Blog Coordinator

« Personal Identity and the True Self | Main | Causal Determinism and Going with the Flow »



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

Hi Eddy,

Great to hear from you. I would love to talk more about these broader issues (including the point you make in this most recent comment), but first, I'm super curious to hear your thoughts on the specific study I discuss in the post. What do you think explains those results?

I'll think more about it (and Morgan Thompson and I say a little about your results in our companion piece to yours:, but let me ask two questions first, one methodological, one about interpreting the results:
1. Did the materials for this study use the full description of Universes A vs. B that specifically state that the only difference between them is human decisions? Or did you use the shorter description of just Universe A (like the one in Bear and Knobe)?

2. For the people who agree to the 'bypassing' question in response to determinism (Universe A), do you think that they are (implicitly?) thinking that in that universe, people cannot make decisions? cannot make decisions based on reasons (or values)? cannot influence what is determined to happen (regardless of their mental states)? I ask because some of these interpretations might suggest these participants are committed to a libertarian (perhaps dualist) view of the mind/self and that's what leads them to think determinism precludes a causal role for certain mental states (because those mental states must derive from that mind/self?), but other interpretations might suggest these participants are making a more substantial mistake regarding what determinism means than what Dylan and I propose.

Hi Eddy,

These are all really helpful points. Here are a few thoughts in response.

1. You suggest in your comment that people are making a 'mistake.' I wonder if you could say a little more about what you mean by that. One possibility is that you just mean that people are thinking something that is false. For example, it might be that people believe that indeterminism is required for certain psychological processes, but that this belief is actually false. That certainly sounds plausible to me, but is that what you mean? Or do you mean something else?

2. Regarding dualism, I don't know if this is common ground between us, but my sense is that all studies indicate that people do not think that mind-body dualism is required for free will. As philosophers, we may think that dualism is somehow required to make sense some of the things that people say, but all the same, the cognitive processes that people themselves use to ascribe free will just don't have anything to do with checking for mind-body dualism. (The researchers who claim otherwise seem always to be people who ignore all of the actual empirical data.) Do we agree on this point?

3. My hypothesis is that what is driving these intuitions is not the way people think about mind-body dualism but rather the way people think about *causation*. On this hypothesis, people think that mental states play some role in free human actions but that mental states do not literally *cause* such action. Thus, in a universe in which everything is causally determined, mental states cannot play their usual role.

4. The point you make about methodology is right on target and very helpful. In this study, participants are introduced to a deterministic universe and told explicitly that human beings in the deterministic universe can make decisions. This deterministic universe is then contrasted with one that differs only in that human decision-making there is indeterministic. Just as you say, the later paper with Adam Bear also shows that people think that some psychological processes are not possible in a deterministic universe, but in that one, we do not use the contrast with a universe in which decision-making is indeterministic. To continue exploring this issue, it would be great to try re-running this study using a Bear-style description of determinism.

Thanks again for all your thoughts! From the very beginning of our work on this topic, I have really valued your comments and suggestions.

Hi, Joshua

I have a couple of brief comments/questions.

1. In the human vs. computer case, the result in the human condition was M=5.4, and in the case of the computer, it was 3.6. But that still suggests a good amount of disagreement between different people assessments in both cases, or else that a good number of people partially agree, but also partially disagree with the statements. Something similar happens in the beliefs and values vs. emotions case. What do you make of that?

2. Regarding your conclusion, might it be that a scientific understanding of free will is actually compatible with the ordinary concept of free will, even if not with some ordinary beliefs about free will?
My concern here is that (some, but not all) people might think that in universe A, there is no freedom, but they also think that A is a universe very different from ours. But what if they thought that's our universe?
In one relevant respect (but not in several others), the following example might be a parallel. We believe that XYZ in Twin Earth is not water, but that's because we believe that water is H2O. But if we stipulated that scientists got it wrong and water is actually (here, on Earth) XYZ, under that condition we'd say that XYZ is also water in Twin Earth.
If people were told that their ordinary belief (granting that they do have that belief) about free will is actually not true in our universe, would the statement that beliefs and values in universe A have no effect also give results like 5.4, or 5.7?
I suspect the number might go down, though I don't know how much. And there is a further question about whether they'll be correct about that; more precisely, what would be the proper assessment in such cases?

Hi Angra,

These are both great questions. A few quick thoughts in response:

1. You are completely right to say that there was some variance in participants' responses within each condition. This is a phenomenon we observe in every experimental philosophy study -- we never find that all participants give the same answer -- and it would be worth thinking more about what it means.

2. The prediction you make here turns out to be completely correct. This issue was investigated in a really nice paper by Roskies and Nichols:

Angra's question about how people's responses change depending on whether they are thinking determinism applies to the actual world is really interesting. In addition to the cool Roskies and Nichols' paper that explores it, we got some results that suggest more 'compatibilist' looking answers in real world cases than alternate world cases in this paper:

I've always assumed that even though some people have incompatibilist intuitions (libertarian conception of FW), if they would revise their view to a compatibilist one in response to coming to believe the world is deterministic (or fully lawlike or whatever), then that provides support for a compatibilist theory. But I'm not entirely sure about this, especially since a most prominent incompatibilist, van Inwagen, famously says he would become a compatibilist if he were convinced determinism is true. Anyone have thoughts on this issue?

Josh, in response to some of your questions:
4: yes, it'd be interesting to try other descriptions of determinism, including Bear-style and maybe ones I've used. As you know, one worry I've had with the A vs B description is that the salient (only) difference between the universes is put in terms of the 'had to happen' language applied to decisions (which Murray and I showed primes some people to think not just that bypassing occurs but that fatalism occurs). It would not be surprising if some people responded to that language differently regarding human decisions (which even most compatibilists think require alternative options) than regarding human emotions or computer programs, which are not associated so strongly with alternative options.

It might also be useful to test whether a clear description of complete causation, but with some indeterministically caused events, would influence responses. If I understand what you are suggesting in 3, you would predict that people should be (almost) as resistant to the possibility of values effecting actions in such a universe as they are in the deterministic universe. Or am I missing your point here?

Regarding 1, suppose some people think the definition of 'determinism' is 'the opposite of free will' and then we ask people: Imagine a deterministic universe; Could people have free will in that universe? Negative responses would not tell us whether those people have incompatibilist intuitions, because incompatibilism is about whether determinism, defined in some specific way both parties can agree on, rules out free will.

Now suppose some people think determinism (at least as they read Universe A) means that reasons can play no causal role in a person's actions. At least among the disputing philosophers who think that determinism does not have that consequence (perhaps because they also agree that reasons can be causes), it would seem that people's responses would not tell us about whether they have incompatibilist intuitions. At least, that's what I was thinking.

But now I'm unsure. Suppose you have shown that (some) people are deeply committed to a theory of decision-making that says some mental states, such as reasons, do not *cause* actions (do you also think the evidence suggests people do not think reasons can be caused?). Then, assuming free will requires that our reasons can play a role in our actions (what role, if not causal?), it looks like people are incompatibilists about free will and determinism (or any theory that says the only events that play any role in our actions are those that play a causal role in them?)

Sorry if I'm getting confused about your view or just confusing myself.

In any case, I can certainly agree with you about 2--and it drives me a bit crazy to see the willusionists (like Sam Harris and Jerry Coyne) insist that everyone thinks free will requires dualism. (Alas, notice that the first issue comes in again here, since some people might think dualism is true, but also be 'theory-lite' enough to think that if our mental states are shown to be physically instantiated, they'd be OK with that.)

Hi Eddy,

This comment raises a whole lot of really interesting issues, but I wanted to focus in on just one of them. You suggest that people might be somehow inferring from the idea of determinism to the idea of fatalism. This certainly seems like a plausible hypothesis, but I was trying to provide evidence against it.

The evidence is very simple. If you look just at the responses of the participants who were asked about the relationship between beliefs and actions, those responses might reasonably be interpreted as indicating that participants are assuming the universe is fatalistic. However, if you look at the responses of the participants who are asked about the relationship between emotions and facial expressions, those responses clearly reject fatalism. Thus, we have reason not to interpret the responses from the participants who are asked about beliefs and actions in terms of fatalism. It seems that there must be some other explanation.

Of course, someone might think that my other explanation is no good. However, independent of whether my explanation is any good or not, would you agree that the overall pattern of results seems to indicate that people's responses are not driven by a belief that the deterministic universe is an entirely fatalistic one?

p.s. So glad you agree with me about that dualism thing! We have got to find some way to stop people from repeating that claim while ignoring all the evidence to the contrary.


On point 1., I think the issue depends on what kind of disagreement there is. For example, if you get M=5.4, maybe that's because everyone picked either 5 or 6 and a bit more picked 5, or maybe a lot of people picked 6 or 7, but a non-negligible minority picked 1. Those are very different scenarios, and I don't know which one is the case.
Assuming that a non-negligible minority picked 1. or 2. (i.e., they disagree with the statement that values and beliefs have no effect on actions), then I would suggest the following hypotheses.

a. The disagreement is the result of a disagreement about what's going on in the actual universe, and the concept of the self and/or causality is sensitive to what is actual (in a sense vaguely reminiscent of rigid designators).
b. The disagreement happens because some of the people are making a mistake in their assessments of what would happen, under their own conceptions of agency, self, etc.. If that's the case, I would be inclined to say that the majority is wrong (though that's because I would disagree with the statement, and I don't see a good reason to yield to majority judgment on the matter).
c. There is more than one ordinary concept of agency and/or the self and/or actions based on reasons, even if one of them is more common than the other or others.
d. A combination of more than one of the previous hypotheses.

Regarding 2., thanks for the link. It supports my prediction to some extent, though the issue is framed differently, so I think it's not conclusive.
That aside, it's interesting to me that here is a considerable amount of disagreement, it seems, about whether people should be blamed, can be fully morally responsible, even when the universe supposed to be the actual universe. Errors are probably common, I think (the alternative seems to be that moral concepts are also different in different people, and different enough to explain the results; I don't think that's probable, but if that were the case, it would seem to support some sort of (perhaps partial) moral error theory, or relativism).


Your hypothesis (3) makes me wonder: have X-Phi'ers investigated whether the folk conception of "cause" is deterministic? Would lots of the folk regard "probabilistic causation" as an utterly novel idea (or worse, an outright contradiction)?

Supposing that they would, my next question is, do the folk also reject the idea that *decision* can cause action? Because to the casual observer, decision is the only mental state that highly reliably leads to a specific action, i.e. the one that was decided on. Masochists steer toward pain, not away. Ascetics avoid pleasure. Macho men confront the things they fear. Etc.

To clarify my last post, I'm wondering about how people use the *word* "cause"; I'm not so much wondering about how they reason about (what most philosophers would call) causation. I found this nice compilation, which unfortunately doesn't look promising for an answer to my question:

Hi, Eddie

Thanks for the link. I don't have access to that paper, but I take it that the results go in the same direccion as the paper Joshua mentioned.
I would say the results support a hypothesis that is relevant in this context and that entails compatibilism, though whether they support also compatibilism would require further analysis (and it likely depends on other pieces of information available to the person making the assessment).
More precisely, let Sen be the hypothesis that whether it's possible that an agent makes free choices in a deterministic world is sensitive to how things are in the actual world, determinism-wise (here, "possible" is the most usual sense of metaphysical possibility).
If Sen is true, it seems that the concept of making free choices is compatible with determinism (else, there would be no such sensitivity, and regardless of how things are in the actual world, it would be impossible that an agent makes free choices in a deterministic world).
The results support Sen. But I'm not sure they support it strongly, given how frequent folk mistaken assessments probably are, on these matters (then again, that's always a difficulty when doing experiments asking people to agree or disagree with statements, etc.)

Regarding van Inwagen's reply, that's intriguing. I don't have the full context, but at least as far as I can tell, the answer seems problematic to me, for the following reasons:
As long as he holds that the concept of free will is incompatible with determinism, it seems to me that van Inwagen is implying that if he were convinced determinism is true, that would provide decisive evidence against his conceptual analysis (which led him to the incompatibilist conclusion), because while he holds it's very improbable that the concept of free will is compatible with determinism, he holds it's even far more improbable that we don't have free will.
But that's a problem for his arguments against compatibilism, it seems to me.
For example, in his Consequence Argument, he says that since it's not up to us what happened before our birth, and since on determinism our actions are the consequences of things that happened before our birth, our actions are not up to us on determinism. But given his implication above (if I got it right), van Inwagen would have to concede the following amendment:
Since it's not up to us what happened before our birth and on determinism our acts are the consequences of things that happened before we were born, our actions are not up to us on determinism, unless determinism is true, in which case even though it's not up to us what happened before we were born and on determinism our actions are the consequences of things that happened before we were born, our actions are up to us..
However, it seems to me that an exception like that seems to at least strongly undermine any intuitive appeal of the Consequence Argument, at least as an argument against compatibilism in the conceptual sense (in this case, compatibilism between determinism and up-to-us-ness).
Do you have an alternative interpretation of his position?

Angra--on van Imwagen--

I take the CA to be an argument *for" incompatibilism on the specifically given premises, not against compatibilism, and it could hardly be so given that compatibilist accounts of what is up to us, what we control, what we can plausibly be accountable for, and so on, is a very open set. Should the CA fail, then by default something else of a FW nature should be true *if* any such account is true. It seems that what van Inwagen says in that case that he'd prefer some other FW account as opposed to FW nihilism.


I'm not sure I understand some of your points; I'd like to ask for clarification on the following issues.
a. I'm not sure how an argument can be for incompatibilism without being against compatibilism as well, if we take compatibilism and incompatibilism to be clearly incompatible with each other. Maybe we have different definitions?
You say it's for incompatibilism "on the specifically given premises"; you mean, that if determinism is true (i.e., the premise that it's true), then there is no free will?
If so, that would seem to me like an argument for lack of free will on the specifically given premises, and an argument for incompatibilism (and against compatibilism). But as I said, I'm not sure I understand your reply correctly.

b. What do you mean by "a very open set"?
Do you mean there aren't many such theories? If so, I agree given the theories developed so far, but I'm not inclined to say that any single one of those theories is probably true.

c. I agree that van Inwagen would prefer some other FW account rather than FW nihilism. However, I still don't see how he'd avoid the implication I mentioned.
Do you think that he's not committed to the view that if he were to be convinced (by some a priori argument, empirical evidence, both, etc.) that determinism is true, then that would be decisive evidence against his conceptual analysis (the one that led him to the incompatibilist conclusion)?
I'm thinking he's committed to that because he seems to imply that on the basis of determinism, he would (properly) come to the conclusion that incompatibilism is false, and so it follows his conceptual analysis is mistaken too.

Hi Joshua,

Great posts and discussion (nice to see Flickers spring to life!), most above my pay grade but I'll toss in a little anecdotal evidence. You ask:

"People surely recognize that mental states play some kind of role in the generation of free human action, but if they don't think that human actions are causally determined by mental states, how is this role to be understood? To begin addressing the question, I suspect we will need to look to a quite different way of making sense of the self."

Here's what one of the folk wrote in a Facebook discussion on free will that has mental states playing a non-causal role. Instead, they are *selected among* by the radically free conscious self when embarking on a course of action:

"We have evolved consciousness which gives us the ability to make completely uncaused choices...Though we are tethered to our experiences and our brains, our reasons for our choices are never the causes of our actions. Our choices and actions happen independently of our reasons, i.e., our reasons do not compel us to act in any way. Nothing compels our choices. We're always free going forward. In every moment choice is open to us. That is what consciousness has enabled - free will among living beings."

