Are all intuitions created equal?

Experimental philosophy has repeatedly shown that philosophers' intuitions are not shared by ordinary people (or are shared by only a subset of ordinary people), and various skeptical conclusions have been drawn from this fact. 

But it could be that not all intuitions are created equal; philosophers' intuitions might just be more reliable than ordinary people's,perhaps because of their expertise (a view often aired by philosophers: see, e.g., Michal Devitt's discussion of Machery et al.).

Indeed, it could be that smarter or more reflective people tend to have intuitions that differ from ordinary people and that are in in line with philosophers' intuitions. That would be some kind of relevant evidence for the view that philosophers' intuitions are more relieable. 

But do smarter and more reflective people have different intuitions? This is what Angel Pinillos, Nick Smith, Shyam Nair, Peter Marchetto, and Cecilea Mun argue in a recent paper (which can be read here) by looking at the intuitions elicited by Knobe's harm and help cases. 

So, do you have smart intuitions? Find out by reading their very interesting paper.

Comments would be most appreciated by Angel and his colleagues.

Edouard

Comments

[Adam Feltz]

These are interesting results. Thanks!

Edward Cokely and I ran a similar study with the Knobe cases and CRT with a similar test; however, we didn't find any effect of CRT or of other fundamental cognitive abilities (e.g. working memory capacity which is known to mediate intelligence).

Perhaps one potential explanation of the differences between the results of your work and our work might be other factors related to CRT. In other work we have found that those who had more philosophical training (judged by the number of philosophy classes they had) tended to be higher in CRT (we report this in a paper that's in press). Indeed, this might be some more evidence that philosopher's intuitions (or at least their judgments which are multiply determined and may reflect considerable intuitive “System 1” and deliberative “System 2” conflict) are different from those of non-philosophers'. Interestingly, Jonathan Weinberg et al. have some evidence that cognitive reflectivity may only make you “wrong in a different way.”

Perhaps the more likely explanation of why our experiment didn't reveal the difference you’ve detected is that we had less power (smaller sample). You had an impressively large sample that would make it easier to detect very subtle differences. A quick effect size estimate does seem to suggest that the effect is fairly small (about 1% of the variance) albeit clearly reliable. Of course, even if this particular effect is small we think it is an interesting finding as the influence of cognitive reflection (or abilities, or more generally individual differences) is an important and largely under-examined issue.

The data and paper that Edward and I have in press also explored a variety of other individual differences. It deals with the Knobe effect and shows predictable fragmentation predicted by extraversion (a highly heritable personality trait). We further show that extraversion’s effect is mediated by specific beliefs and we argue this may be evidence of a socially adaptive (ecological rational) mechanism. In other work we’ve make the argument that heritable traits may issue fundamentally different intuitions, which may in part explain why some philosophical debates have proven so intractable.

Finally, we wondered if you collected demographic data in your sample (i.e. sex differences). In another paper of ours (in re-submission), we found that there can be an order effect in Knobe cases and that women are the ones responsible for reductions in the Knobe effect (help followed by harm) when they are presented with both cases. Additional studies also indicated that when both help and harm judgments are presented (i.e. within subjects) many people (including both men and women) report that they believe the same chairman is involved in both cases. This seems to be an important methodological consideration regardless of any potential order effect issues.

Please feel free to email us for any pre-prints. Congratulations again on some very interesting work and on your innovative approach.

[Eric Schwitzgebel]

Fiery Cushman and I are running an experiment right now (web-based, see earlier link on this blog) that will also compare philosophers' and non-philosophers intuitions along several dimensions -- intuitions about moral dilemmas, in our study. Preliminary results soon, hopefully!

[Thomas Nadelhoffer]

As some of you may remember, Trevor Kvaran, Eddy Nahmias, and I also ran a large on-line survey late last spring which enables us to compare the intuitions of philosophers and non-philosophers. Moreover, we can also further divide philosophers into undergrad majors, grad students, and professional philosophers. We were interested primarily in moral intuitions and intuitions about free will. Unfortunately, we have each been so busy with our independent projects, that we haven't put a lot of time into analyzing the results. Hopefully, we will have something interesting to share after the upcoming break.

[Angel Pinillos]

I look forward to seeing all of these results!

Adam, I looked at the most recent paper you and Edward Cokely wrote (which was very interesting by the way--its more evidence for the conclusion, which I think is right, that the Knobe Effect is a very complicated phenomenon). One thing about running the CRT and the Knobe probe is that, as we indicate in the paper, taking the CRT (and getting questions right) just before responding to the vignettes affects one's responses to the latter. However, responding to the vignettes does not affect subsequent performance on the CRT. In essence, getting questions right on the CRT is "disfluent". So care should be taken when using the CRT in this context.

To respond to some of your points. (1) we have some gender data for other tests we ran. We still have to analyze these and will send you this info when we get it (it wont be until next semester though), and (2) that is interesting that giving people both vignettes at once causes some of them to think that the vignettes concern the same chairman/company. We anticipated this problem so we added to the instructions information that would prevent subjects from making this inference (without changing the content of the original vignettes).

[Max Deutsch]

Very nice. I am not at all surprised that Ángel Pinillos is one of the forces behind such an interesting and impressive paper. I am a bit surprised that he felt compelled to get up from his armchair, however. What is the world coming to?!

I think it’s less clear than Pinillos et al. suppose that traditional philosophy assumes that intuitions are evidence. True, we have Kripke in N&N saying that he assumes so. But, really, it’s not what you say you assume, but what you assume—or better: what you *need* to assume. And I don’t think there are many arguments in N&N that crucially depend on assuming that intuiting p is evidence in favor of p.

What’s the take-home message re: philosophical method, I wonder? Can we now safely philosophize from the armchair, knowing that smarter folk intuit more like philosophers? That would be an overly optimistic conclusion to draw on the basis of this single study involving a single pair of cases concerning intentional action. So how do we proceed? It doesn’t seem as though traditional philosophy has been defended, if the take-home message is that, for each and every philosophical case, we need empirical studies showing us how the smart folk intuit about the case, so that we may then cross-check with the philosophers. And it’s an empirical question how philosophers intuit in the first place.

Relatedly, what would Pinillos et al. counsel, if it turns out that smart folk disagree with the philosophers on a range of cases? For example, I’d be very surprised to find out that Weingberg et al.’s (2001) apparent discovery that East Asians tend to intuit that agents in Gettier cases really know, as opposed to only believe, the relevant propositions has anything to do with general intelligence. Maybe they polled a bunch of dummies, but that seems unlikely. So if even smart East Asians tend to intuit differently from Western philosophers on Gettier cases…well, what?

There’s some interesting speculation about how, exactly, philosophers intuit with respect to Knobe’s vignettes. In a footnote, Pinillos et al. ask us to note that they are “only saying that philosophers’ judgments [re: Knobe’s vignettes] are symmetric,” they’re not saying that philosophers’ judgments are symmetric in a particular way (No-Harm/No-Help, e.g.). Even this guarded speculation requires empirical backing, however. More interestingly, Pinillos et al. have results that suggest that ‘maximally intelligent,’ ‘maximally aware’ subjects (a) don’t exhibit the Knobe effect but (b) tend to answer Yes-Harm/Yes-Help. But how will ‘symmetrical philosophers’ answer? This is an empirical question, but my speculation is that many of them will tend toward the other symmetry, i.e. No-Harm/No-Help. If that’s right, then philosophers intuit differently from the very smart and very aware. Uh oh.

[Angel Pinillos]

Hi Max, Thanks for your insightful (and kind) comments.

First, concerning intuitions and evidence, you raise a good point about what might be going on in N+N and Kripke’s quote. For our paper, however, we don’t want commit to the view that evidence in philosophy is “psychologized” (where it is of the form “it is intuitive that P”). We certainly don’t need to make this assumption.

Second, about the take-home message for philosophical methodology, I would say (and I don’t speak for the co-authors of the paper) that philosophers need not adopt a skeptical attitude towards the reliability of particular intuitive judgments. However, if it is shown that some of these differ from those had by the public (or are only shared by particular groups), then this cannot be ignored. It requires a closer look (perhaps along the lines we suggest in the paper). On a related point, I do think that paying attention to system 1/system 2 considerations discussed at the end of the paper might help us understand the source of intuitions (a point well known in psychology but less well appreciated in philosophy).

Third, on the question of philosophers’ intuitions concerning the Knobe vignettes, we assume in the paper that it is symmetric (but we do not make a prediction about the exact pattern of responses that would yield this symmetry). The evidence for our assumption is that (1) philosophers find the Knobe Effect asymmetry surprising and (2) Traditional accounts of “intentionally” developed by philosophers don’t account for, or acknowledge the Knobe assymetry (although, see G. Harman). I should add that experimental philosophers often take the Knobe Effect to be out of step with what philosophers think, so dialectically it is fair to make this assumption. It is worth noting that “intentionally” might not have an invariant semantics (it might be context sensitive, for instance) so that both “Yes” and “No” answers to each vignette might be acceptable. For more on this see Nichols and Ulatowski (2007) and also Cushman et al. (2008) (references found in the paper).

Finally, we did find that the group that showed maximal awareness and intelligence as measured by the CRT doesn’t exhibit the Knobe Effect: 65% say ‘Yes the Chairman intentionally harmed the Environment’ while 56% say ‘Yes the Chairman intentionally helped the environment’ (there is no stat. sig. difference between these two figures—so there is no Knobe Effect here). However, this doesn’t mean, as you seem to suggest, that those subjects tend to answer “Yes” for BOTH Harm and Help. That’s a different claim. It could be, for instance, that all of those who say ‘No Help’ (44%) also would say ‘Yes Harm’. In fact, the best that we can conclude from our data (mathematically) is that at least 21% (and at most 56%) of those maximally aware and intelligent (as measured by the CRT) will respond “Yes” to both vignettes (when presented on their own). But from this one can’t conclude that a person who is maximally intelligent and aware (as measured by the CRT) is likely to answer YES to both Harm and Help scenarios. That is, there is a difference between saying that a group fails to exhibit the Knobe Effect and making a further positive claim about what pair of answers a member of that group is likely to make.

[jonathan weinberg]

Hi Angel! Some thoughts:

"The evidence for our assumption is that (1) philosophers find the Knobe Effect asymmetry surprising" Well, that's true, but in my (entirely nonscientific) experience, philosophers tend to have the Knobean _intuitions_. It's just that they are surprised that they are having them. For example, on seeing Joshua give a talk on this stuff at the Pacific APA a few years ago, my then-chair really enjoyed the talk, and told me, only half-winkingly, "one great thing about his argument, is that you could have done it entirely from the armchair!" I have heard at least a handful of others say the same sort of thing.

"For our paper, however, we don’t want commit to the view that evidence in philosophy is “psychologized” (where it is of the form “it is intuitive that P”). We certainly don’t need to make this assumption." I heartily agree with that point -- I think that Max (hi Max! This is getting a bit incestuously Rutgersian, isn't it?), and others like Williamson and Jonathan Ichikawa, think you can get a lot of methodological mileage out of rejecting this psychologization. But it sure seems to me that pretty much everything that one might want to say about the methodology of intuitions on a psychologized account, one can just as well say about the methodology of judgments (or whatever) on a depsychologized account, so long as one mutatisizes the appropirate mutanda.

I should put an ad here for one of Ichikawa's recent papers on this topic, which a lot of folks will find interesting (and perhaps a bit vexing, but in a good way):
http://www.st-andrews.ac.uk/~armeth/2008/11/new-paper-draft/

I do want to add a point of what I _think_ is agreement with Max here: just showing that philosophers do something _differently_ from the folk does not suffice to show that they are doing it _better_.

[Angel Pinillos]

Hi Jonathan,

Interesting comment about the anecdotal evidence you mention. My own experience (more anecdote here) is that the few philosophers who have the initial Knobe answer inclination take it back after thinking about it further. Maybe the first intuition is a product of system 1 (quick, unreflective process) but the more reflective is part of system 2 (reflective, conscious etc).

Let me add, however, that in our paper we never say that the system 2 answer is the correct answer (or even that it is the competent answer) to the vignettes. That would require a few further assumptions: (A) system 2 captures competence, (B) competence tracks correctness, (C) there is one "intentionally" concept and (D) the vignettes are not underdetermined so that they may be interpreted in various ways (and this might make a difference to judgments made).

The main point of our paper is that a certain objection to skeptical doubts about intuitions raised by a particular experimental philosophy result (the Knobe Effect) is supported by the results of our experiments. The objection requires this premise: The difference found between philosophers' judgments and the judgments of the folk is due to the fact that philosophers are in a better epistemic position than the folk are in when the latter are placed in the typical experimental conditions created by experimental philosophers.

(Our results do not apply to all X-phi results. Your experiments with the NFC test and the true-temp case don't support the objection, as you point out. Nice experiment btw)

Now about this claim:

"I do want to add a point of what I _think_ is agreement with Max here: just showing that philosophers do something _differently_ from the folk does not suffice to show that they are doing it _better_."

I agree with this. For sure, we don't assume (or argue for) such a thing in our paper.

However, one could say this about our results (although this is not part of the paper, its not clear that my coauthors would agree with it and it would require further argumentation): Our result that people in better epistemic positions (smarter, more reflective, have more information etc) give answers that more closely resemble the philosopher's answer is some evidence that philosophers are on the right track.

[jonathan weinberg]

I do think that, as more & more of this sort of research gets done, we'll find areas of clear superiority of philosophers over Das Volk. (I mean, just consider the Wason Selection Task...) I also expect that we'll find areas where there is no such superiority, and perhaps even some where they do worse (because of theory-grip and the like). I look forward to our community's working to map all this out over the next god-knows-however-many years.

* * *

I would note I little puzzlement at this, in your last comment: "skeptical doubts about intuitions raised by a particular experimental philosophy result (the Knobe Effect)". Has anyone ever tried to use the Knobe effect results as premises in an anti-armchair argument?

(Fwiw, I try to avoid using "skeptical" and its cognates in these circumstances, as I think it leads people to try to map these arguments into traditional skeptical ones, which often leads to some very high-powered missing of the point. I think this happens to Tim Williamson in the relevant chapter of _The Philosophy of Philosophy_, for example.)

[Angel Pinillos]

"Has anyone ever tried to use the Knobe effect results as premises in an anti-armchair argument?"

The traditional philosophers consults her own intuitions regarding some concepts to get at the folk concept. Joshua Knobe, as I understand him, thinks that this isn't enough to do the job. He thinks that the Knobe Effect is evidence that the task of getting at the folk concept should be supplemented with experimental work. This is a type of skepticism about the reliability of the philosophers' intuitions. After all, if it was really reliable then we wouldn't need to do experimental work to get at the concept. This would be consistent with what you (and contributors) call the proper foundation view of experimental philosophy. This seems to me like a type of skepticism about the reliability of intuitions (philosophers') for getting at the folk view.

[jonathan weinberg]

But this seems to me to get the order of the argument wrong, in Knobe's case. _First_ Knobe takes us to have good reason to think that philosophers' armchair intuitions need to be supplemented by experimental methods. _Then_ he demonstrates the side-effect effect using experimental methods. The doubts about the sufficiency of armchair methods motivates looking for the that effect experimentally, but not, I think, the other way around.

(Presumably _Knobe_ had Knobe-effect intuitions, as that was part of the motivation to look for them experimentally! But he would not think that his own case would be sufficient to generalize from, to make the large points about folk psychology that he wants to make. Hence, he had to go after it experimentally.)

And, again, I really would warn against using "skepticism" here. When a scientist says that, for her purposes, ordinary perceptual judgment, uncontrolled and unaided by instruments, etc., is not up to the required task -- is this a claim that would merit the label "skepticism about perception"? It seems to me rather misleading to say so.

[Angel Pinillos]

Well, I think that there is a use of "skeptical", much more tied to ordinary usage (as opposed to a more loaded usage) in which it is correct to say that the results of experimental philosophy have caused many to be skeptical about the reliability of the method of armchair philosophy to get at folk concepts (even on the "proper foundation" view of X-phi). We mean to be invoking this ordinary sense. But you are right that this should be clarified to prevent a misunderstanding.

In the paper we don't say that the Knobe Effect engenders skepticism about intuitions (on the proper foundation view). It is that it engenders skepticism about the reliability of the traditional method of appealing only to philosophers' intuitions to get at folk concepts. Now even if Knobe had antecedent doubts about armchair philosophy, it seems right to me that the fact that the Knobe Effect showed us something of importance for getting at the folk concept that we didn't get just by using philosophers' intuitions, is some evidence that the method of using only philosophers' intuitions to get at folk concepts is less reliable than previously thought. And hence Knobe's results engender skepticism about the reliability of traditional armchair methods.

About your perception analogy, I think it would be right to say that if someone presents evidence that ordinary perception is not good enough to carry out a task X, then this would lead to skepticism not about ordinary perception (as you correctly point out) but rather it would lead to skepticism about the reliability of the method of using only ordinary perception to settle matters about X. This would be our corresponding assertion in your analogy.

[Edward Cokely]

This work is very interesting and exciting. Thanks for sharing and daring to go in new directions!

A few questions:

1. Isn’t it important to distinguish between intelligence and cognitive reflection?

The relationship between CRT and other psychometric measures of intelligence are modest (r = .3-.5). Although CRT is correlated with intelligence it is NOT an intelligence measure. CRT measures something very specific—i.e., one’s cognitive impulsivity and tendency to be more analytic and careful during information processing. This construct has been shown to have convergent as well as discriminate validity (i.e. it accounts for unique variance that isn’t related to other intelligence measures). Indeed, there are many, many things that correlate with intelligence (i.e. mortality, SES, wearing glasses) but these are all theoretically and pragmatically quite different.

Process level research shows that CRT is related to the amount of information search used during judgment and decision making (i.e. CRT predicts longer reaction times and more elaborative think aloud protocols). If there are differences in search and contextualization processes that give rise to differences in judgment this may bear critically on the implications of these findings. To be clear, I don’t believe it makes the current results any less interesting that they may speak primarily to cognitive reflection (and may or may not speak to intelligence). In some theoretically important ways it makes them even more interesting (e.g. see question 2).

2. If one is high in cognitive reflection should we assume one’s judgment is an intuition?

In the paper there was an explicit decision not to address this question. Perhaps you’d be willing to speculate here. Specifically, the cognitive reflection task is said to capture different tendencies to use more deliberative (so called “system 2”) versus intuitive (so called “system 1”) processes. If we accept Kahneman and Frederick’s interpretation, the idea is that reflective people have an intuition and subsequently call on metacognitive (e.g., monitoring and control) processes to alter the intuitive “impression” during judgment and decision-making. As noted, a relationship between CRT and deliberation has been observed in process level data. Therefore, the difference in the relationship between high CRT scores and judgments (e.g., Knobe cases) may not reflect any difference in “intuition” (i.e. Bealer’s intuition as uncritical belief). Wouldn’t it be theoretically quite interesting, and important to know, if the observed differences reflected variability in more deliberative and extensive information processing rather than more automatic, “intuitive” differences?

3. Might the Knobe scenario be affecting CRT?

Rather than the CRT influencing judgment on the Knobe Help cases, might the Knobe cases be influencing CRT (i.e. improve CRT)?

4. (A comment) Dual System theories are not without controversy.

There are many very critical arguments about the limits of dual system theories. One is that dual system theories can post-hoc explain any set of data because they are so vague and don’t make clear a priori predictions (for discussion see Gigerenzer & Regier, 1996). Another is that there is no single dual system theory but rather that mechanistically most dual process theories are very different, save the fact that most posit at least two “modes” (or types) of processing (Evan, 2008). Regardless of one’s opinion, or how widely embraced some ideas are, it would seem reasonable to acknowledge these major criticisms.

Thanks again and cheers on a fine paper…

[Angel Pinillos]

Hi Edward,

These are great points. let me address these (although the comments here will be too brief, unfortunately)

(1) The fact that the CRT test does correlate with measures of intelligence (although to varying degrees) and the fact that CRT measures reflectiveness (which can be plausibly understood to be a feature of intelligence--thanks to Joshua Knobe here) suggests that a higher CRT score is *evidence* for higher intelligence. This is the important claim for us. It is in this sense that we say that CRT is a measure of intelligence. But you are right that this should not be interpreted in a stronger sense. Thanks.

(2) This is an important, and in my view, extremely difficult issue. Certainly, I would not want to assume that when philosophers talk about intuitions, they must be talking about the products of system 1. In other words, Kahneman's use of "intuition" in this context cannot be assumed to be the same as the philosopher's use of that expression. Personally, I suspect that when philosophers talk about "intuitions" they mean or should mean something closer to the products of system 2. After all, System 1 is known to be subject to certain biases.

(3) The mean CRT scores of those who took the CRT after responding to a vignette is the same as the mean scores for those who took the CRT before answering the vignette. This led us to think that answering the vignette did not affect CRT performance.

(4) Duly noted. thanks.

[Max Deutsch]

Hi Angel and Jonathan! If only we could run over to Doll’s and continue the conversation with help from beer.

Angel, you’re right; I did mistakenly suggest that your results show that the very smart, very aware folk incline towards Yes-Harm/Yes-Help. Still, though that’s wrong, your results do show that the very smart, very aware folk intuit things that the philosophers don’t, or anyway might not. For example, your results show that the majority of very smart, very aware folk intuit Yes-Help, right? But the 3 philosophers I just asked intuit No-Help. You say: “Our result that people in better epistemic positions (smarter, more reflective, have more information etc) give answers that more closely resemble the philosopher’s answer is some evidence that philosophers are on the right track.” You have a variety of results, only some of which count as a result that “people in better epistemic conditions give answers that more closely resemble the philosopher’s answer.” In fact, one of your results is that people in an extremely good epistemic position give an answer that differs (maybe) from the philosopher’s. Is this evidence that philosophers are on the wrong track?

Re: philosophical method—a philosopher finds herself wanting to appeal to an intuition. You say she “need not adopt a skeptical attitude toward the reliability” of the intuition. On the other hand, if it is experimentally shown that the folk disagree with that particular intuition, she can’t ignore the disagreement, you say. If she should be worried at all, I’m not sure why she should not be worried by the folk _potentially_ disagreeing with her, even if there’s no survey data to show it. Perhaps she ought to run a survey on the folk to double check—and then a further survey on smart folk to triple check. Anyway, not ignoring the folk when their intuitions clash with yours might involve presenting an argument to the effect that their intuitions are wrong. If you’re feeling especially generous, you could offer an “error-theory” explaining why they screwed up. But all that could be done from the armchair, or so one might think. I get the impression that you think not; the proper response to surveys showing that folk intuitions differ from the philosophers’ is: more surveys, perhaps ones that weed out the dummies. Is there some non-survey-method way to respond to demonstrated intuitive clashes between philosophers and the folk? If not, then a lot of philosophy is going to require a lot of x-philosophy (shudder), since there are plenty of demonstrated clashes. Seems like there’s a new one every other week or so (darn xphiles).

Jonathan, you at least _think_ you may agree with me about something! Hurray! I don’t understand your point about psychologized evidence though. You suggest that even if I’m right that evidence in philosophy is not psychologized, xphi intuition surveys might still be used to criticize various arguments in philosophy. But how can that be? The surveys tell us only that various purely psychological facts do or do not obtain. They tell us that people intuit this way or that. But if I insist that some argument does not depend on these sorts of psychological facts, and I’m right, how is it that the surveys are going to matter in any case? For example, suppose I say Weinberg’s Gettier study doesn’t touch Gettier’s anti-JTB argument because, although the study casts doubt on the claim that it’s intuitive that agents in Gettier cases fail to know, the anti-JTB argument doesn’t hinge on it being intuitive that agents in Gettier cases fail to know. Instead it hinges on premises that have nothing to do with who intuits what. How do you “mutatisize the mutanda” here?

[Jonathan Weinberg]

"Instead it hinges on premises that have nothing to do with who intuits what." Ah, but as stated, this is not true. The _content_ of the premises may not have anything to do with "who intuits what". But it is not correct to say that the epistemic status of the premises _as premises_ does not hinge on exactly those sorts of issues. Everyone needs an account of what premises are epistemically appropriate to use when, and will in part concern how it is that we gain entitlement to such premises, and an inelminable chapter there will concern many psychological facts about human agents. Maybe facts about "who intuits what" won't themselves be premises in the main philosophical arguments. But they will play an essential background role nonetheless.

One way to see the weakness of the "not a premise, therefore not relevant" argument is to just look at our ordinary practices. For example, I tell my wife "The car is parked in the Jordan St. garage," and she raises to me the concern that my memory of such things is pretty lousy; but it would be preposterous for me to reply to her concern with, "But we were discussing the location of the car, not the psychology of my memory faculties. Stay on topic, honey!" We also very frequently do make explicit reference to the psychological nature of our epistemic access to various claims; "I saw you looking on that other student's paper during the midterm." It doesn't matter whether we make sense of that by making the mental state description itself part of the premises of the further argument (about why the student in question is getting an 'F', say), or whether we make sense of it by giving the facts about our mental states some other, non-evidential role in securing for us the really relevant premise, that the student was looking at another student's paper. Either way, facts about our psychology will be epistemically very relevant.

[Edouard Machery]

Absolutely correct, Jonathan.

[Angel Pinillos]

Hi Max. well, when we say that the answers of the smart/aware group more closely resemble the answers of the philosophers than do the answers of the less aware/smart, we mean that they do so with respect to the measure of symmetry. We don't mean that they more closely resemble in EVERY respect. Since the Knobe Effect is about the asymmetry, then this seems like the right measure.

Let me give an example that might help. Suppose it were "inconsistent" to be pro-life and pro-death penalty (I am not at all implying that there is any merit to this, this is just an example). And we found with surveys that (1) most regular folk held both of those views. Philosophers are consistent and it happens that (2) most of them are pro-life and anti-death penalty. However, the smart and aware members of the general population are consistent but (3) most of them are pro-choice and pro-death penalty. Here, we would say that the smart/aware more closely resemble the philosophers in the measure of consistency than the less smart/aware group. This is so even if the smart/aware folk differ from the philosopher other respects (the philosopher is pro-life while the smart/aware folk is not).

About the second point, you say:

"Is there some non-survey-method way to respond to demonstrated intuitive clashes between philosophers and the folk?"

Well, yes there are and there will be more. For example, we can already appeal to semantic/pragmatic confusions (think of uses that are clearly metaphorical for example). Or we can appeal to framing effects and other known biases. As we learn more about these then we can explain clashes without doing more surveys.

[Max Deutsch]

Angel, I didn’t mean to suggest that you claimed in the paper that every intuitive judgment about Knobe’s cases is or will be shared between smart folk and philosophers. I meant only to draw attention to the interesting result that the majority of perfect scorers on the CRT in the CRT-before condition judge Yes-Help. I find that surprising. It also strikes me as a result that could be turned _against_ the Immunity Objection.

Ok, glad to hear that there are responses to philosopher/folk clashes that don’t involve intuition surveys. However, your idea that some of the clashes may be explained away by invoking the semantics/pragmatics distinction is not clearly a survey-free method. Xphiles can say (and have said) that proposals to the effect that some range of results on an intuition survey is due to subjects mistaking implicatures for contents is an empirical claim to be tested by—wait for it—more surveys.

Jonathan, if by ‘the epistemic status of the premises _as premises_’ you mean their status as justified or not, then I think you’re wrong to insist that that status hinges on issues to do with who intuits what. Gettier might justifiably believe that the agents in his cases fail to know even if intuition were completely agnostic about his cases. I say this not simply because Gettier’s premise is silent, content-wise, about what is/aint intuitive, but also because intuitions need play no role in its justification. Of course, it is hard to disagree with the claim that ‘psychological facts about human agents’ will play an ‘essential background role’ in a story about epistemic justification. But, remind me, how do we get from here to the thought that xphi intuitions surveys ought to strike fear into the hearts of analytic philosophers? About your mistreatment of your wife—if you admit to your wife that, yes, your memory isn’t very good, and so relying on it alone to reveal the location of the car isn’t a very good idea, but you then point out to her that you have this other bit of evidence re: its location (maybe you just found a Jordan St. garage stub in your pocket) and she continues to go on and on and on about your bad memory, well, now she’s the one behaving badly.

[Jonathan Weinberg]

"if by ‘the epistemic status of the premises _as premises_’ you mean their status as justified or not" I don't. There's a lot more to epistemology (and a _lot_ more to methodology) than the question, is this justified or not? There's all sorts of ways in which someone might be justified in their belief in a particular proposition, but still not be entitled to use it as a premise in a particular argument. (I don't mind granting that philosophers generally have personal justification for their intuitive claims, fwiw.)

This bit seems to give away your game entirely: "if you admit to your wife that, yes, your memory isn’t very good, and so relying on it alone to reveal the location of the car isn’t a very good idea, but you then point out to her that you have this _other_ bit of evidence..." (emphasis added) Wait, who said that my memory was serving as evidence in the first place? Where'd _that_ come from? My original "text" to my wife was simply a claim about a location. No appeal to memory in that text at all. Maybe appealing to "I remember that p" as itself a piece of _evidence_ isn't the right way to cash this sort of thing out; or maybe it is. But the important thing is that facts about human (or, in this case, just my) memory are unavoidably relevant to the epistemic status of the substantive, not-about-psychology-at-all claims. That's where the mutatis mutandis comes in for philosophers. Even if there's not such thing as _intuition_ per se, a la Williamson or something like that, it's still the case that a story needs to be told about how to distinguish claims we ought, and those we ought not, appeal to as premises; and that facts about our psychology will be central to that story.