Blog Coordinator

« A Paradox in Punishment (and Reactive Attitudes)? | Main | My Brain Made Me Do It? (More on X-phi and bypassing) »



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


Thank you for this fascinating post. Perhaps I am one of the people you mention, who have challenged you about "sophisticated" incompatibilist intuitions.

Here is my reaction: I am surprised that you don't quote Derk Pereboom, or Galen Strawson, or Neil Levy, or Einstein, or Spinoza, or (I think) Shaun Nichols. I've been reading fw/mr/agency philosophy for a decade, and I don't recall seeing the names you cite pop up *that* often (Feinberg, Swinburne, Malcolm). You don't even quote prominent non-skeptical incompatibilists, like Van Inwagen and Robert Kane - even though I wouldn't surprised if one or more libertarians were incompatibilists for bad, bypassing-type reasons.

On first reading, I do not think any of these sophisticated incompatibilists commit this (refined) error that you outline. The error that you paint strikes me as a genuine error. Determinism does not entail that our reasons do not track our biology, such that our biology rules us in a way that our reasons do not. In my understanding of determinism, reasons map to underlying biology, and so determinism does not entail that reasons are bypassed in any way.

I don't think Neil Levy thinks about determinism that way. I don't think Pereboom thinks about determinism that way. I don't think Galen Strawson thinks about determinism that way. I'm not as good of a philosopher as them, but I assure you that I don't think of determinism that way.

In other words: I still think that you've (I want to say completely) failed to understand why the best incompatibilists *are* incompatibilists.

I want to press upon you how important this insight is - if I am right. I don't want to impress it on you to prove that you are wrong. Perhaps you are right. Nevertheless, even if you are right, it is still fascinating to think that you are (arguably) failing to understand your best opponents.

Let me sketch a brief argument about why the best incompatibilists *are* incompatibilists: constitutive luck. The kinds of people we are, the kinds of desires and goals we have, are ultimately matters of luck. It is not the case (a la bypassing) that, in childhood, we were fully formed beings, with fully formed goals and desires, but, because of determinism, these goals and desires were bypassed. That is *not* the argument about constitutive luck. Rather, the argument is: growing up, we were, earlier and earlier, something like proto-people. Our goals and desires were not fully formed. At one point, we were not fully people, and we had zero or minimum goals or desires. Then, as we aged, we were *endowed* with goals or desires (Smilansky's words). This initial endowment, of initial goals and desires, did *not* bypass earlier goals and desires. The initial endowment did not bypass any other goals or desires, because there were no other goals or desires, prior to the initial endowment - that's why it's the initial endowment.

Nevertheless, even without bypassing, this initial endowment is problematic, because it is a matter of constitutive luck. Suppose, for example, that I was a proto-person (a fetus, say). I could have been endowed with a desire to be a serial rapist. Or I could have been endowed with a desire to a be serial philanthropist. Which endowment I got was a matter of luck. My poor proto-self had no mechanism of self-control or self-analysis to accept or reject this initial endowment - it was simply given to me, and became part of me, and I became a rapist, or a philanthropist, because of it.

*That* is the story of constitutive luck. That is the worry that motivates me, and Levy, and Strawson, and (I believe) Pereboom, and Nichols, and Spinoza, and Einstein, and Russell, and Darrow, etc. The story does not involve bypassing. Perhaps you are right that incompatibilism is wrong. But, if so, you have at least (I believe) reached that conclusion without understanding the concerns of your best opponents.

There are counter-arguments to the worry about constitutive luck. You can argue that constitutive luck is an incoherent concept. You can argue that, when the fetus is endowed with rapist goals and desires, it deserves to be punished later and held responsible for its later rapes. You can argue you many things. But "you're committing a (sophisticated) error about byspassing" should not be one of them.

I agree with Kip that most sophisticated incompatibilists do not in fact make the error you attribute to them. For instance, I don't see how the Consequence Argument involves this kind of error at all, and yet many are incompatibilists because of an adherence to the CA.

I think that Daniel Dennett has some nice arguments against the exclusion approach, although he targets folks like Malcolm rather than Kim (and Kim-style argumentation). I do think some incompatibilists are very much taken by the Kim-style argument: see William Hasker, for instance.

I have a different error theory for the incompatibilists. (I won't say: My error theory is cooler than YOUR error theory.) Some folk tend to fail to distinguish two kinds of freedom or control: freedom to do otherwise/regulative control and acting freely/guidance control. Or, if they have distinguished them, they assume that they necessarily are co-instantiated. But the FSCs show that one can have acting freely/guidance control without freedom to do otherwise/regulative control, and in those situations the relevant agent is morally responsible. Thus, the error is to suppose that the CA's efficacy w.r.t freedom to do otherwise/regulative control implies a general incompatibilism, and specifically, incompatibilism about determinism and moral responsibility.

I don't claim however that this applies to sophisticated incompatibilists who have obviously thought about the FSCs. Such folks are not making an error of conflation or confusion; they are, in my view, wrong about something, but then I suppose I must believe that about everyone with whom I disagree. I don't suppose that entails I have, or must have, an "error theory" for them. (It would be nice, but...)

I agree with much of what Kip said. I doubt many sophisticated incompatibilists are guilty of misunderstanding the implications of 'determinism' as the thesis is typically understood in contemporary debates about FW and MR--roughly, past + laws = unique future.

Indeed, I doubt good philosophers like those you cite, Eddy, are guilty of misunderstanding the implications of determinism so defined. More likely, I hypothesize, is that they are using the word 'determinism' to refer to similar, but importantly different, theses, ones that on dualistic views of the mental might very well involve bypassing.

Hey Eddy,

I'll echo Kip somewhat in thinking that these offered quotations don't look like the best case the incompatibilist could make. For, e.g., isn't the reply to Malcolm's quote obvious? Reductionism doesn't entail eliminativism all by itself. So just because we have intentions that are realized by electro-chemical brain-states doesn't mean we don't have intentions. What else could the brain-state be doing? It's just false that I would climb the ladder without the intention of doing so (under most ordinary circumstances).

Still, I'm not sure that the constitutive luck line Kip's suggesting is so inimical to Eddy's view. Isn't the difference between bypassing fully-formed person's capacities and the case of the "proto-person" similar to the difference between an excuse and an exemption. Sure, if the proto-person lacks the capacities altogether, then it may be true that there is nothing to bypass. But the point, I take it, of pointing to bypassing is that the capacities are not engaged. If you don't have them, they can't be engaged either.

"Thus the movements of the man on the ladder would be completely accounted for in terms of electrical, chemical, and mechanical processes in his body. This would surely imply that his desire or intention to retrieve his hat had nothing to do with his movement up the ladder."

Do we have any idea why intentions (reasons, decisions, etc.) are subject to this sort of confusion and other aspects of ourselves are not?

I don't think anyone would have written "This would surely imply that his *legs* had nothing to do with his movement up the ladder." I doubt anyone would have written "This would surely imply that his *ability at climbing ladders* had nothing to do with his movement up the ladder." Perhaps someone could be tempted to write "This would surely imply that his *stubbornness* [or some other character trait] had nothing to do with his movement up the ladder."?

It is hard for me to judge because I'm so far from sharing the intuition that mechanical causes bypass intentions and the like, but there must be something about intentions, or common beliefs about them, that makes them particularly liable to be subject to this sort of mistake.

I'd think that an error theory concerning beliefs in X would, in the first place, show that beliefs in X are false, i.e., in error. Perhaps it might then explain why, although such beliefs are false, many people believe in X. Eddy, I don't see that the argument you've sketched includes the first part. Perhaps it might better be called a non-warrant theory, since it purports to show that certain beliefs in incompatibilism lack warrant or justification. Not as catchy a name, I realize.

Clearly I was not clear enough, for which I apologize: I am not trying to explain away all of the *most* sophisticated incompatibilists' intuitions, much less their *arguments*. And I'm not trying to psychoanalyze the best incompatibilists like Pereboom, Kane, Strawson, or van Inwagen (at least, I won’t try in public!). I have engaged with their arguments (well, not the constitutive luck one) and do not “fail to understand my best opponents.”

Let me try to be a bit clearer:
(1) Explaining some apparent incompatibilist intuitions, from folk through scientists through at least some philosophers: It appears relatively easy for people to understand determinism in a way that suggests bypassing, and this seems to infect the way these people understand the dialectic. I think Dylan and my evidence suggests that many ordinary people think this way, and it’s clear that many scientists who think free will is an illusion think this way. Indeed, it’s a natural way to think given the very different psychological systems humans use to understand and predict the events of the ‘mechanistic’ world and the events of the intentional world (though I don’t think the latter are understood as *non-causal* as, for instance, Knobe suggests in the paper I link to).
Until we have a naturalistic theory of mind that genuinely explains the reconciliation of mental processes and physical processes, it is very easy to think that the mental (e.g., conscious or rational) is bypassed when told that the physical explains everything (e.g., one way determinism is often understood). Sellars ends his essay like this: “We can, of course, as matters now stand, realize this direct incorporation of the scientific image into our way of life only in imagination.” Until we have a theory to take us beyond imagination, it’s very easy to see competition not reconciliation.
Randy is right that I haven’t offered an argument against, say, the exclusion argument, or for, say, a naturalistic theory of rationality, each of which would help establish that the bypassing worry is in fact false (see my question 3). Nonetheless, if this is the worry driving someone to think determinism undermines free will or responsibility, then it is an error relative to the debate among philosophers.

(2) Explaining more sophisticated incompatibilist intuitions (not the most!): I think it is plausible to think that some philosophers have been moved to adopt incompatibilist theories by *beginning* with bypassing intuitions. I can find more quotations if you want, but there are quite a few introductions to “the problem of determinism” that suggest the truth of determinism would entail that our lives proceed along a set path and we can only observe which way things go (e.g., Carl Ginet’s image of the amusement park ride in which “If one turns the wheel in the directions suggested by the environment—directions in which the car is actually going—one can easily get the feeling that one is steering the car—even though one knows all along that he is not. A child might think he actually was steering the car.”) Anecdotally, when I present the problem to students, I find the most resistance to compatibilism comes from students who adopt this image of determinism as epiphenomenalism. If this is how someone gets bit by the incompatibilist bug, it would not be surprising if they then find the more sophisticated arguments compelling. OK, I’m starting to psychoanalyze…

(3) Explaining the most sophisticated incompatibilist intuitions? Well, I’ll suggest two sketches to answer my second question. In setting up the Consequence Argument, van Inwagen is explicit about describing determinism in terms of the *physical* states of the universe and the *physical* laws, saying “the laws of nature would be just as they are even if there had never been any human beings or other rational animals.” Granted, the argument can be developed without these emphases, but I think it is more intuitive to accept NL when L is the laws of physics than when L is explicitly understood also to include whatever regularities are instantiated in rational action. I think responses to CA using Humean laws are more intuitive when one considers that the laws of nature might *not* have been just the same without rational animals, and I think responses to CA from Perry, Horgan, Lewis, and others aim to remind us that propositions describing the past and laws of nature include us and move through us, rather than pushing us as if we were not part of the systems these propositions describe. As Charles Hermes says in a recent paper (that I may discuss later this month): “nobody has ever been pushed around by a proposition.”

Indeed, I think the intuitive force of Frankfurt-style cases (FSCs) is that they help illuminate that causal processes can “move through” you appropriately, even if determinism is true (or situations in which, relative to certain contexts, one lacks alternatives). They help us see that rationality is compatible with determinism (that reasons-responsiveness does not require ‘regulative control’). So, John I’m trying to coopt your cooler error theory!

And a second sketch (which I hope to develop more in another post): Manipulation Arguments may work (to convince some people) by pumping a bypassing intuition. When we’re told an agent Manny’s action is fully caused by another intentional agent, it is very easy to think that Manny’s *own reasons and reasoning* is bypassed by the manipulator’s reasons and reasoning. Yes, the cases ask us not to cling to this intuition by ensuring us that Manny maintains (and exercises) his compatibilist capacities. But that stipulation is not easily reconciled with our intuitions about intentional manipulation (and people’s intuition that Manny is unfree drops as the cases drop the reasons and reasoning of the agent or forces determining Manny—as recent studies have shown). Is the best explanation for most people’s *intuition* that Manny lacks FW and MR that, as Pereboom suggests, “his action results from a deterministic causal process that traces back to factors beyond his control,” or is it that his action is ultimately caused by another agent’s reasons and reasoning?

Shall I even try a sketch of how bypassing intuitions creep into the constitutive luck case (as Matt suggests in his comment)? It would begin by pointing out that the way Kip describes it, it sounds like proto-persons are endowed with a fully formed character (e.g., rapist) and then that character pushes them along through life, bypassing any capacities for rational control the person may have to influence that character or its manifestation in action. Yes yes, I know that the capacities for rational control are also endowed as a matter of constitutive luck, but I’m suggesting something about the way our intuitions get primed.

Whew, enough for now!

Hi Eddy,

Let me say that I agree with most of what you say. I agree that presenting things in physical terms, instead of "reason" terms, makes people vulnerable to a bypassing error (and there is data proving that). I agree that many (most?) skeptics/incompatibilists around the world commit a bypassing kind of error - I didn't always realize that, and I learned it through your research, and updated my beliefs accordingly. Also, it might have been to harsh to say that you don't understand your best opponents - as you've reminded me, your bypassing arguments are not necessarily directed to your best opponents.

I could write volumes in response, but here are some further thoughts:

Probably the most telling sentence in your last comment is:

"Kip describes it, it sounds like proto-persons are endowed with a fully formed character (e.g., rapist) and then that character pushes them along through life, bypassing any capacities for rational control the person may have to influence that character or its manifestation in action."

The whole point of the example is that there is no bypassing. To inject bypassing into the example (as you seem to have done here) suggests one of two things:

A. you're simply reading me uncharitably
B. you think about rationality in a way that makes it *impossible* for a serial rapist to be rational

You're a gentleman, so I'll assume that the case is B.

I have long suspected that differences between compatibilists and incompatibilists are due, in many cases, to differences about understanding rationality. After all, compatibilists are so impressed that we can have rationality in a deterministic world. If rationality gives us *a lot*, then this is truly impressive (as you seem to think). If, instead, rationality gives us a little, then this is less impressive.

It is not surprising, then, that Shaun Nichols (who strongly leans incompatibilist, as I read him) defends the view that psychopaths can be rational (see his 2002 article). In other words, he's defending a *thin* view of rationality. As I understand you, Eddy, you're adopting a *thick* view of rationality.

All I can say in response is: it is possible to be perfectly rational and also a serial rapist. It is possible because: a person might have rapist goals, with no countervailing goals, and the person might rationally pursue those goals successfully. In this kind of situation, there is no bypassing, and that is what worries skeptics like Pereboom (or so I think). If you want to further understand your best opponents, I recommend (re)considering Nichols' article, and his arguments why psychopaths can be rational (i.e. without bypassing). If we're wrong that psychopaths can be rational, then I'd like to know why...?

The "bypassing” error theory and alternative definitions of "can", to pick the two most common examples, are sleights of hand designed and implemented irresponsibly by compatibilist professors to distract and inspire unnecessary doubt in students, who, to their annoyance, so immediately (and correctly!) see the power and plausibility of manipulation cases and the Consequence Argument.

Eddy, I think you’re right that the bypassing story is most likely to apply to the “initial intuitions” that get sophisticated incompatibilists going. It’s not a debunking explanation of their explicit arguments (or even various subsequent intuitions) except insofar as it suggests that those are post-hoc rationalizations of the “initial intuitions.”

Here’s an even more sophisticated version of the mistake you end on: rational causal explanations don't get bypassed or excluded by just any physical causal explanations, but specifically by *complete* physical causal explanations. That's an easy transition. And from there someone might work themselves into thinking that indeterminism gets us off the hook if indeterministic causal explanations don't count as complete. Personally, I can't see why they would be any less complete in the relevant sense - the best instances of both don't leave anything *more* to be *explained*. But that seems like one way in which an initial intuition about bypassing of the rational can easily develop into a worry about determinism, and I wouldn’t be surprised if it applies to some even better incompatibilists who’d explicitly profess not to be worried about exclusion. (Are we so sure – given our data on the folk – that it doesn’t fit some proponents of the CA if we replace ‘*completely* caused by’ with ‘*necessary* consequence of’? And wouldn’t worries about bypassing of rational causation – the unique way in which a person herself can bring about an action – help explain the temptation to posit a special type of (agent)-causation?) At this level of sophistication, the bypassing story doesn’t directly apply to the subsequent intuitions and arguments, but as an account of where they came from, it might still be part of a debunking explanation.

Of course, the best opponents won't be making any mistake, ever. They’re like that. And the second-best won’t be making any non-question-begging mistake. But then (as Justin suggests) the best opponents might not really think it's determinism, as such, that has much of anything to do with what's uniquely threatening to MR and FW, either. (Personally, I’ve never been clear on how indeterminism helps with the luck problem. Why would deterministic constitutive luck pose any more of a threat than indeterministic constitutive luck? You just get saddled with the same initial endowment either way. Yes, perhaps we can tell some story about how we can only harness indeterministic causal processes to (re)-shape ourselves, but… in either case, it’s the luck that’s threatening.) And the bypassing story simply doesn’t need (isn’t meant) to explain away “incompatibilist” worries that aren’t about determinism.

Hi Eddy,

When one situates the bypassing claim in the larger cognitive scientific context, I think it becomes clear that it has some more work to do. Your claim seems to be that we easily slip from the intentional stance to the physical stance. But there's lot of work showing that we - children and adults - have the opposite tendency: to utilise the intentional stance even when we know it is not justified. Think of work on hyperactive agency detectors and the like. So how do you reconcile this work with your account? It seems that the intuitions ought to go in the opposite way to the way you want.

Dylan, what philosopher thinks (a) constitutive luck threatens free will and (b) indeterminism would help? I can't think of anyone besides myself who makes a big deal out of constitutive luck, and I emphasise that on this score (actually on all scores) incompatibilism at its best is as good as compatibilism (ie, not very good at all).

The biggest potential problem with your entire by-passing explanatory framework is that I don’t think you ever adequately got at the empirical issue that I suspect is driving both your results and incompatibilist intuitions more generally—namely, the issue of sourcehood or ownership.

For simplicity’s sake, I will use myself as an example (but I will pick up on Kip's earlier theme). As you know, I, like Kip and others, happen to think that determinism threatens free will and responsibility. One reason I think this is that I believe that to be “genuinely” free is to have the sort of contra-causal freedom that nearly everyone, including you, thinks is incompatible with determinism. Just because compatibilists don’t seem to care that something gets lost, metaphysically speaking, if the universe is determined, it doesn’t follow that others shouldn’t find determinism unnerving.

According to a standard incompatibilist line of reasoning, if determinism is true, I am not the proper source of my actions and behaviors. Indeed, in some deeper sense, the more objectively I try to view my actions and behaviors the more they start looking like mere events that are not *mine* in the right sort of way.

There is still a difference, of course, between intentionally shaking my hand and having a spasm in my hand even in a deterministic world. But there is nevertheless a sense in which I am not completely free in either case (even if I am more free in one case than the other!).

So, while I can be free and responsible in some lesser sense in a deterministic universe—which we can define in terms of reasons responsiveness, reactive attitudes, punitive practices, etc.—there is a deeper sense in which everything in my life was *in the cards*.

Let’s imagine, for the sake of argument, that since lots of incompatibilist philosophers seem to share this very intuition, that at least some of the folk might just share it as well (especially those with a dualistic and theological bent). Now if you present these natural incompatibilists with deterministic scenarios and then ask them whether the agent’s conscious decisions “made a difference”, etc., they may say “no.” But it’s not necessarily because they don’t think the agents in the scenarios acted consciously, voluntarily, intentionally, etc., but rather because they don’t think the agents in a deterministic scenario *own* their beliefs and behaviors in the right sort of way. Indeed, I think the way you attempted to operationalize by-passing actually made it more likely that people felt the sourcehood worry (since the bypassing items contain talk of what the agent “ends up being caused to do”).

In short, my suggestion is that worries about sourcehood and not worries about epiphenomenalism are primarily driving both your results and people’s incompatibilist intuitions more generally. From what I can tell, the sourcehood model is no less consistent with the extant data (and might even do a better job of capturing all of the salient data—see below).

This is obviously testable. In the event your model is correct, if you present people with deterministic scenarios and ask them whether the agents acted consciously, intentionally, etc., they should say “no." If my model is correct, people should have no problem saying the agents acted with their folk psychological capacities intact. I suspect the reason your participants responded as they did had more to do with the oddity of the "no effect on what P ends up being caused to do" phrasing of your by-passing items than anything else. If you just ask Ps instead: Did she intentionally (or consciously) x, y, or z? They I predict they will say "yes". But they will still deny that what *happened* to the agent was “up to her” in the right sort of way (however fuzzy this may be).

A further prediction: If you then asked people to explain their answers, the ones who answered along the lines I suggested will then say that the agent couldn’t have done otherwise given the past, the laws, and the state of the universe. They will also be the people who most strongly endorse the libertarian items from Part 2 of our Free Will Inventory (and perhaps even the dualism items from Parts 1 and 2).

In short, I think there's a real possibility that genuine incompatibilist intuitions about agent-causal freedom and sourcehood (i.e., causing without being caused!) may ironically be driving your bypassing results. It's not that people think our conscious wills get bypassed if determinism is true. Rather, it's that my will ceases to be *mine* (somehow) if you could have predicted all of my behavior shortly after the Big Bang with perfect accuracy. I still have the power to behave intentionally, voluntarily, and the like, but there is the lingering sense that luck (i.e., fate?) reigns supreme in a deterministic universe. Or so it seems to me. It's an open question how many people share my view...but I am confident that there are more than you typically let on.

Another related issue that complicates your interpretation of the data is that people have a hard time distinguishing determinism and fatalism. Indeed, as you know, when we were designing our scale, there were several fatalism items that loaded more cleanly and strongly with the items from the determinism subscale than some of the determinism items! This suggests that philosopher's concepts of determinism and fatalism don’t track the folk concepts (which is an issue worth considering and addressing in its own right).

To the extent that people think that to be determined is to be fated, it’s no small wonder that they conclude our actions aren’t “up to us.” It’s not because they don’t think we can be conscious and volitional creatures in a deterministic universe; it’s that they correctly appreciate that if determinism is true, everything is, in some important sense, in the cards.

If I am right, the metaphysical chatter you’re picking up with your bypassing studies is just a downstream ripple being caused by deeper worries about sourcehood (worries which you invite them to have given the way you operationalized bypassing).

p.s. You try to use your error theory to explain away what you take to be the pseudo-incompatibilist intuitions that are allegedly driven by by-passing. Perhaps you could say a bit more about what you count as a genuine incompatibilist intuition. For instance, 80%+ of the 1,500 or so folks we’ve surveyed agreed with an explicit statement of contra-causal free will. Does this not count as a genuine incompatibilist intuition? If not, why not? If so, why isn’t this part of what we ought to be trying to explain?

Hi Neil, that was really just a response to Kip's claim in his first post and that the best incompatibilists are incompatibilists because of constitutive luck. My point was just that I don't think the (in)compatibility of constitutive luck and FW/MR has much to do with the (in)compatibility of determinism and FW/MR, and it's only the latter that Eddy and I have applied the bypassing story to.

Kip, I was responding to your image of your being endowed as a fetus with a character (serial rapist or philanthropist) and because “it was simply given to [you], and became part of [you], [you] became a rapist, or a philanthropist, because of it.” To me that image suggests that it was inevitable that you would become, say, a philanthropist, and to me that is easily understood to mean you would become a philanthropist regardless of how you might reason along the way (or in a way that suggests your reasoning is just rationalizing). I understand you are not explicitly trying to suggest such bypassing, and that when you think about constitutive luck, you presumably aren’t making a bypassing mistake. Call me uncharitable. But that image seems to leave out the crucial role of your *reasoning* being a *difference maker* in what you become (i.e., how you reason along the way is a critical causal condition of whether you become a philanthropist or not). For what it’s worth, I do think rapists can have capacities for rationality, and if they do, and have the opportunity to exercise them at relevant points in their life, then they can be responsible for what they do. Whether psychopaths are rational depends on whether you think they understand moral reasons and whether you think understanding moral reasons is required to be rational. I think they probably have diminished degrees of such understanding. On my view, that should diminish their responsibility (yet my prior post aimed to problematize that conclusion).

Brent, I hope you’re not in charge of who gets the Hemlock for corrupting the youth! (Luckily, you must not think we compatibilists are doing anything ‘irresponsibly’ in the desert sense, right?) I hope to have a post related to debates about ‘can’ but I’ll just say here that I think you have it backwards—it’s incompatibilists who have the “alternate definitions of ‘can’.” Suppose determinism is true: Is it true or false that a coin that landed tails could have landed heads? That a dog who dropped a Frisbee could have caught it? That a competent child that missed a math problem could have gotten it right? Tricky questions—obviously true in the capacities sense of ‘can’ and the ‘in general’ or ‘next time’ senses of ‘can’, but not obviously true in the ‘on this particular occasion’ sense of ‘can’ (obviously false on the ‘holding fixed the entire past and laws’ sense but the question is why we should think that sense ever provides the right context). I think it is true for the ‘on this particular occasion’ sense too on both the most useful and most intuitive analysis of ‘can’. If so, then the incompatibilist needs to explain why the ‘can’ of human choice and action should be analyzed differently.

Neil, I don’t think our bypassing theory suggests or predicts “that we easily slip from the intentional stance to the physical stance.” Rather, we suggest that if you describe determinism (or mechanism) in a way that strongly suggests that there is a complete causal explanation C for action of decision X *and C makes no reference to mental states or processes* (and/or if it suggests that X *has to happen*), then people are susceptible to concluding that mental states or processes do not play a causal role in X. Recall that part of our theory is that the Nichols and Knobe description of determinism (Universe A vs. B) is problematic in ways that prime bypassing. In other descriptions of determinism (and physicalism—as I’ll discuss in next post), and in concrete scenarios that highlight an agent making a decision, bypassing judgments themselves are much lower (and FW and MR judgments are much higher).

I’d have to think more about how the evidence for our hyperactive agency detection comes into play. My initial thought is the real problem it poses for our view is the one that critics (including Thomas) have suggested in various forms from the start: that without doing a lot of work to turn *off* our agential thinking (and concrete cases do not do that), people simply inject any scenario (even deterministic ones) that discusses agents with their understanding of agency, and since these critics believe that our understanding of agency includes libertarian elements, they suggest that most people are assuming such elements even in deterministic cases. Something along these lines seems to be going on in Rose and Nichols’ paper, in a recent chapter by Knobe (I’ll link to in my next post), and in Thomas’ comment.

Dylan and my bypassing picture says some people’s implicit thinking goes:
Determinism --> Bypassing --> no FW/MR

They suggest it instead goes:
Agency --> FW/MR, including libertarian elements, hence ignore determinism (or don’t recognize its inconsistency with libertarian elements)
AND/OR: Determinism --> No agency or decisions (which includes libertarian elements) --> No FW/MR, including libertarian elements --> hence Bypassing

Sorting out which model is correct is an important next step (I’m not yet convinced by Rose and Nichols’ model or evidence, nor by some other papers circulating that challenge Dylan and my model, but I don’t know if those can be discussed.)
In general, the hyperactive agency evidence might help support our critics, except that…

I’m not convinced that people’s understanding of agency includes libertarian elements (in the relevant way). I also think it still counts as an error if people infer from determinism that there are no decisions (as Rose and Nichols use in their model) or if they infer from determinism a sort of fatalism that all sides agree is a mistake (e.g., that what happens has to happen regardless of what we try to do). Here’s the place where my disagreement with Thomas is most significant and where some of our recent evidence is most challenging for my claim that people’s understanding of agency is not libertarian. Here’s a few responses to his thoughtful criticisms (not aiming to be complete at this point):

First, I’ve never disputed that some people have (genuine) incompatibilist intuitions. In our early work, we said it looked like about 1/4 to 1/3 (which seems to line up with the proportion of philosophers who are incompatibilist). Dylan and I are probably over-zealous when we claim it’s only 1 in 10 (but we do think that some of the reported intuitions should not count as “genuine”). I want to stress that, while I have suggested (and believe) that the numbers do matter for the rhetorical point of disputing the long incompatibilist tradition of saying compatibilism is counter-intuitive and for shifting burden of proof onto more metaphysically loaded theories (I want to hear from people about this!), I am more interested in figuring out the underlying psychology and which, if any, intuitions are reliable or informative for philosophical theorizing. Hence, I think it matters which of the models I just laid out is right.

Second, Thomas offers some really interesting ways to test between various models. I’m not sure what people would say about whether “In Universe A, a person does not perform actions intentionally [consciously]” or (in concrete case) whether “Bill does not intentionally [consciously] steal the necklace.” I’d guess that fewer people would agree than agreed to our bypassing questions, though I was surprised by how many agreed to those (note that the second study we ran used the less awkward—but still grammatically problematic!—“In Universe A, what a person wants [believes, a person’s decision] has no effect on what they end up doing.”) I have to think about what our theory should predict about these questions (Dylan?). I think someone could have the bypassing response to determinism but still think people consciously act, in that they are aware of what they are being caused to do (and have no control over).

Third, Thomas and I do have evidence clearly showing that the majority of people agree with statements such as “The fact that we have souls that are distinct from our material bodies is what makes humans unique” and “To have free will is to be able to cause things to happen in the world without at the same time being caused to make those things happen.” Agreement with these statements suggests most people have dualist and libertarian intuitions. I am not convinced. I think they suggest that many people agree with dualist and libertarian-looking *theoretical* claims, for some people perhaps because of their religious upbringing. But I think most people are in fact ‘theory-lite’ about the metaphysics of mind, action, and free will, meaning that their intuitions and, in general our psychological systems governing agency and responsibility attributions, do not involve any commitments to contra-causal or substance dualist metaphysics. I’ll likely discuss this more in the next post, but an analogous case would be that most theists might agree to theoretical claims such as “Our souls are non-physical” or perhaps “God does not exist in time,” but I suspect most would not “apply” those beliefs in relevant cases or practices—e.g., they expect to have something body-like in the afterlife and be able to recognize grandma, and they think God acts within time, etc. I agree that things get tricky when we try to decide which evidence should *count* more in assessing what people believe about the nature of the mind, free will, God, etc.

This is already long and I have to stop now, but let me know if there are any comments to which anyone wants me to respond more fully.

What John said, pretty much.

Jaegwon Kim's exclusion argument is influential, and representative of a strain of thought with deep roots, and can obviously motivate an incompatibilism. And I wouldn't want to call philosophers wrestling with that issue "unsophisticated", either.

So some incompatibilisms, sure. But the CA? PAP? I admit I never thought about the plausibility of NL declining "when L is explicitly understood also to include whatever regularities are instantiated in rational action." I guess I never gave NL much thought when the blatantly unacceptable transfer principles were there to distract me. Why, pursuing your point would be like going after Al Capone for tax evasion instead of murder -- on second thought, maybe not such a bad idea.

But Humean laws? That's cheating, in my book: a "determinism" so alien to ordinary thought about causality as to disconnect with any intuitions that causality may conflict with FW. Give me Carl Hoefer style laws (bidirectional and ontologically non-derivative) any day. The folk's conception of causality may need to be knocked down a peg, but not obliterated; that takes all the fun out.

Wait, the causal exclusion argument? I can see how it might motivate worries about free will, but how can it motivate incompatibilism? If supervenience and reductive physicalism is true, indeterminism won't help; if one or other is false, there is no threat whether or not determinism is true. Right? Am I missing something?

Hey Thomas, I wouldn’t be surprised if people do think that humans have contra-causal and “libertarian” powers in the actual world. With Eddy, I just doubt that they take those to be required for free will, moral responsibility, and the things those are most are importantly connected to. Even if people think the compatibilist-friendly variety is a “lesser sense” of freedom than what we happen to enjoy (though I’m not convinced many “fuller” senses are coherent), they can still think it’d be sufficient for the truth of our ascriptions, associated social practices, the appropriateness of the reactive attitudes, and the like. In other words, what we get with contra-causal or libertarian powers is a sort of “super freedom” (sort of like being supererogatory) – something that might be kinda neat, but isn’t necessary for the things we really care about. (I like the ambiguity of ‘super’ between being over-and-above and not relevant to responsibility at all, and being extra, in the sense that it might make us a bit – but not much! – more responsible than enjoyment of just the compatibilist capacities.) I suspect a lot of non-philosophers do think we happen to have “super-freedom,” but that if we hadn’t (or, even indicatively, if turns out that we don’t), we’d still be perfectly responsible, blameworthy, etc. (Maybe super-freedom is a sort of God-given, honorific bonus. Maybe it’s even required for something else, though I’m not sure what.) In that case, even if those count as libertarian, they’re not natural incompatibilist intuitions - they’re intuitions about what exists, not about what’s co-possible with what. Using hypothetical cases separates those issues more clearly, whereas some of the items on your free will scale may be prompting answers about existence, rather than co-possibility…

I really like the idea of asking questions about acting consciously, voluntarily, and intentionally, and I guess I would predict that we’d get fewer mistakes with those than we did with the bypassing questions that we used. But I bet we’d get higher bypassing scores again as long as we continued to ask about causation. I’d predict more agreement with:

“Alfred’s action was conscious.” (“Alfred acted in the way he did consciously.”)
“Alfred’s action was intentional.” (“Alfred acted in the way he did intentionally.”)

but less agreement (higher bypassing scores) with:

“Alfred’s intentions caused him to act as he did.”
“Alfred’s conscious mental states caused him to act as he did.”
“Alfred’s conscious intentions…”

I like Eddy’s story about why people might agree more that “Alfred’s action was conscious.” Someone can be consciously aware of what she’s doing without having any control over it. Explaining higher scores for “Alfred’s action was intentional” (if we got them) would be trickier. One possibility that’s consistent with our story, especially if fewer people agreed that “Alfred’s intentions caused him to act as he did”: maybe some people think Alfred’s action was intentional in the sense that it was caused by *some* of his mental states (perhaps at the most extreme, like a sort of reflex), and so counts as intentional, but that the important rationally-connected ones, like his conscious intentions, beliefs, desires, and decisions, still got bypassed, explaining why he's nonetheless not responsible. (Of course, one could test that explanation, too…)


It is very curious to me that, even when I try very hard to avoid bypassing in my example, you still read bypassing into it. My original example included these statements:

* "At one point, we were not fully people, and we had zero or minimum goals or desires."
* "This initial endowment, of initial goals and desires, did *not* bypass earlier goals and desires."
* "The initial endowment did not bypass any other goals or desires, because there were no other goals or desires, prior to the initial endowment - that's why it's the initial endowment."
* "Nevertheless, even without bypassing, this initial endowment is problematic, because it is a matter of constitutive luck."
* "My poor proto-self had no mechanism of self-control or self-analysis to accept or reject this initial endowment [i.e. no mechanism to "bypass"]

Despite *all* of that, you still read the example to include bypassing! Isn't that strange?

Again, I'm highlighting this not to criticize your bypassing theory directly. As I said before, I agree that it explains a lot of why many incompatibilists are incompatibilists for bad reasons. It wouldn't surprise me at all, for example, if my mother *did* think that determinism entailed bypassing and, therefore, rejected compatibilism.

Still, I'm highlighting this issue because it is striking, to me, how much you and your best opponents (Pereboom, Levy, Galen Strawson, etc.) seem to be talking past each other. It seems to me that all of these well-respected opponents do not commit the bypassing error. Moreover, even when someone (like me) bends over backwards trying to explain that a thought experiment does not involve bypassing, you will inject bypassing into it!

Suppose, for example, I am right and that you project a bypassing error onto your best opponents, even when they don't commit it. Does this prove you wrong, and them right? No. You could be right for other reasons, and they could be wrong for other things. Nevertheless, if I am right about this, it is fascinating to think there is such an empathy gap between you and the others. Of course you don't agree with each other - you have (collectively) failed to understand each other.

This brings me to my last point: in your most recent comment, you try to explain why you read bypassing into my example (despite my repeated warnings not to). Then you acknowledge that, in principle, the example can be read without bypassing. Well, in that case, can you see how someone like me or Levy would be worried about constitutive luck? After all, it was not the rational rapist's fault that, as his body/brain/DNA was being fashioned, he ended up being a rational rapist. Whether he became a rational rapist, or a rational philanthropist, was a matter of luck. Do you see that potential threat? And, even if *you* think it is a mistake to regard that (i.e. constitutive luck) as a threat, do you see how others could (mistakenly) think it is a threat - *without* committing the more egregious error about bypassing?

Eddy and Dylan,

If you presented Ps with a scenario involving an agent in a determined universe and then present them with statements like "Alfred's action was conscious" and "Alfred's action was intentional," I predict very few Ps would disagree.

Let's assume I am right. Then why think that the Ps are making the bypassing error? If I ran the same paradigm you ran along with my revised statements and found that Ps are perfectly happy to ascribe conscious, intentional behavior to agents, then why think they are making the bypassing error? What is it precisely you're suggesting they think is being bypassed? It can't be consciousness, but then your suggestion that worries about epiphenomenalism explain the results doesn't make sense.

I also predict that if you give Ps determined agents who end up acting involuntarily, Ps will track this salient folk psychological difference even across deterministic scenarios.

If either of these predictions panned out, then it seems to me this would not be good for your by-passing thesis.

Dylan suggests that perhaps you could use items like the following to continue to elicit the bypassing error:

“Alfred’s intentions caused him to act as he did”

“Alfred’s conscious mental states caused him to act as he did.”

But why think these represent a better way at getting at whether Ps think agents in deterministic universes act consciously, intentionally, voluntarily, etc.?

Here again, wording it this way is not only odd, but it invites Ps to make the very mistake you claim to have identified. But why invite them to make a mistake when (a) ordinary language will do the trick, and (b) you are going to use the mistake to discount their intuitions?

p.s. Eddy, I requested that you provide an account of what you'll count as a genuine incompatibilist intuition. You still haven't provided such an account. Instead, you gestured at your "theory-lite" hypothesis. But that won't do. If you're going to be in the business of discounting and dismissing intuitions as illegitimate or confused, you need to provide a clear account of what you'll accept as evidence. Indeed, until you provide a clear statement of what counts, you won't even be able to test your own "theory-lite" hypothesis. So, perhaps you could start there?


Despite your ingenuity (and persistence), I just don't see the bypassing intuition as playing a major role in explaining the pull of the Consequence Argument. Rather, the pull is from the fixity of the past (and laws); the past is over and done with, and thus we have no "handle" on it. That's the intuitive engine behind the Consequence Argument. Similarly, the idea is that we have no control over the laws of nature. It is not that they push us in ways that bypass our motivational systems; their moving *through* our motivational systems is even more problematic. This would imply that stuff that is out of our control is "inside" us in a deep way!

Perhaps the analogy with the Divine Foreknowledge Argument will help. I have for a long time been struck by the parallels between the CA and the DFA. They are both driven by the engine of the Fixity of the Past. Of course, there are differences. But it seems to me that they are fundamentally similar. Assuming that we take it that the laws are out of our control (admittedly, a contentious assumption), then it seems to me that one should accept the CA iff one accespts the DFA. And the appeal of the DFA can't be explained in terms of a bypassing error, can it? That is, God's foreknowledge doesn't push or cause us at all, and thus doesn't bypass our normative orientations and motivations systems, does it?

Like you, Eddy, I'm a compatibilist about causal determinism and moral responsibility. And I agree that *some* folk and *some* sophisticated incompatibilists about determinism and freedom to do otherwise are indeed moved to some extent by a bypassing mistake (or perhaps more charitably, a worry about exclusion). And I do agree that some manipulation scenarios involve bypassing, and further that it would (thus) be an error to extrapolate from "no responsibility" in these scenarios to "no responsibility under determinism". Where I still disagree with you, despite your persistent and admirable and resourceful efforts, is about the CA. I'm inclined to accept it; at least I take it very seriously. I don't think there is an obvious fallacy or blunder, or an subtle fallacy or blunder. I don't think it is *obviously* sound, though, and I'm moved to some extent by the arguments of Lewis, Perry, and others. But of course I don't think that freedom to do otherwise is necessary for moral responsibility, so I don't think one has to take a stand about the CA.

Out of curiosity: why do you think it is so important to establish some sort of defect in the CA? Why not say that reasonable people can disagree, the issues are subtle and complicated, and remain agnostic--here I'm actually with Al Mele. Then one can go the "actual-sequence" route for moral responsibility. And the case for incompatibilism now becomes harder to establish. (I'm not saying "impossible to establish": going all the way back to my 1982 article in the Jnl of Philosophy, "Responsibility and Control", I pointed out that even if moral responsibility does not require freedom to do otherwise, it doesn't follow that (say) casual determinism is compatible with moral responsibility, since causal determinism "flows through the actual sequence". Thus I mapped out the possibility of Source Incompatibilism, although of course others, especially Michael McKenna and Derk Pereboom, have done much more to develop and explore (and, in Pereboom's case) defend the doctrine.

Thomas, I think participants are likely to read “Alfred acted intentionally” in contrast to “Alfred acting accidentally” and so want to agree – if they do, which I don’t think I quite share your predictions about – on that basis. (I think they’ll tend to agree with that statement more than they do with the originals, but I’m not sure ratings’ll, go, say, above the midpoint. I don’t know.) So, I’d also expect people to keep track of the accidental/intentional distinction across deterministic cases.

Even if they’re making more subtle distinctions, people might think that you can act intentionally without having an intention, and take the latter to be more important for moral responsibility and free will than the former. Maybe they see Alfred’s action (or “activity,” in Velleman’s sense) as intentional but not as a “full exercise of agency,” compatible with giving lower MR and FW ratings. And it’s of course plausible that the distinction between those types of “action” should be cashed out in terms of which mental states were properly operative in their production (e.g., conscious, deliberative, rational mental states vs. unconscious associative impulses, desires, and at the extreme nearly brute neurological states). If determinism is *only* confused with the bypassing of higher-level, “genuinely” rational causation, and people don’t confuse it with bypassing of lower-level types of mental causation (that can still produce more-than-accidental actions), that’d still explain the full pattern of (hypothetically expanded) results. Knobe’s new paper actually provides some preliminary evidence for exactly that: people don’t confuse determinism with bypassing of physical processes *or* the non-rational causation of things like facial expressions by emotions. (Don’t we smile intentionally in such situations, or at least non-accidentally, but without the intention to smile? But aren't we not very responsible for doing so?) I’d actually predict that MR/FW judgments in response to deterministically-caused-but-otherwise-intentional-looking actions would look a lot like those for actions produced at more associative, lower-looking levels like that.

“Why [then] think they’re making the bypassing error?” Because of our results! I think the above would provide a fine explanation *if we did* find considerable agreement with “Alfred’s action was intentional” in this (still hypothetical) study, but I’m not convinced of your alternative explanation of our (actual) results: that people do strongly agree that “Frank’s decisions/beliefs/desires had no effect on what he ended up doing” just because of an artifact of the language we used. Is it really that grammatically odd? (And won’t you still need some story if they agree that “Alfred’s action was intentional” but also that “Alfred’s intentions did *not* have any effect on what he did"?) What about “Frank did A because he deliberated about and decided to do A”? That’s not odd, right? I think that might be even better. Admittedly, we didn’t phrase things that way, but I’d expect comparable numbers in response to that as we got with the original questions, and due to the same type of thought: “No, Alfred didn’t A because he deliberated and decided to A; you already told me he did A *because* of the past and the laws of nature (i.e., not A)."

Actually, John, couldn’t it be that the CA turns on something like the thought just voiced, rather than the fixity issue? It seems easier to me to make the bypassing mistake when thinking about explanation, and not directly about causation, at least. Isn’t it possible that the intuitive force of the “necessary consequence” bit is that people tend to think a complete explanation of A in terms of X and Y precludes Z from doing any further explanatory work (a general intuition that might persist even when we’re clear that X and Y only really explain A because they bring about Z, which in turn brings about A)? Anyway, I think a lot (more) of the pull behind the CA comes from philosophers elaborating a peripheral (not moral-responsibility relevant (but larger issues here…)) element of the folk concepts (in the ways suggested in my last comment). They pick up on something that’s there, but mistake it for something central. Indeed, it seems like the easiest way to get CA-like arguments rolling is precisely by expanding the set of facts that we hold fixed when evaluating whether one “can/has the ability” to do something beyond all recognition – way beyond the set of facts we ever hold fixed when actually using the concepts in responsibility-relevant settings. (We thereby artificially restrict the number of counterfactual worlds where Alfred acts otherwise that count/decrease the potential candidates that could make “Alfred could have acted otherwise” true; a restriction that moves our conversation off into the periphery, at best, of our (MR and FW-relevant) concepts of “can” and "ability.")


For starters, if you truly are interested first and foremost in the descriptive project of getting a handle on people’s baseline beliefs and attitudes and related concepts, then it might be more helpful if you spend less time trying to distinguish the *real* and *genuine* intuitions from the phonies (whatever that means). Instead, you spend a lot of time and energy trying to explain away otherwise interesting and important empirical findings that happen not to fit your compatibilist narrative.

For instance, two general types of data have been collected thus far—namely, responses to primes and responses to explicit statements. As you know, while some deterministic primes elicit mostly compatibilist responses, others elicit mostly incompatibilist responses. The findings on this front—which happen to be your focus—are mixed. But when you look instead at people’s responses to explicit statements, people give more thoroughly incompatibilist responses.

You’ve known this for years, yet it rarely gets a mention in your work. Why isn’t this part of the data you’re trying to explain away with your error theories and the like? When you do mention the issue in passing (either in print or in person), you often seem to follow Mele in drawing a folk intuition vs. folk theory distinction. I am happy to grant the intuition vs. belief distinction for the sake of argument. But why think that people’s intuitions are more philosophically relevant that their explicit philosophical beliefs and theories? From the purely empirical standpoint, data about both intuitions and theories provide us with useful information concerning the role that the beliefs and attitudes about free will play in our daily lives. So, it would be nice to hear more from you concerning why you focus on one type of data and largely ignore the other.

At this point, it is clear that most people have explicitly incompatibilist libertarian *beliefs* about free will. It’s also clear that for many people, these beliefs are positively correlated with their beliefs about the soul and negatively correlated with beliefs about determinism. What error theory do you have to explain away these findings? As far as I can tell, you don’t have one. Instead, you have your “theory-lite” hypothesis. But this has never seemed to me like a promising avenue for you to take.

Here is what you say:

“I think they suggest that many people agree with dualist and libertarian-looking *theoretical* claims, for some people perhaps because of their religious upbringing. But I think most people are in fact ‘theory-lite’ about the metaphysics of mind, action, and free will, meaning that their intuitions and, in general our psychological systems governing agency and responsibility attributions, do not involve any commitments to contra-causal or substance dualist metaphysics.”

Take, for instance, our aforementioned explicit statement of contra-causal freedom. On the one hand, we found that this statement elicited overwhelming agreement. On the other hand, we found that these specific beliefs about free will—which most people have—are negatively correlated with their beliefs about determinism and positively correlated with their beliefs about the soul (just as most philosophers would have predicted). Yet, in the face of this mounting evidence, you nevertheless proclaim that people’s intuitions and psychological systems do not involve any incompatibilist commitments. How would you test this? After all, if I remind you of the data we already have, you will say that even though folk beliefs and theories appear to be incompatibilist, these commitments and attitudes aren’t “deep” enough to influence our thoughts and behaviors or don’t include “libertarian elements (in the relevant way).”

Here again, it would be illustrative if you could clearly explain what you would count as a relevant, genuine, incompatibilist intuition (or belief) about free will. For instance, here is what I would count as clear evidence for a compatibilist belief or intuition (at least in the philosophical sense):

Ps are presented with a deterministic scenario wherein agents don’t have the unconditional ability to do otherwise. Ps who appreciate that the agent’s couldn’t have done otherwise nevertheless think the agent is free and responsible.

As you know, one lingering problem with our earlier work on beliefs about free will is that it was really difficult to ensure that Ps appreciated that agent’s genuinely couldn’t have done otherwise (which is presumably why so many Ps missed our manipulation check). In fact, our team never did reach an agreement on how best to interpret our findings on this front or how to ensure that people were understanding and internalizing the deterministic scenario as we intended. But unless and until you can show that people are willing to give up on contra-causal freedom while at the same time holding on to compatibilist control and responsibility, you haven’t shown what you’ve claimed to have shown. Or so I have been saying for the better part of ten years now.

There is an additional empirical difficulty with your “theory-lite” hypothesis—namely, that it’s hard to see how to distinguish it from another rival hypothesis—namely, the flip flopping hypothesis. According to the later, people do have a deep, genuine, metaphysical commitment to libertarian free will and responsibility. However, like Inwagen, when forced to choose between their incombatibilism and their freedom, they keep a revised version of the latter. Let’s assume that at least some of the Ps in our previous studies were flip-floppers. Do they count as compatibilists on your view? You seem to think so, but I am not so sure. Instead, they are default incompatibilists who have flexible enough beliefs and attitude to tolerate revision. If this is right, then some of the respondents you’re counting as compatibilists are actually incompatibilists-in-transition (or conditional compatibilists). These are incompatibilists who would give up incompatibilism before they give up free will!

The final related worry about your “theory-lite” hypothesis is that it enables you to shield your view from any conflicting data. After all, you can always shrug off the data by saying the recorded intuitions, beliefs, and theories aren’t “deep enough” to be reflective of people’s deep compatibilist core. Instead, you’ll just say that what we’re picking up is surface incompatibilist chatter. The deep and important compatibilist stuff, on the other hand, is safe and secure high up on compatibilist ground. But why think this? Or, more accurately, what evidence would you accept that people are actually both incompatibilist and “theory-heavy”?


You say:

“Why [then] think they’re making the bypassing error?” Because of our results! I think the above would provide a fine explanation *if we did* find considerable agreement with “Alfred’s action was intentional” in this (still hypothetical) study, but I’m not convinced of your alternative explanation of our (actual) results: that people do strongly agree that “Frank’s decisions/beliefs/desires had no effect on what he ended up doing” just because of an artifact of the language we used. Is it really that grammatically odd? (And won’t you still need some story if they agree that “Alfred’s action was intentional” but also that “Alfred’s intentions did *not* have any effect on what he did"?) What about “Frank did A because he deliberated about and decided to do A”? That’s not odd, right? I think that might be even better. Admittedly, we didn’t phrase things that way, but I’d expect comparable numbers in response to that as we got with the original questions, and due to the same type of thought: “No, Alfred didn’t A because he deliberated and decided to A; you already told me he did A *because* of the past and the laws of nature (i.e., not A)."

A few things:

First, yes, your original wording is positively bizarre. What I want to know is why you chose that wording when straightforward questions about conscious action would have worked just as well? If the hypothesis was that deterministic scenarios elicit some people to mistakenly assume that agents don’t have conscious, intentional control over their behavior, why not just ask about that directly? Why chose instead the “end up being caused” language? Clearly you made an explicit decision concerning how to word these items and so I am hoping you can shed some light on why you thought the odd wording you selected was preferable.

Second, if, as you predict, Ps agree with statements like “A intentionally x-ed” while at the same time agreeing with statements like “A’s intentions did *not* have any effect on what he did," why think this supports your version of the by-passing hypothesis? At this point, it would start looking like it’s not conscious and intentional agency that people find in conflict with determinism. Rather, it’s the deep, homuncular, agent-causal self that is being bypassed. On this view, while I may still act either intentionally or unintentionally in a deterministic universe, I don’t ever have the unconditional ability to do otherwise—which means “I” (i.e., the dualistic self) can’t cause things to happen without being caused to cause those things to happen. In short, people correctly take determinism to imply a sort of cosmic causation that is incompatible with my unconditional power to do otherwise.

But if this is right, it’s not the by-passing you were after. Instead, this would seem to me to support a rival by-passing hypothesis—namely, the hyperactive self model by Knobe. After all, the difference between the two views (as far as I can tell) is that while Knobe and colleagues think that issues revolving around the homuncular self drive our free will beliefs, you think it’s issues at the folk psychological level. My suggested experiment is a way of testing between these two rival theories.

Notice, too, that if it turns out Knobe’s by-passing model is the right model, then this would undermine your error theory. After all, the reason your error theory works (at least in principle) is that everyone—compatibilists and incompatibilists alike—can agree that determinism does not entail epiphenomenalism. So, when you provide purported evidence that people slide from the former to the latter, this enables you to explain the intuitions away. But if I am right, people are not inferring epiphenomenalism from determinism. Instead, they correctly (if unclearly) understand that determinism undermines the unconditional ability to do otherwise. Consequently, it undermines the freedom of my deep self—which has the executive power to fully and independently legislate what I do. While I can still act consciously, etc., what I can’t do is act differently even if everything had been the same. But if this is the worry that undergirds P’s responses, then there is no error theory to be had. After all, everyone agrees that determinism is incompatible with the unconditional ability to do otherwise (just as everyone agrees that determinism doesn’t entail epiphenomenalism. The issue is whether people ordinarily think free and responsible agency require the unconditional power to do otherwise (in addition to the basic suite of folk psychological capacities).

My closing question for you and Eddy both is: What would you count as evidence that a person's default metaphysical beliefs and attitudes about free will revolve around the belief in the unconditional power to do otherwise? Once you've spelled that out, I will get to work. Indeed, I already have a draft of the study we've been discussing uploaded to Qualtrics. Hopefully, I can collect some pilot data next week. But before I collect and share any data, I want you and Eddy on the empirical hook!

p.s. Here is an excerpt from a draft post I wrote last summer but never published. It sheds a little more light on the work Eddy and I have done on the Free Will Inventory. Some of you might find the results relevant to the discussion at hand:

“For the past two years, I have been working with Eddy Nahmias, Chandra Sripada, and others to develop a new psychometric tool—called the Free Will Inventory—for measuring people’s beliefs about free will, responsibility, determinism, and dualism. During this time, we have collected data concerning the intuitions of more than 1,500 general population participants through Qualtrics’ panelist recruitment service. These participants were not only gender-balanced, but they tended to be older, more conservative, and more religious than the undergraduates who are usually participants in the studies run by psychologists and experimental philosophers. Our main reason for relying on general population participants rather than undergraduates is that we thought their intuitions would be both more diverse and more representative of the folk than the participants used in previous studies.

The Free Will Inventory presents people with a series of 29 statements and asks them to state their agreement using a 7 point Likert scale ranging from strongly disagree (1) to strongly agree (7). For present purposes, I am not going to discuss all of our results. Instead, I am going to (a) limit my attention to our most recent study—which involved 330 general population Qualtrics panelists, and (b) limit my attention to the statements that are relevant to the debate at hand about folk dualism and folk libertarianism. Moreover, for ease of presentation, I am lumping together those who strongly disagree, disagree, and somewhat disagree under the heading “disagree” and I am lumping together those who strongly agree, agree, and somewhat agree under the general heading “agree.”

Here are participants’ responses to some of the items from the Free Will Inventory that are especially germane to our present discussion:

• “The fact that we have souls that are distinct from our material bodies is what makes humans unique”: 11% disagree; 16% neither agree nor disagree; 73% agree (mean=5.41; standard deviation=1.64).
• “Human action can only be understood in terms of our souls and minds and not just in terms of our brains”: 21% disagree; 18% neither agree nor disagree; 61% agree (mean=4.82; standard deviation=1.62).
• “Each person has a non-physical essence that makes that person unique”: 6% disagree; 13% neither agree nor disagree; 81% agree (mean=5.63; standard deviation=1.29).
• “The human mind cannot simply be reduced to the brain”: 14% disagree; 18% neither agree nor disagree; 68% agree (mean=5.13; standard deviation=1.57).
• “The human mind is more than just a complicated biological machine”: 6% disagree; 9% neither agree nor disagree; 85% agree (mean=5.84; standard deviation=1.33).
• “Free will is the ability to make different choices even if everything leading up to one’s choice (e.g., the past, the situation, and their desires, beliefs, etc.) was exactly the same”: 6% disagree; 15% neither agree nor disagree; 79% agree (mean=5.44; standard deviation=1.30).
• “To have free will is to be able to cause things to happen in the world without at the same time being caused to make those things happen.”: 17% disagree; 36% neither agree nor disagree; 46% agree (mean=4.53; standard deviation=1.45).

In short, the majority of participants in this study (and in several of our previous studies) appear at least on the surface to endorse precisely the kinds of dualistic, anti-reductionistic, and libertarian views attributed to the folk by the scientific skeptics. On the one hand, the majority of participants agree that we have souls and minds that cannot be fully explained by or reduced to the brain. On the other hand, the majority of participants agree that we have agent causal powers and the unconditional ability to do otherwise.
At first blush, these findings suggest that those who criticize the scientific skeptics on the grounds that their claims about folk dualism and folk libertarianism are mistaken have more work to do. However, upon closer inspection, things are perhaps unsurprisingly a bit more complicated. Consider, for instance, participants’ responses to the following two statements:

• “If it turned out that people lacked non-physical (or immaterial) souls, then they would lack free will”: 36% disagree; 32% neither agree nor disagree; 30% agree (mean=3.82; standard deviation=1.68).
• “If it turned out that people lacked non-physical (or immaterial) souls, then they would lack moral responsibility”: 38% disagree; 28% neither agree nor disagree; 35% agree (mean=3.82; standard deviation=1.68).

Here the results are admittedly pretty mixed. In each case, we find a roughly even split between those who disagree, those who agree, and those who neither agree nor disagree. So, while the responses to the earlier items suggest that most people endorse dualism and anti-reduction and adopt a libertarian conception of the ability to do otherwise, the responses to these latter two items suggest that there is consistently broad disagreement between participants when it comes to the relationship between free will, responsibility, and the immaterial soul. Further complicating matters is the fact that most participants also state that we can be free and responsible in the face of scientific advances:

• “People could have free will even if scientists discovered all of the laws that govern human behavior”: 7% disagree; 20% neither agree nor disagree; 73% agree (mean=5.39; standard deviation=1.98).

So, while our latest findings do provide some support for the skeptics’ claims concerning folk dualism and folk libertarianism, they also suggest that people are less worried about scientific threats to free will and responsibility than the skeptics have assumed.

There are at least two lessons to be drawn from these mixed results: First, the scientific skeptics have rushed to judgment when it comes to how people ordinarily think about both the relationship between free will and the soul and the relationship between free will and science. Second, Mele, Nahmias, and others have yet to establish that most people are compatibilists rather than dualists and libertarians. Resolving the debate between these two camps is a task for another day."

Thanks Eddy for laying out some sophisticated arguments against sophisticated incompatibilist intuitions!

I think there is something right about your claim that a bypassing mistake plays a role in the initial appeal of the Consequence Argument. Now, John is right to point out that there much more to the CA than this. The CA also has non-bypassing leeway themes and sourcehood themes going on—it is a complex argument. But bypassing does, I think, play a *limited* role in its initial appeal. Here is a quick way to bring this out. Here are certain key premises of the CA.

S cannot change: The state of the universe in the remote past is X
S cannot change: X and the laws of nature jointly entail that S does A
S cannot change: S does A

We usually substitute a simple action in for A, say A1 = “S votes for Clinton”. Let’s now instead substitute a complex sequence that characterizes compatibilist agency. For example, let A2 = “S mentally represents various options and evaluates the reasons for undertaking these options, and then S selects that option that is most worth doing.” (I hope no one is skeptical that A2 can be realized in deterministic universe. I think surely it can.)

I think that the intuitive force of the argument changes when switch from A1 to A2—the A2 version of the argument doesn’t have much bite against compatibilism. The best explanation for the change in intuitive force of the argument is that A2 blocks a bypassing interpretation of the CA. That is, A2 forces you to recognize that the CA does not imply the agent’s compatibilist reasoning faculties are being bypassed.

Let me be clear that I reject the CA for multiple other principled reasons (rule beta is flawed, abilities should be given a condition reading, etc.). But the CA can be flawed for multiple reasons. *One* of the problems with the CA is that it generates part of its intuitive force by taking advantage of a (largely tacit) bypassing mistake.

The comments to this entry are closed.

Books about Agency

3QD Prize 2014: Marcus Arvan