Blog Coordinator

« Agents, Mechanisms, and Self-Deception | Main | Featured Author: Pereboom »



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

Hi Joe,

Thanks for the interesting post!

I think it is always important to get clear on whether we are talking about general abilities or powers in some more abstract sense or particularized, more specific abilities or powers. I think scientific reproducibility pertains more clearly to the former than the latter.

Also, suppose scientist S does experiment E with results R at T1. Now we are at T2. Even if we have a particularized power in mind, the question is simply: Can S at T2 reproduce R? That is, it isn't a question of "could have done otherwise", and thus one doesn't get into the standard sorts of problems for (say) a libertarian conception of ability/freedom/power.

Of course, if the scientist *does not* reproduce R at T2, we might wonder whether she could have, and that gets us closer to the worries you raise.

I do agree that on a libertarian interpretation of the particularized "can", and on the assumption of causal determinism, ordinary claims we would typically want to make go out the window. Theoretically, then, we have choices (ha--or at least we have to make choices...)

How about this argument, given what you've said, John:

1/ There is a sense of "able" (or contexts in which ability claims are uttered) related to scientific reproducibility.
2/ That sense can't be the libertarian understanding of "able" otherwise we get unintuitive results (if determinism is true, no scientific results are reproducible unless they are in fact reproduced).
3/ Thus, there is a perfectly respectable, non-libertarian sense of "able."

Then I ask: Why can't THAT be what we're talking about when we're talking about the ability to do otherwise -- the sense in which scientific experiments are "able" to be reproduced (even though they won't be)?

Lastly, notice that if we take your approach, John, there are further problems. You'd have to say (given you are an incompatibilist about the ability to do otherwise) that REALLY scientific results that aren't in fact reproduced are NOT reproducible other than in some loose, abstract sense. That seems unintuitive.

In other words, my claim is that the sense of "able" related to the reproducibility of scientific experiments needs to be robust, if we are going to support our intuitions about the scientific method. Here is a context where a compatibilist view of the ability to do otherwise works much more favorably than an incompatibilist (or libertarian) view.

What exactly is it about a moral context, or a context in which we're talking about "free will," that is so different?


Thanks for pushing on this--and I like your argument.

I *do* think that "there is a perfectly respectable, non-libertarian sense of 'able'". That is, I certainly agree that there are compatibilists construals of even the particularized "able", and they are indeed perfectly respectable. Given the list of proponents of compatbilism among contemporary philosophers, including Keith Lehrer, John Perry, David Lewis, and Terry Horgan, it certainly would be bizarre if they weren't onto something plausible. And Semicompatiblism is officially agnostic about the relationship between causal determinism and "could have done otherwise".

Granted, I do find it overall more plausible to suppose that causal determinism rules out freedom to do otherwise (in the relevant sense), but nothing really hangs on this for me--or nothing hangs on it w.r.t. moral responsibility.

Right, if incompatibilism about causal determinism and the relevant notion of freedom to do otherwise obtains, then, if causal determinism holds, then no scientists who does not reproduce data R could have done so. But I think we could construe the relevant notion of "reproducibility" more broadly to mean something like: it is compatible with the laws of nature that the results be reproduced, and in a suitable number of "replays", holding fixed the relevant features of the experiment, the results would indeed be reproduced.

Of course, this is vague and no doubt problematic in lots of ways. The basic idea I would push is that the notion of reproducibility here is not to be analyzed in terms of what a particular scientist could (in the free will sense) have done. Again: I have no doubt that this is just a gesture, but does it point in the right direction?

I like everything you say above except the last few comments: "The basic idea I would push is that the notion of reproducibility here is not to be analyzed in terms of what a particular scientist could (in the free will sense) have done. Again: I have no doubt that this is just a gesture, but does it point in the right direction?"

I need to think more clearly about the matter but my current view is that there is no particular/general ability distinction; all abilities are in some sense general. Particular ABILITIES are a myth. It is confusing because what matters in MR are particularized claims: S does a at T. We wish to know whether S could have done otherwise. My view is that we can answer this question by providing a general account of ability. It is by virtue of having a spectrum of general abilities that I am able to do otherwise in the relevant sense.

No doubt this is confusing but it might also be beside the point. What I'm interested in doing, research-wise, is replying to arguments for (classical) incompatibilism (the Consequence Argument, the Basic Argument, etc.), and in those contexts the relevant phrase is not particularized. It is usually something general like "no one is or ever was able to do otherwise."

It still seems unintuitive to say that in some strict, or fundamental, or important (perhaps even particular) sense, NO ONE is able to do otherwise if determinism is true. The example of the repeatability of scientific tests illustrates this, I think.

But likely we'll go back and forth forever on this issue! That is why I want to make a further point: that eventually the incompatibilist about the ability to do otherwise is going to have to give some kind of robust account of compatibilist ability to respond to queries like the one about the repeatability of scientific experiments.

To speak loosely, you are going to have to have a (classical) compatibilist account of ability in your back pocket (as you do). Why not just move it to the front pocket?

Thanks, again, Joe.

Well, consider the notion expressed by "comprehnsible". Let's say a certain sentence is comprehensible. It doesn't follow that any given individual will actaully comprehend it on a particular occasion. Now suppose causal determinism is true and a particular individual does not comprehend it. It is tricky here because it is not clear that acts of comprehension are actions in the sense in which we are interested in actions--but give me a break and concede this. (You are a nice guy, Joe.)
The on incompatibilist assumptions it follows that that individual could not (in the sense at issue) have comprehended it. But surely it would not follow, even for an incompatibilist, that the sentence is incomprehensible. "Comprehensible" picks out a general, more abstract kind of possibility here--not the free will can. Or at least this is arguably the case.

So, yes, even an incompatibilist will need to invoke broader, more abstract notions of possibility. Why not just use some such notion--say one that is compatibilist-friendly, that abstracts from particulars about the past or laws of nature--to analyze free will possibility?

Again: that is a perfectly respectable approach, one adopted by a host of super-smart philosophers, and one to the rejection of which nothing in Semicompatibilism commits me.

But I do think that my freedom in the free will sense is the power to add to the given past, holding the laws of nature fixed. I think this is strongly endorsed by common sense, and I think one gets into trouble denying it. (For this basic idea, I'm indebted to my PhD dissertation supervisor, Carl Ginet. In a recent article forthcoming in Phil. Studies, Garrett Pendergraft and I point to some difficulties with denying this commonsense view: "Does the Consequence Argument Beg the Question?", Phil. Studies, online first publication).

I see your point Joe, and endorse it: science seems interested in reproducible results, and the ability of science to reproduce results entails that those results are reproducible as types in tokened variations of the typed-conditions of experiment (and thus any abilities disclosed in nature are generalized as well). But libertarians need tokened (in-)abilities relative to a specific spacetime point (of decision-making relative to laws and the past)), and yet they then generalize that to "no one is ever able to do otherwise." But aside from these libertarian specialized and tokened senses of ability and inability relative to specific circumstances of where the laws and the past are fixed, the language of abilities (and inabilities) appears to be about general claims, and the libertarian seems to help herself to that typical usage but without grounds to do so.

BTW, my reading of van Inwagen's fix of Beta in his "Mystery" article attempts to avoid this whole controversy by invoking a sense of Beta that only refers to an actual or at least modally-centered deterministic world. Therefore any resultant (in-)ability claim is tied to only one possible situation, and inherently particularizes any (in-)ability claim about (not) having a choice. That is a consistent fix, but only in a petitio fashion. Logically consistent moats can't legitimize the islands thus isolated if they have nothing to do with the larger topography of the issue.

You make some great points, Al, and clearly I'm moving in that direction. (Not too surprising since my views have been shaped by over conversations over the last few years.)

To express the thought in my own words (if I have it right), what I dislike about the particular/general ability distinction is that eventually you are going to be led to adopting something like John's extension principle in order to account for particular abilities: the ability in question is the ability to add to the past and the laws. But this principle strikes me as question begging. Since John and Garrett have addressed this charge in print, I'll leave this discussion for another time.


Thanks for keeping this discussion going! I appreciate your comments and your taking the time to give them.

Notice that you're doing a kind of cost analysis, which I like. There are some theoretical benefits for compatibilist accounts of ability, but also some minuses; and the same goes for incompatibilist accounts of ability (all I mean by this is an account that takes something like the extension principle (see my comment on Al's post) as specifying a necessary condition for ability). I think that this approach favors compatibilist accounts of ability, and the examples we've been discussing support my claim. The examples seem to me to provide the basis of a common-sense argument for compatibilist accounts of ability. Of course, you suggest that there is a fundamental difference between terms like “comprehensibility” and “repeatability” – on the one hand – and “ability” in the sense understood in free will contexts (presumably some kind of particular ability).

Let's take your (brilliant) case of "comprehensibility." To restate my point, by your own admission you MUST give a compatibilist account of comprehensibility. But I also think the account must be robust. In other words, the account must ground our intuitions about what it is for something -- a book, say -- to be comprehensible. What it is to say that someone is genuinely able to actually comprehend the book. It would be unintuitive to say, in the end, that NO ONE is REALLY able to comprehend a book -- unless it has been comprehended (provided that determinism is true). The point is not just that comprehensibility is amenable to a compatibilist account, but that an incompatibilist account is precisely the wrong kind of account in this case.

"Comprehensibility" is a very respectable variation of "ability." It is amenable to a compatibilist account of ability. Incompatibilist accounts of ability seem wrongheaded in this instance.

Suppose I say that now it is your turn. You provide for me an example of an ability for which an incompatibilist account is well suited -- again, all I mean by that is an account that adheres to the extension principle. Now you can't provide examples from free will contexts because that would be question begging. Or at any rate, there would likely be disagreement between us about the cases. It would seem that one approach might be to see how the two theories stack up when it comes to our usage of ability terms in other contexts. In that case, a compatibilist account of ability would stand up nicely to any incompatibilist account.

I guess ultimately the question is, What is it about free-will contexts that causes the need to require the extension principle in order to account for the robustness of our abilities? It is not as if adherence to the extension principle increases the robustness of abilities in general. What is particular about the case of free will?


And just to add to Joe's point/question: Are there *any* (other) contexts in which we use the concepts of abilities, opportunities, can, possibilities, etc. and in which anyone (including incompatibilists) would say either (a) that the concept(s) do not properly apply if determinism is true, or (b) that people are mistaken in thinking the concept(s) apply if determinism is true, or (c) that ordinary people understand the concept(s) in such a way that they are committed to determinism being false? (I put "other" in parentheses to indicate that clearly some people think free will/MR is one context in which people say these things, even though I think they shouldn't.)

This reproducibility property seems closely related to the old problem of statistical regularities ("laws") of human behaviour, here applied to the experimenter.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan