Defining Free Will
Defining Free Will
An Example of How to Sidestep Merely Verbal Disputes
I’d guess that many people have some intuitive idea of whether they believe in free will. Another hunch of mine is that many of those same people would struggle with formulating a tight, coherent, and defensible definition of free will despite having an opinion on its existence—it’s a fairly nebulous concept for them. So, here I want to outline some ideas and strategies philosophers use for concretizing that notion a bit without getting totally sidetracked into logomachy (i.e., what’s known as “merely verbal disputes”).
The Ability to Do Otherwise
One major conceptualization of free will is the ability to do otherwise. If S did X at t1, free will (in this sense) means that there has to be some meaningful way in which S could have done Y at t1 instead. The tension with something like determinism is manifest: if, say, some supreme being or supercomputer could predict with certainty at t0 that S would X at t1, how could it be that S would instead Y at t1? Moreover, if S couldn’t do otherwise, how could S be blameworthy for X-ing? Compatibilists (people who believe that free will and determinism are simpatico) proposed a clever solution.1
Classical Compatibilist Theory of Free Will: S has free will and could have done otherwise when it’s the case that, if S had wanted to do something else, S could have done something else.
This makes intuitive sense: it certainly seems like people who are able to do what they want have significantly more agency than people who can’t, and it’s tempting to distinguish the former as someone acting under their own free will. In a limited sense, I think Classical Compatibilism actually succeeds: it shows how even if some particular event is determined, someone can possibly retain agency and blameworthiness in an important way. The obvious defect in Classical Compatibilism, however, is that determinism isn’t causally localized to a particular event—it infects all the pertinent surrounding causality, too. Thus, for example, who gives a shit if you wanted to do otherwise when what you wanted to do was also predetermined?
The free will literature underwent a sea change in the 1960s, partly because of an influential paper by Harry Frankfurt.2 Above, I mentioned the puzzle of how it seems unfair to blame someone for something if they couldn’t have done otherwise. This more explicit version of this argument is often something like this:
If determinism is true, then no one has the ability to do otherwise.
If someone could not have done otherwise, then they cannot be blameworthy for what they did.
Therefore, no one has moral responsibility.
Variations of (2) are known as the Principle of Alternate Possibilities. Classical Compatibilism, discussed above, tried to reconcile determinism and free will by disproving (1); Frankfurt-inspired Non-classical Compatibilism would try to disprove (2).3 Thinking about this from a different angle, Non-classical Compatibilism would focus on salvaging a version of free will other than the ability to do otherwise. How does Frankfurt attempt to disprove (2)? Here’s an example:
Frankfurt-style Counterexample to PAP. Suppose Alice—who lives in a world where she has free will and generally has the ability to do otherwise—wants to kill Bob. Charlie, an evil neuroscientist, also wants Bob dead but knows that Alice is already planning to assassinate him. To be sure Alice follows through, Charlie implants a device in Alice’s brain that would override any wavering by Alice and compel her to assassinate Bob. The next day, Alice kills Bob without wavering and the device never interferes.4
In examples like this, Alice couldn’t have done otherwise (because of Charlie’s device), but this doesn’t seem to let Alice off the hook: Charlie’s device was irrelevant to what transpired. The fact that she couldn’t do otherwise wasn’t the reason why she killed Bob. Here’s what Frankfurt writes:
“There may be circumstances that constitute sufficient conditions for a certain action to be performed by someone and that therefore make it impossible for the person to do otherwise, but that do not actually impel the person to act or in any way produce his action.”5
Alice would have killed Bob even without Charlie’s device and be blameworthy, and the device didn’t interfere at all, so why would that device’s presence suddenly relieve Alice of being blameworthy? Now, I can’t even begin to convey the voluminous literature devoted to arguing whether these Frankfurtian counterexamples are legit.6 My personal view is that they’re bunk, but that’s likely a minority view.
Backing into a Definition
So, what the hell is free will if not the ability to do otherwise? Well, even if moral responsibility doesn’t require the ability to do otherwise, it still requires some kind of agency. Even Frankfurt admitted that “[i]t is generally agreed that a person who has been coerced to do something did not do it freely and is not morally responsible for having done it.”7 Thus, some philosophers have devised a clever way to back into a working definition of free will by thinking about it as whatever the agency is that’s required for moral responsibility,8 and many papers mainly focus on moral responsibility rather than free will.9
Of course, some people retain the belief that the autonomy required for moral responsibility really is the ability to do otherwise, so maybe it’s all the same in the end. Still, this approach can be pretty useful for philosophizing: much of what might otherwise be a purely definitional controversy can be linked up to a more substantive disagreement about what exactly is necessary for people to be blameworthy/praiseworthy.
Unfortunately, this utility has some costs, because all the inside baseball can be misleading. Sure, compatibilists believe in free will, but the compatibilists’ idea of free will is a kind that’s accessible even when your every move is predetermined, and modern compatibilists often (mainly?) believe that free will and determinism can co-exist only because they think that free will is something that doesn’t include the ability to do otherwise. Crucially, this is all pretty opaque to outsiders, and maybe non-specialists would view this as a surprisingly modest conception of free will. Thus, whenever you hear something like that most philosophers believe in free will (because most of them are compatibilists), take it with a grain of salt.
The Classical Compatibilist landscape was somewhat more complicated than what I present here; for a detailed overview, consult Chapter 3 in McKenna, Michael & Pereboom, Derk. Free Will: A Contemporary Introduction, Routledge (2016).
Id. at 102.
See id. at 147 (they use slightly different terms and phrasing); See also Kadri Vihvelin, “Foreknowledge, Frankfurt, and Ability to do Otherwise: A Reply to Fischer,” Canadian Journal of Philosophy, 38 (2008): 348.
This is pretty generic, but there are copious examples out there and clever modifications to rebut counterarguments, etc. You can find more of this stuff here.
Harry Frankfurt, “Alternate Possibilities and Moral Responsibility,” Journal of Philosophy 66 (1969): 830.
See Chapter 5 of Mckenna and Pereboom, supra note 1, for a whirlwind tour of the tangled literature.
Frankfurt, supra note 5, at 830.
See, e.g., McKenna and Pereboom, supra note 1, also use a definition like this. After writing much of this I found Kevin Timpe conveying some similar thoughts (even referencing the Chalmers article on verbal disputes) while undertaking a somewhat different project of distinguishing between leeway and source conceptions of free will.