In studies you could ask folks to agree/disagree on a Lickert scale with a statement something like this, see what you get. I suspect you'd get many disagreeing with such unbridled libertarianism, but I also suspect that the notion of the self as an uncaused arbiter - a point-like controller-essence for which reasons are merely options, not causes - explains at least some of your findings.


Exactly! This quotation perfectly captures the idea that I think we need to be exploring further.


The question about probabilistic causation is a great one. I actually don't know of any work on the topic.

In any case, I should clarify that I wasn't trying to call into question the idea that people think of actions as being caused by decisions. Rather, I was trying to call into question the idea that people think of decisions as being caused by mental states (beliefs, desires, etc.).


You are completely right to ask about the distribution. The answer is that most participants completely agree with that statement, but a small percentage completely disagree. Not a single participant gave a response at the midpoint of the scale.

Hi again Angra--and apologies to Joshua for thread-meandering--

If the CA constituted a proof of incompatibilism, then naturally compatiblism would be shown false. Still, that fact does not entail that the CA is in any way directed at compatibilism. So, if the CA fails, it just fails, and there is no logical consequence except that both compatibilism and incompatibilism remain options (or neither for the possibility of nihilism).

My remark about "the open set" is that since all compatibilist views collectively just deny incompatibilism, then there are a perhaps unending number of ways for compatibilists to account for the needed features of FW--that's why compatibilism is the more fertile ground for continual research and revision (not in Vargas' sense).

I've actually written about why (in my view) PvI made his statement about compatibilism in a chapter for a pop culture book (X-Files and Philosophy). His theological world-view requires some strong sense of personal responsibility for the purposes of theodicy. I see this same sort of big-picture motivation in his famous move toward mysterianism, which at least allows the conceptual space to hold out hope for incompatibilism even with an absence of proof for it.


Given your reading of the results of your experiment, do you think the respondents who agreed with the second statement would also reject rationalizations of actions as explanations?

Put more as a challenge, can we reconcile their agreement with (what I would think is) the obvious fact that when we reference another's reasons for some action it is a species of explanation?


Thanks for the info. That's very interesting.
Since you got M=5.4 out of 41 people, that entails 11 people completely disagreed, and 30 completely agreed that values and beliefs had no effect on actions. (in the values and beliefs vs. computers' programs study).
In the other study (i.e., values and beliefs vs. emotions), M=5.7, and you asked 40 people, so 9 people completely agreed and 31 completely disagreed.
So, the percentage of people who completely disagreed with the statement was 26.8% in one study, and 22.5% in the other.
While those are minorities, they're not negligible, so I would be inclined to suggest the hypotheses I mentioned in my previous reply.
On that note, I would like to ask about the conclusions you draw from the study (and suggest a modification, at least if I got the numbers right).
You consider the responses to be good evidence if we consider certain cognitive phenomena are sufficient for free action, we would be abandoning the ordinary notion of free action and replacing it with a very different one.
However, it seems to me that if one takes the results as good evidence showing what the ordinary notion is, then the conclusion would be that there are two (if not more) ordinary notions, one that is held by most people, but another one held by something like 1 in 4 people, or at least more than 1 in 5, which is a minority but not a minuscule one like the percentage of philosophers and/or cognitive scientists (there is the question of whether one counts 22.5-26,8 % as high enough to be an "ordinary notion"; I'd say it's ordinary enough, but in any case, regardless of what one calls it, it would be the notion held by more than 1/5 and probably closer to 1/4 of the population).
That said, I don't think this is enough evidence that there are two notions at play. It may well be that the concept is sensitive to what the actual world is like, and there is disagreement about what the actual world is like, or simply that many people are mistaken.


Thanks for the explanation.
I have to admit I still think he's committed to the conclusion I mentioned in my reply to Eddie, but perhaps the problem here is the concept of "evidence", so let me change that for a perhaps less controversial wording:

It seems to me he's committed to the following conclusion: If he were convinced that determinism is true, then that would make it rational for him to assign extremely low probability to the conceptual analysis that holds that free will is not compatible with determinism.

If that's right, I would say that that strongly undermines his argument, for reasons similar to the ones I suggested.

Hi Matt,

This really goes right to the heart of the issue. What I am proposing is that there is an important difference between merely suggesting that mental states explain action and actually claiming that mental states *cause* action.

Consider the two sentences:

(1) a. Inflation decreased because unemployment increased.
b. John went to the kitchen because he believed there was a beer in the fridge.

I assume that most people would think both of these sentences sounded right. This indicates that people think unemployment can explain inflation and also that mental states can explain action.

Now consider:

(2) a. The increase in unemployment caused the decrease in inflation.
b. John's belief that there was a beer in the fridge caused him to go to the kitchen.

My sense is that people often use sentences like (2a) but that they do not use sentences like (2b). I am thinking that this is because even though people see mental states as in some way explaining actions, they do not think that mental states can literally *cause* actions. Does that sound plausible to you?

Hi Angra,

First off, my apologies if my earlier comment was misleading. I did not mean to say that every single participant chose either the very highest or very lowest point on the scale. All I meant was that all participants chose either relatively high points or relatively low points, and not a single participant chose the point right in the middle.

In any case, the point you make here is a good one, and of course it could be used equally well to argue against virtually all claims about people's ordinary intuitions (both in experimental philosophy and in more armchair philosophy). When philosophers make claims about people's ordinary intuitions, it may be correct that the majority of people have those intuitions, but there will always be a substantial minority who would give the opposite response.

One obvious question would be about the pattern of intuitions across people. If 80% of participants give a particular answer in a study, it might be that approximately 80% of people are such that they would consistently give that answer, or it could be that all people are such that they would give that answer approximately 80% of the time. Various intermediate positions are also possible.

In any case, I very much appreciate your suggestion that the variance in people's responses might be due to differences in their beliefs about how action is generated in the actual world.

Joshua, sorry I misunderstood your previous reply, and thanks for the clarification.

Regarding people's ordinary intuitions, that's a good point. I'm thinking it might be useful to distinguish between prima-facie intuitions and intuitions after considering the matter more carefully, reading different arguments, etc.
The experiments I've seen seem to measure the former, but intuitions might change (and in some cases probably would, if philosophers are an indication and the meaning of the words they use doesn't change) after further considering the matter, without changing the meaning of the words.

Then again, perhaps a certain percentage of people are already slightly familiar with at least some of the philosophical issues at hand. On that note, I think it might be interesting to ask participants whether they were more or less familiar with that sort of question, or alternatively, to ask people from different countries and cultural environments (as you did when asking about a universe where humans were non-deterministic, etc.), though of course that requires more studies and thus resources, so it's understandable that it's not the norm to do studies in that way. I just think it might be interesting to set up something like that for future studies.


Thanks for that clarification; let's focus on whether mental states cause decisions. Let's further suppose that "cause" here means "deterministically cause". It seems compatible with determinism that mental states don't cause decisions. Mental states are not sufficiently specific. They only probabilify decisions, not "cause" them in this strict sense.


I've said this before on Flickers, but now it seems relevant given your last response to Matt. Do you think that the folk tend *not* to attribute cause-explanations for decisions because they mistake the alternative logical possibilities that they imagine for future actions for a claim that they have accessible alternative metaphysical abilities to engage those imagined possibilities? I'm thinking that alternative future logical possibilities--I imagine I might go left or I might go right through that door--then slides into thinking that I *can* do both. My own thinking is that this is one major basis for many empirically-measured incompatibilist intuitions.

The comments to this entry are closed.