Blog Coordinator

« X-Phi bleg | Main | Why Folk Intuitions Matter to the Free Will Debate and Why we Should Study them Empirically »



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

Here is what I think is happening:

Both incompatibilists and compatibilists are acknowledging that human thinking about free will is infected with errors, but they are using that information in radically different ways.

Skeptical incompatibilists have always assumed that human thinking about free will is error ridden, and that forms the basis for skepticism about free will.

What we're seeing Nahmias et al. do is use these errors (i.e. cognitive biases) to undermine incompatibilist arguments/research. Previously, traditionally, incompatibilists have used experiments to suggest that the folk deny free will while fully understanding the experiment stimulus. Nahmias et al. are trying to show that the folk actually don't understand the stimulus, thereby undermining the conclusion that the incompatibilist wants to draw.

I definitely agree with Nahmias et al. that many people can deny free will for reasons less sophisticated than philosophers like Pereboom. But, as Knobe helps to show here, the reasons for this misunderstanding are more fundamental, and interesting, than Nahmias's results might otherwise suggest. Knobe's results suggest to me something like a fundamental, biological-based cognitive bias toward human exceptionalism about human thought processes (i.e. that human thinking isn't governed by classical physics). That bias would be similar (if not coextensive) with biases like the fundamental attribution error.

In sum, no matter what, all parties to the debate will have to acknowledge that human thinking about freedom and decision-making is very flawed. The question is how each party will accommodate that information. Skeptics like Pereboom can use the information to explain why people mistakenly believe in free will. Ironically, neuroti-compatibilists like Nahmias can use the information to try to undermine incompatibilist arguments. In other words, in defending realism (i.e. common sense) about free will, compatibilists can rely on evidence that human thinking about free will is even more erroneous than we initially thought. That strikes me as a very ironic, and clever, use of the evidence.

For that reason, I sometimes think of compatibilism as a "Phew!" view: after the compatibilist acknowledges all of the cognitive biases affecting human thinking on this topic, and acknowledges the possibility that the world is deterministic, the compatibilist can nevertheless say "Phew! Belief in free will survives all of those assaults! We dodged so many bullets!"

Kip has some really good insights. I'll try to spin off them in a bit more holistic way.

People tend to operate behaviorally and reflectively within the context of some kind of world-view. Generally these are (in my estimation) theistic-overall (tending theistic), theistic-specific (committed theists), supernatural-overall (non-specific theist sympathetic), non-supernatural overall (tending to disregard mysticism by some reflection or neglect), and non-supernatural-specific (non-theists and willful agnostics). I could, I think, also file these categories under epistemic degrees of committment, but I'll forgo that.

Where people are in terms of world-views I would argue guides how they respond to specific questions about free will in relation to responsibility. Committed people (theistic-specific and non-supernatural-specfic) probably tend to sort questions about free will brusquely into categories of who-we-are-as-children-of-God or not, and frame their answers to naturalistic causal explanations of human decision-making and action accordingly. When naturalistic causes intrude in such questions, then we (as children of God) are exempt from their influence and are thus responsible, or are subject to them (as natural beings) and are not responsible. Everyone else is not systematically committed to answers to such questions, and thus their answers vary as context of question and their own mood strikes them.

One challenge for X-phi is the "great unwashed" in terms of a lack of willful committment one way or another. How does one sort out this (probably huge) influence on survey results? How can one ascertain that world-view is a direct influence on response to such questions, or that a lack of coherent world-view is a measurable variable of the nature of response?

Maybe the big question is: how can we measure the lack of certitude on such matters and that impact on responses as against certitude that guides responses? I can barely frame the question; I certainly haven't the faintest how to answer it with reliable empirical foundations.

Yes, Josh, these really are an interesting set of results. Thanks for sharing them with us!

I am tempted to agree that people are making mistakes in the Nahmias/Murray cases. Is it possible that folks are hearing “everything that happens is completely caused by something that happened earlier” as “everything that happens is completely caused by something that happened in the remote past”?

And I really like your own effort to get at the possibility of a tension in ordinary thinking between reasons and causes. I’m left wondering if the folk are feeling this tension only with respect to practical reasoning. I wonder, that is, what they would say in response to a prompt about theoretical reasons--something like:

In universe A, the considerations that actually support conclusion p have no effect on what people end up being caused to conclude about p.

You interested in hazarding a guess?


That's a really nice question about theoretical reasoning. I'm not quite sure, but my guess is that you'd actually get the effect there too.

You've also got an intriguing suggestion with the idea that people think immediately of causes in the remote past. I'm not sure if that can explain the whole phenomenon, though. For example, how would it explain the difference we find between, e.g., the mechanical effects of emotion and decisions based on reasons?

V. Alan,

I like the idea that it's important to distinguish people who have different sorts of religious views, and I definitely agree that that is a question worth exploring. Are you thinking that these questions about religion will prove important in explaining some of the actual results from studies in the experimental philosophy of free will? That is certainly possible, but one thing I find striking about these results is that we seem to get very similar answers even in very different cultures with different religions.


Nice point. If we find that there is en error here, it's not at all clear who should be declaring victory -- the free will skeptic (who could say that our belief in free will is founded on an error) or the compatibilist (who could say that it is only an error that leads us to move to incompatibilism). Definitely an important issue for further work!

"...if they hear that a behavior was completely caused, they immediately conclude that it could not have been the result of a decision based on beliefs and desires."

This suggests that many folks think there's something contra-causal about human decision-making. It's consistent with what I hear all the time in conversations with non-philosophers: that if we're completely caused then we don't make *real* choices. The agent's reasoning in advance of behavior isn't really *her* contribution since in Universe A she herself is fully caused to be the way she is and in her reasoning. She is merely a pass through, not an originator. This suggests that what at least some of the folk have in mind as necessary to make real choices is to be causa sui. As Kip suggests, it's "exceptionalism about human thought processes."

So instead of the by-passing hypothesis, I suggest the pass-through hypothesis to explain these results. In thinking about Universe A, people rightly infer that reasoning is completely caused, hence (on their incompatibilist view) can't be a real, agent-originated contribution to behavior. Therefore it is judged not to have any effect when compared to a real, honest-to-goodness contra-causal decision, which doesn't exist in Universe A but does in our universe, they suppose.

Joshua (if I may)--

Thanks for responding. Yes, I do think that religious commitment makes a difference. But my bigger point is that people tend to have what I call world-views--how everyday events in lives fit within some "big picture". And my further point (by my armchair distinctions above) is that some world-views are better defined/structured/guiding than others when it comes to responding to pointed questions about human choice, action, and responsibility (and maybe even more mundane things like taxes and gun-control). My (also armchair) sense is that some responses have a greater coherence and coordination given a more defined/structured/guiding world-view. For example, a classic Calvinist would (I take it) tend to respond in a way that diminished the role of free will (unless they were making distinctions between a "human" take on such matters as against a God-like perspective, which again would be hard to measure with a questionaire that did not make allowances for such subtleties). Most common folk do not have such well-defined big-picture views (what I called "sympathetic" or "tending" in commitment), and so I'd expect that their responses are less coherent and more context-relative. What I wonder is if there is a way to control X-phi questioning to determine if my hypothesis about fixity or non-fixity of world view as correlative to coherence or noncoherence of response is correct, irrespective of culture or sub-culture.

And like the song, you can call me Al or Alan. (I'm forced to use the V. Alan because there is another Alan White--not Alan R. either--who currently publishes in philosophy.)

Tom (and Joshua)--

So if a question is taken personally--it's all about him/her--then that one will typically resist denial of incompatibilist-style FW. But if the question is not taken personally, then might responders creep toward compatibilism? In partial support of this claim, one fact that I have (tediously) repeated on this blog is that after the John Hinckley trial, Fed lawmakers walked away from FW considerations of judicial responsibility at the Fed (and many states') level by legislatively renouncing the relevance of the Model Code's second prong, which assesses whether one could have conformed behavior to the law. (That's Wolf's asymmetrical freedom to choose the good BTW, which is itself completely consistent with some compatibilist and incompatibilist interpretations of that prong.) Whether that renunciation supports FW nihilism or some compatibilism that subverts incompatibilist interests is up for interpretive grabs. I think it is due to knee-jerk politics myself, and falls compatibly with many FW positions quite by accident. But my point here is that what happened post-Hinckley should be full-forced examined by X-phi. Our insanity laws relative to FW at both the Fed and state levels are comprehensively radically incoherent, and perhaps X-phi could help us understand what's going on. I suspect it's just basal collective incoherence about human belief (world-view) systems. And my guess is that the pragmatics of politics presides now about how we deal locally and nationally with questions of FW and responsibility. But overall my point is that in law, there is no systematic consistency and coherence of what FW means with respect to responsibility--that is just a fact. A fact that X-phi needs to (help) explain.

These are really nice studies. Four things.
1] In the N&M studies, I wonder if you’d get a difference if you subbed ‘end up being caused to do’ with either ‘choose to do’ (which might bias them a bit) or simply ‘with what they do.’ In Josh’s studies, I wonder if you’d get a difference if you subbed ‘end up performing’ with ‘perform’ or ‘choose to do.’ Would any of these substitutions be out of bounds for some reason?
2] I wonder if you’d get a difference if, instead of asking whether wants etc. ‘have no effect,’ you asked whether wants etc. ‘are among the causes of’ what agents do. This would tell us a bit more about how the folk are reading ‘have no effect.’
3] In the N&M studies, I wonder if you’d find a difference between people’s wants and beliefs having an effect vs. people’s decisions if you presented the questions separately. I have a flitting intuition that if asked only about decisions, people might still think that decisions have an effect on what agents do.
4] This is a bit off topic, but the emotion study is interesting. I wonder if you’d get a difference between emotions causing actions and emotions causing facial expressions. Just curious about folk views on the relations between emotions and action – whether emotions are in a category of things we have no control over or whatever.


