinely reproach compatibilists like Dennett for redefining 'free will' or for watering it down?

Also, Inwagen claims that 'free will' is a philosopher's term of art and is hardly used outside of a philosophical context. Is this how the majority of other philosophers see it? I'm not so sure myself. Dennett, for example, sees free will as a central folk concept belonging to our manifest image.

  • 1
    To some extent yes, to some extent no. I am not a compatibilist, but as I understand the matter, the main thing is that we genericize the meaning of the phrase "free will" to "the capacity required for moral responsibility," and then the strength of compatibilism is in those versions which describe such a capacity, irrespective of whether that capacity involves a "robust" ability-to-do-otherwise. Of course, then, we can go on to ask about that ability on its own terms, or we might define "moral responsibility" to retain a need for indeterministic free will, etc. Oct 10, 2023 at 16:45
  • @KristianBerry "Moral responsibility" is tangential to free will, not part of the definition. Plenty of free will deniers believe people still have a capacity for moral responsibility.
    – causative
    Oct 10, 2023 at 17:09
  • Yes, Inwagen's usage is strange here. Compatibilism defines free will in a different way from how libertarian free will defines it. There are materialist compatibilists, and materialist free will deniers. They do not disagree over the material circumstances of the brain and mind, they only disagree over how best to describe those circumstances. Therefore, they must be using different definitions.
    – causative
    Oct 10, 2023 at 17:13
  • @causative granted, again to some extent. But when it comes to debates about the conceptual analysis of "free will," modulo the desire to justify attributing this will to us and in light of compatibilist options, I think the relevant genericization of the semantics does sometime occur (or that is my reading of Strawson's "reactive attitudes" or Frankfurt's "higher-order meshes"). Oct 10, 2023 at 17:46
  • 2
    What van Inwagen says is very controversial and he explicitly frames it this way:"I am presenting a paper that consists largely of advice", "having set out a philosophical problem, and having, tendentiously, identified this problem with the problem of free will", "definitions everyone should use - or so I say", etc. What he means by "this thing" is some objective referent that all the participants of the debate aim to pick out, but do so in flawed ways that he wishes to correct. But many of those participants will reject his advice, his formulation of the problem and his definitions.
    – Conifold
    Oct 10, 2023 at 19:36

3 Answers 3


Van Inwagen's definitions appear to be self-contradictory. If one's actions are determined, then basically by definition one "cannot do otherwise". In determinism, there is a causal network which determines one's actions, hence there IS no ability to do otherwise.

Most compatibilists use a different definition, where free will is something like "the ability to do what one wishes, without being externally constrained". These sorts of compatibilists often also make van Inwagen's argument that what they mean by free will is exactly what libertarians mean by free will, but they do so with a different definition than van Inwagen.

The argument that compatibilists are redefining the word to mean something different from common usage, is a common response from both libertarians and incompatibilism determinists.

The most common counter from the compatibilists is that libertarians cannot define a libertarian version of free will, that libertarian free will is incoherent -- and the only coherent concept of free will -- compatibilist "free will" -- is what is actually meant in standard English usage.

In my experience reading compatibilists, Van Inwagen's claim is not unusual, just his definition.

  • "that libertarian free will is incoherent" - that's the argument from incompatibilism determinists, who argue that we should just reject the idea altogether, and not add legitimacy to libertarian free will by accepting some other definition of free will.
    – NotThatGuy
    Oct 11, 2023 at 8:08
  • @NotThatGuy -- yes this argument is commonly used against libertarian free will. Compatibilists tend to take the argument in a different direction than incompatibalist determinists do, and most compatibilists I have read assert their definition is what regular people mean by free will. Van Inwagen's argument is therefore common, but van Inwagen's definition is not.
    – Dcleve
    Oct 11, 2023 at 17:15

Insofar as conceptual analysis/engineering in general is concerned, most or maybe even all strong attempts at such would involve some redefining of terms. If this were not needed to be so, then whence would there be the need for the reformulation of abstract notions in the first place? If the pre-theoretic definitions of concepts, such as they are, were satisfactory, what would there be to analyze, much less engineer, anew?

So the main issue should be how much of a gap there is between proposed definitions. If someone redefined free will as free choice as free action and then claimed that a being is free if and only if it can fly and swim as well as crawl-or-walk, or worse that freedom belongs only to those with inherent teleporters, that would probably be too drastic of a redefinition-of-terms. Or if we claimed (quasi-Stoically) that true freedom belongs only to those who can defy and then suppress every desire or passively-undergone influence, that might seem to be too much to ask for as well (though c.f. the perhaps-paradoxical desire-for-no-desiresD of some branches of the bodhidharma).

By contrast, then, the gap between compatibilism and libertarianism is not necessarily too wide. Yes, sometimes the compatibilism gloss of the ability-to-do-otherwise seems overly weak on account of the "if one had chosen otherwise" clause (which also might seem to miss the point of that word "chosen," here). Yet for the sake of linguistic charity and public reason, as well as caution in the condemnation of others for their concrete behavior, there are moral grounds for at least entertaining the possibility that compatibilist alternative possibilities are genuinely alternative enough to map to what many people do pre-theoretically mean by "free will" in such a context.

The upshot is that the pre-theoretic discursive community of users of phrases like "free will," "free choice," etc. need not be entirely internally consistent, either. Some Christians, say, might be monergists who assimilate "proper" free will to a highly compatibilist framework, whereas synergists might oscillate between a weaker compatibilism or a weaker libertarianism, and then Pelagians should prefer the highly libertarian standpoint (let us suppose). It's all up in the air, and will be until we find a vantage to look upon the problem from that makes an important enough difference, ultimately (whatever "important enough" should mean, eventually).

DThough note that one might reframe the problem as one of first-order vs. higher-order desires: the successful arahant is the one who has overcome all lower-order impulses, even if for the sake of a higher-order intention to escape samsara, and it is not a weakness to be given to intend that escape.

  • i know next to nothing about the debate, so may i ask whether the denial of free will - whatever it means - always goes hand in hand with denying moral responsibility? fwiw, i'm not sure what the latter would amount to, a license to do whatever we please while treating others like harmful or harmless automata? i don't know why that appeals, and guess naively that all that is necessary for moral responsibility is qualia, not the ability to choose differently (but to feel differently). that may be pragmatic and inane, but why does ethics depends on the will?
    – user67675
    Oct 13, 2023 at 6:18
  • 1
    @prof_post normativity is usually taken to be at least weakly action-guiding. If taken to be strongly so, via ought-implies-can, strong free will seems to follow. Bernard Williams appealed to moral dilemmas to undermine such a thesis, but so a neither-weak-nor-strong option is to assume that sometimes ought implies can, and go from there. Oct 13, 2023 at 10:10

Before redefining free will we must define it first. The most common definition is the ability to choose one's own actions.

In determinism all events are completely determined by the previous event. That means that there are no events that are determined by an agent's choice. No agent causation, no choices, no alternatives to choose from.

Free will by most definitions is about making choices and that is completely excluded from determinism. Free will and determinism are totally incompatible.

Compatibilism has to redefine both to make them compatible.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .