Blog Coordinator

« Libertarianism and (Non) Evidential Assessment | Main | Gardeners—I mean Flickerers—in the NYTimes on Experimental Phil and FW »



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

Every Xphi experiment on Free Will I have seen, including this one, uses the same protocol. Subjects are asked if you can be morally responsible in a universe in which:

“Every decision is completely caused by what happened before the decision—given the past, each decision has to happen the way that it does.”

But, as Hume taught us, a world where everything is caused by what went before is not the same thing as a world where each decision "has to happen the way that it does". So what are subjects being asked to imagine"

1) A universe in which any future state of the world can be inferred from the laws and a complete description of world at an earlier state. (That is, a deterministic universe in the standard philosophical sense)?

2) A universe in which there is a metaphysically necessary connection between cause and effect?

3) A universe in which Logical Fatalism is true?

And which, if any, of these is meant to be described by a world in which "every decision is completely caused by what happened before the decision" ?

What does "completely" caused mean? What would an "incompletely" caused decision be like?

Suppose I were a libertarian who held that every decision is caused by prior uncaused choice, deliberation or act of will. In that case I would believe that every decision was caused (completely!) by what went before and that every decision "had to happen" given that past. But in that case my belief that freedom was possible in the world as described would not reflect my views about free will and determinism, would it?

When I have raised such worries with experimental philosophy proponents in the past I have been told that the "naïve subjects" who answer these questions are too stupid to be sensitive to any of these highfalutin metaphysical nuances. I am usually told this right after I am told how useful and important it is to find out what they think about these metaphysical issues.

If you ask a stupid question you get stupid answers. If you ask a muddled question you get a muddle. None of this experimental work shows what the subjects think about determinism, free will or anything else because it is unclear what question the experimenters were asking or what question the respondents took themselves to be answering .

Once again (cf. the only question raised by experimental philosophy is which is worse: its philosophy or its experiments.

"When I have raised such worries with experimental philosophy proponents in the past I have been told that the "naïve subjects" who answer these questions are too stupid to be sensitive to any of these highfalutin metaphysical nuances. I am usually told this right after I am told how useful and important it is to find out what they think about these metaphysical issues."

That's awful, disgraceful--no wonder you have such a low opinion of experimental philosophy. Which experimental philosophy proponents have said this to you? I want to talk to them, show them how inconsistent they're being. Obviously you must have some documentation of this, right? (Especially since you're using quotation marks.) They told you this in an email, or a blog post, or something like that, right? Something? When did this happen?

Well, for one example, you could look at Justin Coates here:

Don't you remember, Tamler? You participated in that discussion?

I do remember that discussion. But I'm struggling to see how you regard it as an example of what you described.

Again, you wrote:

"When I have raised such worries with experimental philosophy proponents in the past I have been told that the "naïve subjects" who answer these questions are too stupid to be sensitive to any of these highfalutin metaphysical nuances. I am usually told this right after I am told how useful and important it is to find out what they think about these metaphysical issues."

In your example, it's JUSTIN who raises the objection and you who agrees with him. I hate to nit-pick but that doesn't seem precisely like what you were describing.

It might be more productive to turn from history to philosophy.

Quite a few people have pointed out that if the descriptions of universes employed in many of these studies are meant to convey the information that these universes are deterministic, then the descriptions are poorly crafted. If our students offered these descriptions, we'd likely correct them. Why not use better ones in the studies?

The definition of determinism that we most commonly use in discussions of free will is itself open to an interesting challenge, one that Scott Sehon raises in a recent Analysis paper. Any interest in discussing that?

I'd like to see this blog used to make some progress.


When I raised my concerns that the Xphi Free will scenarios were confused and confusing, Justin defended them saying:

"That said, I definitely appreciate the difficulty in crafting scenarios that are (1) accessible to the folk and (2) capture the philosophical theses in question. Moreover, there are additional methodological constraints on researchers imposed by the IRB; one such constraint is that all questions must be on a rather low reading level (I can't remember what grade level specifically, but it's quite low). In recognizing that these scenarios must be written on low level, I think we can move past some concerns that they are ungrammatical or wouldn't pass muster in a college classroom (of course, they seem a great deal more sophisticated than many of the papers I've graded). Experimental philosophers must give scenarios that are standards appropriate... "

I would say that anyone who does not read this as showing a high level of disdain for the intelligence of the "folk" is… well, reading on a rather low level.

But it does seem to me that your defense of Xphi is avoiding my more serious challenge. The important issue is not whether Xphi philosophers harbor condescending attitudes towards "the folk" but whether their experiments are any good.

Let me make the point again. If one is going to present scenarios to subjects and ask them "what do you think?" the scenarios should be as unambiguous as possible. An obvious methodological upshot of this requirement is that the experimenter should never ask a question of the form:

"What do you think about a Scenario where X, in other words, where Y."

The obvious problem is that even if the experimenter thinks that X and Y describe the same scenario there is no guarantee that the subjects do and so their responses may be skewed in any number of directions.

But this is precisely what the Nichols protocol does:

“Every decision is completely caused by what happened before the decision—given the past, each decision has to happen the way that it does. "

Do the subjects understand what is on the left of the dash as equivalent to what is on the right? If they don't are they responding to the first scenario, the second, or some hybrid of these different interpretations?

And this is not an abstract worry. Anyone who has ever taught an intro class on this subject knows how difficult it is to get students to disentangle hard-determinism from logical fatalism. This form of question positively invites this confusion.

This, I submit, is bad science.

Even assuming that the folk understand the scenarios and definitions, what good philosophical purpose would their intuitions serve besides commencing OUR own philosophizing re. FW? (And, for what it's worth, I always seem to have at least a handful of students who do grasp the issues, especially PAP and Determinism.) If there isn't one, then why shouldn't I just consult my own intuitions? After all, besides being a semi-renowned philosopher, I'm a member of the folk too: I follow baseball, take care of my lawn, and get angry over gas prices just as well as the next guy. At risk of stirring up my erstwhile adversaries at the GFP, I must again say that I can't get past the notion, implicit in Socrates’ elenchi, made explicit by Aristotle, and reiterated at the GFP and the Leiter Report by David Velleman, that the folk’s intuitions may point us in the direction of the truth, but are not philosophically sacrosanct.


I agree that it's extremely difficult to craft a good description of determinism for a study--and there a host of legitimate objections one could raise against the ones found in the literature. Not only that, as I wrote in my Compass piece, I don't think X-Phi studies on free will should be describing determinism at all, since the premises that appeal to intuitions in incompatibilist arguments do not contain such descriptions. I was just objecting to your strawman characterizations.


For that reason, I don't think it's more productive to hone the definition of determinism for future studies. I think that would be a step backward, further entrenching the current framework which I think is fundamentally flawed.

No incompatibilist argument that I'm aware of is of the form:

1. Determinism is true.
2. If determinism is true, then there is no such thing as free will and moral responsibility.
3. So, there is no such thing as free will and moral responsibility.

Rather, the arguments contain premises with incompatibilist principles (like TNR, or Rule B, or Strawson's 'if you're going to be MR for what you do, you have to be MR for the way you are') that, when combined with the truth of determinism lead to the incompatibilist conclusion. Either that, or the arguments will appeal to intuitions about cases, but again, cases that don't involve a general description of determinism. In all of these arguments, the premises that appeal to intuition do not involve descriptions of determinism, so there is no need to describe determinism in the studies. My worry is that the current framework has become "industry standard" and so future researchers will feel compelled to follow suit.

Tamler, I don't quite see what you're saying. In the argument you presented, premise 2 is a statement of incompatibilism. (Or, rather, 2 with a necessity operator is incompatibilism.) Arguments for that thesis are typically prefaced with a definition of determinism -- the authors aim to tell us exactly what they're arguing about. The typical premises are the fixity of the past and the fixity of the laws.

G. Strawson's argument, of course, isn't about determinism; it's an impossibility argument. So no need to define determinism there. But if we ARE offered an argument for incompatibilism, wouldn't we want the author to tell us what she takes determinism to be?

Is it your view that impossibilism, and not incompatibilism, is the thesis that's interesting? I find them both interesting.

Two replies in one comment.

Good idea. Why does that author say that determinism as standardly defined makes God's existence "logically impossible"? As standardly defined determinism is a contingent thesis and is consistent with there being worlds in which God exists. Is his point that the truth of determinism as standardly defined is incompatible with the existence of a God with certain powers? If so, why doesn't he put it that way? And once we sort out what might even possible be true in the point, is the point new? (see, eg, Jim Stone's Phil Studies paper "Free will as a gift from God" as I recall the title).

If one wants to know what subjects think are and are not consequences of determinism, I still don't see how we don't need to provide a prompt that accurately reflects determinism.
You say that in arguments for incompatibilism "the premises that appeal to intuition do not involve descriptions of determinism". But what is it for a *premise* to appeal to intuition?

Further, let's take one such argument for an example. Van Inwagen's normal formulation of the consequence argument features determinism as an assumption for conditional proof, 2 premises, and 2 proposed rules of inference beyond standard logical principles. Are some or all of these "incompatilist principles"? Which ones? Just rule Beta? More than that? Not every incompatiblist accepts all of van Inwagen's premises and proposed rules of inference and not everyone who accepts all of them (and certainly not any one in particular) is an incompatiblist.

But back to there being no need to give a prompt that correctly captures the thesis of determinism. Assume that it's correct that one needn't do this because of some feature of incompatibilist arguments that I apparently haven't understood. Is it necessary to give a prompt that fails to capture the thesis of determinism? I mean, how does the imprecise prompt help things? What's wrong with the idea that if a prompt is given it should be one that correctly states the thesis? I take it that you agree, and for this reason plus the reason that I ask about earlier (the part about "premises appealing to intuition" you think such attempts at describing determinism should be avoided? But that somehow we'll manage to still get at people's views about the consequences of determinism (rather than distinct cousin theses of determinism perhaps)?

Randy and Fritz,

First of all, I should have been clear that I'm referring to arguments about moral responsibility rather than arguments about free will (that's why I referred to rule B rather than beta...)


I understand that the arguments as a whole contain a description of determinism. But, as I see it, the premises that appeal to our intuitions are not the same premises that describe determinism--and moreover we don't have to have a description of determinism in order to evaluate those premises' intuitive plausibility. The role that determinism plays in these arguments (as I see it) is this: Assuming we find those premises to be intuitively true, when combine them with an accurate description of determinism, we get the incompatibilist conclusion. But that's a matter of logical analysis plus an accurate understanding of the implications of determinism. It is not directly related to the key intuitions that X-Phi studies are supposed to be testing.

So, for example, here's a simplified version of an argument for incompatibilism about moral responsibility.

1. The truth of determinism entails that all actions we perform are caused by processes that trace back to factors beyond our control.
2. If an act is the result of processes that trace back to factors beyond the agent’s control, then the agent is not morally responsible for that act. (A rough approximation of The Transfer of Non-Responsibility (TNR) Principle. )
3. The truth of determinism would entail that no one is morally responsible for their actions.

Now, I take it that both compatibilists and incompatibilists agree with premise 1. Their disagreement is over the plausibility of premise 2. But we don't need a description of determinism to evaluate the plausibility of (2). Rather, we need is an accurate understanding of the TNR principle. That might involve either an appeal to an intuition about the fairness of such a principle or intuitions about particular cases in which acts are clearly caused by processes that trace back beyond our control.

So (as I see it), X-phi studies should be probing for intuitions about the TNR principle in general or about particular cases, and not about the compatibility question itself. I hope that makes some sense. I hate to do the "as I've written" thing but I lay it out a little more clearly in my Compass piece Experimental Philosophy and Free Will, in the critical section.


My point was not that we should give imprecise or ambiguious descriptions of determinism. My point was that we shouldn't be giving any descriptions of determinism, rather we should probe more directly for the intuitions that make incompatibilist arguments successful or not. As for what it means for a premise or principle to appeal to intuition, doesn't van Inwagen admit that he relies on intuition for Rule B? About principle beta—the ‘transfer of powerlessness’ principle that he later applies explicitly to Rule B as well,he writes:

I must confess that my belief in the validity of Beta has only two sources, one incommunicable and the other inconclusive. The former source is what philosophers are pleased to call "intuition".... The latter source is the fact that I can think of no instances of Beta that have, or could possibly have, true premises and a false conclusion. (Van Inwagen, 1983, pp. 97-99)

In my view, both of those sources involve appeals to intuition--how else are we supposed to assess whether something is truly a counterexample?

I move that someone start a new thread so that we can talk about the Sehon article!

In one sense, that article cracks me up since the absurd conclusions that follow -- determinism is impossible, the current definition of "determinism" are flawed -- only follow on the assumption that the consequence argument is sound. One could look at the paper as a reductio to the consequence argument, especially since standard compatibilist responses to the argument (Lewis, Perry) would appear to work for Sehon's interventionist God problem, as well.

After all, all Sehon really shows is that if determinism is true, then even a God cannot do otherwise. But that is absurd. The consequence argument and the incompatibilist assumptions connected with it appear to be the real problem.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan