What can 'folk philosophical' surveys show?

I started writing a response to Matt Weiner's response to my comment on Neil's response to Eddy's 'testing free will' post, but then I thought, 'hey, I'm an author here -- I don't have to just respond in a comment, but I can make a post. So, here goes.

The question that Neil raised was: why should we care about studies that put the folk into the armchair? If we think philosophers' armchair intuitions are worthless, ought we not think the same about those of untrained civilians'? (I'm paraphrasing here, so Neil, please correct me if I've misinterpreted you.) I offered a quick response to that, which Matt made trouble for, so let me try to give a non-quick version of that response:

One thing that one might be trying to do, in undertaking a survey of the folk's intuitions, is to undermine current analytic philosophical reliance on intuitions. I take it that that was what Shaun Nichols & Steve Stich & I were trying to do in our "Normativity & Epistemic Intuitions" paper -- we weren't trying to build any epistemic theory on the basis of the intuitions we surveyed, but only to show that the method of relying on intuitions has issues of variability & inconsistency that it has not even begun to reckon with. Now, if one is trying to use surveys to attack the armchair, then you want to consider whether your subjects are understanding your materials appropriately, and intuiting on the case with the understanding that you want them to have. Because if group A intuits that p while considering a case, and group B intuits that not-p while considering that case, you don't really have a conflict of intuitions unless you think that A and B have essentially similar understandings of the case in question. E.g., if you give a clairvoyance case, and members of A are just antecendently unlikely to think of clairvoyance as remotely reliable, whereas those in B consider clairvoyance to be a basic part of their cognitive apparatus, it will fail to be metaphilosophically interesting that A's withhold knowledge in some clairvoyance-involving cases in which B's attribute knowledge. The more esoteric & complicated your materials, the more that this worry comes to the fore. (The worry does double back somewhat on the fan of intuitions, however: the more esoteric & complicated the intuition needed to make your point, the less confidence you ought to have in the point in the first place!)

Now, that's just one thing that one might try to do with 'folk philosophical' surveys, and we see that that primarily destructive metaphilosphical goal entails some methodological considerations. But what else can one aim to achieve with such surveys? Can the data of such surveys themselves not simply displace, but in fact replace philosophers' intuitions? I can see how that might make sense if one is pursuing a more experimentally-respectable version of Jackson-style conceptual analysis. If the facts you're interested in about KNOWLEDGE or FREE WILL are the facts about what the folks think using concepts like KNOWLEDGE and FREE WILL -- then of course surveys of the folk will have some relevance. I think Neil is right, though, that we'd be better off attending less to highfalutin & artificial folk intuitions, and more to the everyday folk practices involving & surrounding such concepts. Intuitions about fairly boring, non-esoteric sorts of cases would be generally fine, but as soon as you start asking people to evaluate cases with content far removed from their ordinary experiences, you have to wonder what is really driving their answers.

So I've tried to suggest two theoretical goals that can make at least some use of these surveys:
--the destructive project of metaphilosophical anti-intuitionism (and, I suppose, the project of trying to rebut this project!);
--the descriptive project of folk philosophy (on a par with psychologists' descriptive studies of folk psychology, folk physics, etc.).

What are some other philosophical projects for which one might imagine such survey evidence being directly relevant? (Note that there may be lots of projects for which it is indirectly relevant, in that the surveys give evidence for some psychological theories over others, and different philosphical theories may have different presuppositions with regard to those psychological theories -- cf. the recent thread on automaticity and moral psychology. But I'm wondering here about when survey data can speak fairly straightforwardly to a philosophical concern.)

Comments

[Ron Mallon]

Interesting question... It seems to me like lots of philosophical arguments are empirically "exposed," but the issues would not be easy to address with the sorts of surveys that some of us have used so far. I have in mind arguments that connect folk theories with behavior, where the empirical issues are complicated enough to require rather substantial cross-disciplinary interaction between philosophers and psychologists.

Here are two examples from ethics and applied ethics, wherein arguments progress on the basis of assumptions about the impacts of a certain way of thinking or talking:
(1) In meta-ethics, some people worry that anti-realism about ethical values, at least if widely accepted, will send the world to "hell in a handbasket" (where does this expression come from, anyway?). But there seems to be a substantial empirical argument here, perhaps one that goes like this:
(i) people are intuitive ethical realists
(ii) the endorsement of real ethical principles explains real ethical behavior
(iii) the undermining of ethical realism will undermine real ethical behavior.

Needless to say, there is substantial empirical content here. (i) depends on folk ethical 'intuitions', and (ii) endorses a substantive theory connecting these intuitions with behaviors.

[Incidentally, it looks to me like surveys about ethical intuitions with regard to (i) fall simultaneously under both of your categories (destructive and descriptive). The destructive part occurs if the folk don't match the ethical realists assumptions about them. The descriptive part is important, because we might want to know how a new theory or belief might relate to a widely held folk theory.]

Here's a second case:

(2) In contemporary race theory, various arguments depend on whether the practice of labeling and thinking about people by race can be dissociated from more robust racialist and racist thinking. If the two *can* be dissociated, you might think that "race" talk is (or could be made to be) relatively innocuous, while if the two cannot be dissociated, you might think that "race" talk should be discarded ASAP. Unfortunately, while there is a interesting psychological work on this subject, it has not been engaged much (to my knowledge) in the context of the concerns of race theory. Instead, the arguments go on relying on (what look to me like) armchair assumptions about how the folk think and could be made to think about race.

Both of these cases show, I think, that there is a real opportunity and need for philosophy to engage psychology in a way that goes beyond the direct links between survey data and philosopical assumptions about the folk. The best existing example of this sort of engagement, I think, is Doris's work on the virtues.

Of course, there is still lots of room for surveys (of the sort we've both employed) to directly impact philosophical issues. But I worry that this is because philosophers have been particularly unsophisticated in their judgments about the folk. The optimist in me thinks people will eventually get the picture and stop doing this, leaving such surveys only with descriptive interest. In contrast, I think lasting work in "experimental philosophy" will, like Doris's work does, try to connect up the descriptive project with behavior and explore its philosophical implications in a way that goes beyond what you call "direct" connections.

[Neil]

People's folk physical intuitions can be quite misleading, but it does not follow that we ought to disregard the physical intuitions of those whose intuitions are in something approaching reflective equilibrium. Einstein, for instance, explicitly said that physicists needed to have a 'feel' for their work. Or think of Kuhn, and of how science education is aimed at getting to see the trainee to see things in the right Gestalt. Here's the heretical conclusion (from an experimental philosophy point of view)): who cares what the folk think? Finally, we tend to underestimate the extent to which we are concerned with terms of art, not with getting folk concepts right. I'm not convinced, for instance, that 'knowledge' and 'belief' have the same referent in epistemology as they do in ordinary language. The first philosophy book I ever read in full was from a different tradition: Sartre's *Being and Nothingness*. There he writes that to say that one believes something is to say that one doubts it, because if you did not doubt it you would say you knew it. That seemed to me right when I was 20, and it has never been criticised in the literature on Sartre from within his own tradition. So even across the Western philosophical tradition there is ambiguity as to reference. How can we be sure there is no equivocation in the survey of Gettier-style intuitions?

[Jonathan Weinberg]

Ron: I think we are (unsurprisingly) in pretty much total agreement here. One of the most fundamentally interesting junctures between philosophy & psychology is where we try to expose & evaluate significant empirical presuppositions of philosophical theories. And much of my point, in this post, was that surveys just aren't the right tool for that job. There are some jobs that surveys can help us with, but they are fairly limited (though still, I hope, compelling!) I'm here to destroy analysis-by-philosophers-intuitions -- but not to rebuild it on the spurious foundation of analysis-by-everyone-else's-intuitions. (This also suggests that I am in pretty substantive agreement with Neil's 'who cares what the folk say' attitude.) I think maybe you took me to be offering an advertisement for direct applications of surveys, but really I am unsure that there is much for surveys to do, outside of a minor role in the much larger, harder domain of empirical psychology. But I ended my post on a question, to see if anyone had any nice ideas about some uses surveys could be put to, but which I hadn't yet encountered.

Neil: Bringing in Einstein here seems to me to compare apples and oranges. I think there are good psychological reasons not to treat Einstein's expert intuitions about physics to be on a par with folk physics -- indeed, they are not even on a par with Einstein's own 'folk physical' intuitions. Trained physicists probably have (at least) two different psychological mechanisms they can draw on, in approaching a physical problem: their mostly-innate folk physics, and their mostly-acquired, er, 'real' physics. But the development of the latter system does not seem to overwrite the former, and physicists placed in various sorts of task environments without recourse to paper, pen, and calculator perform pretty much as badly as completely untrained subjects. (Or so, at least, is my understanding of the data.) Your "how do you know they don't mean something else by 'knowledge'?" question is a perfectly good question about confounds, and you control for such possible ambiguity the same way you do in other applications of this methodology in psychology: you ask other questions whose point is precisely to detect such understandings. E.g., we asked subjects about a case involving a true belief accompanied by a strong feeling of subject certainty, in order to peel away any subjects who may be operating with that notion of knowledge. Interestingly, only one subject in the entire study considered truth-plus-subjective-certainty sufficient for knowledge.

(Btw, Sartre as you quote him does not seem to me to be in any real disagreement with the analytic epistemological tradition, but rather he seems to be making a fine Gricean observation.)

[Neil]

I wasn't aware of the data on physicists - the claim that their folk physical intuitions co-exist with their trained responses. It weakens the analogy I wanted to make, but not the point. My claim was that philosophers' intuitions should count for more than those of the folk, because they are trained intuitions, the product of reflection upon various cases. These intuitions remain defeasible, because philosophy is a distributed enterprise. The tradition has a method checking intuitions: it holds them up to the scrutiny of the entire community. We do not rely upon the deliverances of our individual introspection. If (as in the Gettier case on the evidence we have) the overwhelming bulk of the intuitions of those people whose theories and judgments are closest to wide reflective equilibrium is that such cases are not knowledge, then they are not knowledge. The interesting result would be if East Asian philosophers who have reflected on the same kinds of cases and engaged in the same kjnds of debates had the kinds of intuitions you describe. That would be an interesting result, calling for a revision in our concepts and perhaps a distinction between kinds of knowledge. Note the disanlogy, on your account, between physics and epistemology. The mechanisms which generate physical intuitions are informationally encapsulated and innate; hence their resistance to revision. But epistemological intuitions are not like that at all.

I am sympathetic to the view that surveys are pretty useless for philosophy. I also agree that whenever philosophical theories have empirical implications, we should test them. Philosophy of mind needs to be cognisant of neuroscience, moral psychologists need to read social psychology, debates in meta-ethics might be informed by consideration of psychopathologies, and so on. But we can eliminate appeal to intuitions from philosophy only by eliminating large parts of its traditional subject matter. How can we discover what knowledge is without appeal to intuition? What kinds of empirical information would help here? I would go further, and argue that cognitive science itself, indeed all the natural sciences, is reliant covertly on intuitions about, for instance, similarity and difference.

[Matt Weiner]

I'm glad to see this post--I had feared that my efforts at starting trouble had been for naught!

I'd like to agree with Neil, I think, although if Neil is a heretic about experimental philosophy I'm an am-haaretz (ignoramus). It seems to me that there are two different uses of the word "intuition" at issue here. One comes up in Gettier-type cases--people say, "Intuitively, these cases aren't knowledge." What they mean is: There's this word "know" that everyone uses, we're competent speakers of English so we can use it, and so by looking at our intuition we are revealing facts about this ordinary word "know" that everyone uses. And I agree that that is empirically exposed, and if your purpose is to get at some non-technical sense of knowledge you'd better make sure that that's what you're getting at.

But the role of "intuition" in many free will debates seems to me fundamentally different. When someone says something like "Intuitively, if you can't do otherwise you're not responsible" or "Intuitively, in the Frankfurt case Jones is responsible for his action," they're not explicate what the folk mean when they use the word "responsible"--they're trying to say what matters for responsibility. To mangle a quote from Wittgenstein, intuition is an unnecessary shuffle--it's the philosophers' way of saying "this argument works" or "this argument stinks."

One of the differences between the free will case and the knowledge case has to do with how people talk about the importance of the concept they're discussing. In Gettier cases the initial philosopher's reaction wasn't "We need this connection for knowledge to be something that's worth having," it was "Wow, this doesn't look like knowledge," with analysis of why knowledge might be worth having coming much later. In free will cases the argument always goes something like "The kind of 'freedom' that the compatibilist discusses may be well defined, but it's just not important. You couldn't really ground praise and blame on that sort of 'freedom'." And I take it that the people making that argument, if they were to discover that the folk did ground praise and blame on that sort of freedom, would say that the folk were making a mistake.

(Compare surveys showing that people think that a word is more likely to end 'ing' than to have 'n' as its next-to-last letter--no one's going to think that's right, no matter how overwhelming the survey is!)

So I don't necessarily think that compatibilists should be worried if a survey shows that most all the folk have incompatibilist intuitions. The response could be--"They haven't thought about these problems hard enough to figure out what's important to moral responsibility. We have, and we find these considerations convincing--our opinions count for more because they've been hardened in the fire of philosophical debate." Compatibilists maybe should be worried that about half the philosophers working in the field just don't see the point of their arguments, and vice versa, but that's another story....

[Jonathan Weinberg]

I think Matt is definitely on to something about the disanalogy between the role that intuitions about cases plays in 'S knows that p' on the one hand, and the free will debates on the other. I'm on the committee for a defense-of-libertarianism dissertation right now, and I find the level of the argument takes place almost entirely at the level of clashes between highly abstract principles.

I'm definitely starting to get a clearer sense of Neil's position now. (I had though it was, 'who cares about the folk's intuitions, since all intuitions are bunk'. But I see now that it is, 'who cares about the folk's intuitions, since they haven't had the kind of training & experience that we have as philosophers'.) The argument is, roughly, that we defer to experts' intuitions all the time, as in Neil's example of modern physics -- to which one could immediately start adducing all sorts of other quotidian examples, like the expert intuitions of your doctor or your car mechanic. And there are lots of intuitions by philosophers that I think do have this status: intuitions that such-and-such a counterexample can raise only technical, not deep, problems for so-and-so's theory; that a particular counterfactual analysis of X will suffer from the same problems that exploded a counterfactual analysis of Y; that a particular idea for a dissertation won't be marketable; and so on. In all such cases, we can see ourselves as having had significant experience with some relevant training set, and that for some sufficiently large number of cases, we have means other than the intuitions to tell when we've gotten things right (or, perhaps more importantly, when we've gotten things wrong).

But why think that philosophers' intuitions about cases involving concepts like KNOWLEDGE or CAUSE or FREE WILL have that expert status? We don't have more knowledge than, say, your average engineer, and we certainly don't cause things more than, well, almost any other profession. And I presume we have exactly as much free will as most everybody else (however much that is). What is the basis for our claim to expertise? To put it differently, and tendentiously: by what right do we claim our intuitions to be analogous with those of expert physicists, as opposed to analogous with those of expert astrologers?

Please note that I'm not just offering that up to, as it were, epater le bourgeoisie philosophique -- indeed, I am prepared to argue that astrology really & truly is a better analogy here. What physicists have, but philosophers and astrologists don't, is (as mentioned above) a track record of checking their intuitions against the world. Astrologers don't because they refuse to; analytic philosophers don't because, as near as we can tell, they can't; but, regardless of their reasons, both groups are in the same, fatally-leaky methodological boat.

[Neil]

Jonathan asks "why think that philosophers' intuitions about cases involving concepts like KNOWLEDGE or CAUSE or FREE WILL have that expert status? We don't have more knowledge than, say, your average engineer, and we certainly don't cause things more than, well, almost any other profession. And I presume we have exactly as much free will as most everybody else (however much that is). What is the basis for our claim to expertise?"

(I know its possible to quote from comments much more neatly than that - indented and italicised - but I don't know how. Anyone?)

Well, as Jonathan mentions, mechanics have better intuitions about cars that other folk, and doctors have better intuitions about somatic disorders than, say, philosophers. Yet a good mechanic may not spend more time driving than a non-mechanic, and a doctor probably has exactly the same number of bodies as I do (s/he may not even get sick more). Its because she thinks about and works with these things regularly that she acquires the relevant expertise. When I drive my car, I am concerned with its performance in a way that is quite different from the way that the mechanic is (in principle, a mechanic probably need not know how to drive). Its reflection upon a very wide range of cases, actual and possible, and upon the best arguments for and against a view, that leads philosophers to acquire relevant expertise. Notice that we do not just say "Well, it seems to me that in that kind of case the agent doesn't have knowledge". Instead, we say, "that's not knowledge because getting the right result by chance is not sufficient to justify it; if you don't agree, consider these cases". (It might be true that the intuition precedes the explanation of why it is not knowledge, but if no explanation is forthcoming, we drop the intuition, eventually. In that case, the intuition serves a heuristic purpose). That's not to say the intuition is dispensable; note that when I say that luck of the wrong sort precludes knowledge, I urge your assent by asking you to consider other, less controversial cases. John Dupre writes somewhere that the average scientist knows as much about science as fish know about hydrodynamics. I think that's (more or less) right; doing an activity gives you special expertise about the aims and methods of that activity, but not about the higher-order questions that philosophers are concerned with. Philosophical expertise comes from doing philosophy.

I agree with Jonathan that part of the reason that analytic philosophers do not test their theories against reality is that they can't, at least directly. Note, however, that this puts them in much the same kind of boat as physicists, Our physical theories only impinge upon experience at the edges, in Quine's memorable metaphor. This is not to deny that there is a difference of degree here, but only of degree.

Finally, I'm glad Jonathan made explicit the claim that I took to be implicit in much experimental philosophy. If I understood him correctly, he was arguing that appeal to intuitions is necessary for philosophy, but entirely unreliable as a means to gain knowledge. He is not, therefore, recommending we do philosophy by other means, but that we abandon philosophy (or am I giving insufficient weight to the word 'analytic', Jonathan? Do you mean the kind of philosophy which consists in conceptual analysis? Outside Canberra, few people do this, and not all than many inside do it either). Experimental epistemology (for instance), is not epistemology by other means, but a change in subject.

This has an odd consequence, though. Philosophers like Cummins who distrust intuitions argue that if we rely upon them we are not doing metaphysics (say), at all, but folk-metaphysics, which is a branch of psychology. Folk-metaphysics is not concerned with the nature of reality, but with the nature of people's ideas about reality. Experimental philosophy, as I understand Jonathan's recommendation, explicitly embraces Cummins' picture but turns it around: it says metaphysics is impossible since we cannot avoid reliance upon intuitions, so let's do folk-metaphysics instead.

[Josh Glasgow]

Ron's comments about race theory are right on target, and they speak to the point that this is one area in which surveys actually *are* directly on point for developing, rather than merely destroying, a philosophical position. (Caveat: I've got a paper under submission defending just this methodological claim, and I'm working with a psychologist to do just such a survey.) The reason, as I see it, why surveys of folk intuitions are relevant here, is that this is one area of research where Jackson-style conceptual analysis is important. And the reason this kind of analysis is important here is that the race debate is not just about what might constitute race, but about the actual ordinary, folk conception of race - in particular, whether it maps onto anything real and whether it should be eliminated from public discourse and practice. Given that this - the *public* conception of race - is the topic of philosophical debate, we need to know what "we" (the public) are conceiving when we talk about race. So there's a whole host of philosophically rich work to be done in terms of coming to a better, empirically-grounded rather than armchair, understanding of the folk conception of race.