Very interesting. I think Tom Clark has it right here--the folk notion of belief and desire seems to entail that something cannot be a *real* belief or desire if it is caused as per universe A. This also explains why my undergrads all have the intuition that computers don't think because they are "just programed" and also that *because* we really think, we are not programed. Folks are not making any error here--they are consistently applying their ordinary contra-causal notions of belief and desire.

About your emotion case. I wonder if in universe A someone can perform an action out of love or hate? Can one feel pride in one's actions? Shame? Or are those notions contra-causal as well, when they are tied to action?(Questions about "crimes of passion" and moral responsibility leap to mind here...)

Lots to say here about earlier points and Josh's cool followup studies, and I'll try to say more soon. But two quick things for now:

1. In follow up studies, Dylan and I made the questions less awkward and they had this form: "In Universe [A/C], what a person wants [believes] has no effect on what they end up doing."
The main effects were the same (i.e., bypassing responses not only correlated highly with no FW/MR responses, but they mediated them, suggesting such misreading of determinism caused apparent the incompatibilist responses).

2. I am really, really surprised to hear people saying that the folk think that your beliefs and desires can't really be yours if they are fully caused by prior events. I'd be interested to see someone design a study to demonstrate that result. I don't think you'd get it. Here's a quick way to test it. Give them deterministic universes (like Josh's Universe A or our rewinding Universe C) and ask agreement about:
"In Universe [A/C], people don't really have beliefs [desires]."

I also think you'd only get agreement to the following question if the subject was also misreading the scenario to involve fatalism or bypassing of mental states:
"In Universe [A/C], people don't really make decisions [choices]."

I think people are OK with mental states being caused (even fully caused), as long as they are caused in the right sort of way. Being fully caused by the distant past (which, I think, is a false description of a consequence of determinism), or by one's brain states, easily leads people to think their mental states are not being caused in the right sort of way, because it's hard to see how full causation by such things also allows full causation by the proximal things that matter, like my previous thoughts and deliberations and events in the external environment.

People tend to have what I call a Single Explanation Assumption (SEA). (More on that later.)

For what they're worth, here are some quotes from a recent exchange on free will at that illustrate the conflict in some people's minds between the idea of being fully caused and being an agent that really controls, decides and chooses. This person doesn't conflate determinism with fatalism, but (despite my best efforts to persuade him otherwise) still believes we don't make *real* choices. This seems more like pass-through than by-passing, but of course it's just an N of 1.

"If we are caused to an action, how can we have control over that action?"

"I can’t get behind the idea of humans and agents if we accept that humans are fully caused. For the record, I do believe we are fully caused. If we are fully caused, then in a given situation we couldn't have “decided” otherwise. If that’s the case, why is this different than a ball “deciding” to roll downhill? I contend they are the same, and if they are the same, then calling us “agents” if we don’t call balls agents is just a semantics game."

"If I’m truly fully caused, then the molecules in my brain are acting/reacting only as they can, and no other way. If the atoms that make up those molecules are fully caused, and fully material then they are acting only as they can. If the baryons that make up those atoms… and the quarks that make up those baryons…etc. are all fully caused all the way back to the creation of our universe then……well then my free will is an illusion."

" perhaps I failed to address the precise definition of fatalism you [Clark] point out here. Of course my actions make a difference, I buy that...if fatalism means my actions don’t matter then I am not a fatalist. So I’m not determined to conflate determinism with fatalism (although, really, could I?) — excuse my little fatalism joke. But determinism means that 14.7 billion years ago it was determined that I was either going to be a fatalist or not. “I” have no choice in the matter because that there is an “I” is an illusion anyway."

"My interpretation of choice is that it implies contra-causal free will... When we get to what it means to make a decision, I look at a fully-caused decision and I reject calling that a decision in any real sense."

I have found this thread interesting and helpful. As always, I think Eddy's work (sometimes with others) is careful and of high quality. And perhaps it is interesting to know what the folk think about this and that; if anyone is going to capture this, it is Eddy (and his collaborators).

Of course, I know that many have thought about this kind of question, but I keep wondering: why exactly should we care what the folk think? So, for example, if the consensus among the "folk" is that a mental state, such as a belief, cannot really be yours if it is in place in you as a result of causal determination, why not simply say that the folk are mistaken about this?

Yes, it is of some interest, and perhaps of some philosophical interest, what the untutored or uneducated folk think about such matters. But surely we don't want to stop there, or take the folks' views as controlling or dispositive of the philosophical issues, do we?

(Sorry: I was just using my worm-can-opener.

Here's one thought about the potential relevance of folk intuitions to these kinds of debates. Say the bypassing hypothesis (or some other error theory for incompatibilist intuitions) stands up to further scrutiny. It might be open to compatibilists to claim that incompatibilist philosophers are subconsciously making the same mistake as the folk, or at least that incompatibilists' present views are traceable to an initial mistake they made in their pre-theoretic days.

If it sounds implausible that expert philosophers might fall victim to the same mistakes as the folk, consider the possibility that the psychological mechanisms underlying even philosophers' intuitions might be insensitive to the kinds of explicit clarifications they make when they try to isolate the right independent variables (e.g., determinism) in the relevant thought experiments. In that case, incompatibilists can say all they want that they aren't taking determinism to imply epiphenomenalism or fatalism -- the subconscious mechanisms that produce their intuitions still might.

And even if philosophers are expert intuiters in the sense that the source of their intuitions can accurately represent even highly unusual scenarios, there might still be a genealogical story that could be told that would undermine the credibility of incompatibilist philosophers' intuitions. Maybe incompatibilists repeated the bypassing mistake so many times in their youth that the only reason they retain the incompatibilist intuition is due to some sort of learned behavior or motivated reasoning? I'm not sure such a story could successfully be told, but it doesn't seem to me obviously impossible.

Anyway, I think there are other ways evidence about folk intuitions could bear on philosophical debates. This just seems like a route that's particularly close at hand for those wanting to progress the debate using the present findings. That being said, I don't even know if compatibilists would be open to the kind of argument I've outlined here. But it's a thought.

I think Adam is completely correct: that's one reason why these debates might be relevant in philosophical theorizing. There is a very large body of evidence that people often reason from, rather than to, conclusions: we look for reasons to support our initial judgments. In some areas of philosophy, especially ethics, we need to look for debunking explanations because the reasons offered for a view are so bad (opposition to gay marriage, for instance, or to certain kinds of cognitive enhancement). I don't think this is the case in the free will debate, but even here, when debates issue in dialectical stalemates, tie-breaking genetic arguments might have some force.

But I also don't think that these issues require some extra justification. These debates pass the Levy test for whether a debate is philosophical: they engage the attention of philosophers in professional contexts. I doubt any rival test would do better in telling us why interpretation of Mencius, settling whether theory of mind is simulation based, discussion of the role of genes in phenotypic traits, assessing whether qualia have semantic content and deciding whether some contradictions are true all count as philosophical. They count as philosophical because enough philosophers find them interesting enough to engage in professionally. I actually am not especially interested in this debate, but it just goes on the long list of philosophical topics in which I'm not particularly interested.

Hi John, thanks for the high praise. In the meantime, your question "why exactly should we care what the folk think?" has been raised often enough that I think it's worth giving a systematic answer. I'll write something up and make a new post.

But I hope people will still think more about Josh's questions and results. I'm more curious to hear what others think than to hear what I think about them.


Good--I agree that the most promising use of data on folk intuitions is to identify "psychological mechanisms" so that we can be aware of them and perhaps sensitive to their influence in our philosophical work. Your comment is helpful in reminding me of this point, which also was developed nicely in the introduction to the Experimental Philosophy collection edited by Nichols and Knobe.

But I also think it is helpful, or might be, to keep in mind that this is indeed the most illuminating use of data about intuitions, rather than the "first-order" issue of what is correct to say about the relevant problem. So I think Josh Knobe introduced this thread by saying that he thinks Nahmias and Murray are onto something "deep and important about the way the folk think about how people understand human action"; yes, but this is different from something deep and important about human action. Maybe one can get from the first point to the second, but there is ground there to be covered.

For what it's worth, I think that folk surveys can be important because:

1A. they tend to show how people use words
1B. the meaning of those words is governed (entirety or partially) by how people use those words
1C. specialists and philosophers do not have a monopoly on those words - they are part of the larger community, and it is the usage of the larger community that influences the meanings of words

For example, I don't think that compatibilists should be able to say "well, we are using our own special compatibilist definition of free will" if that definition is inconsistent with how the public in general uses the term.

The other reason is that:

2A. surveys and research can expose cognitive biases in the way that people think about philosophical problems
2B. these biases largely explain the origin and solution (if any) to the aspect of the problem arising from the biases

Regarding the latter point, the relationship between philosophers and cognitive biases would be similar to the relationship between scientists of perception and visual illusions. Even if the scientists no longer see the visual illusion, or suffer less from it, they still want to study how it affects people in general.

And - although these two ideas (semantics and cognitive biases) seem narrow - I actually suspect that they explain and motivate most of what is interesting in philosophy.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan