I have a friend, and often we agree to hang out. One out of every 10 times we make plans, he doesn’t show up, and when I call him it turns out he forgot completely, or remembered but didn’t feel like going.

Despite the fact that this friend is reliable 9/10 of the time, overall he is very unreliable.

If I have a car that won’t run 1 out of every 20 days I try to use it, it is highly unreliable, despite the fact that its successes vastly outnumber its failures.

And since you correctly point out that we can’t compare the number of intuitive successes with intuitive failures — they are assuredly both infinite, and since we can only make countably many sentences in English, this suggests that they are not only both infinite but both of Aleph-Nought size — it seems that your argument states that intuition doesn’t always fail, but nothing more.

However, I think the issue of the role of intuition in philosophy goes slightly deeper than whether or not it is reliable. For decades now, there has been a movement in philosophy to adopt concepts or arguments based (at least in part, if not in whole) solely on intuition. Why is X morally wrong? Let’s appeal to intuition. How can I argue my stance of moral absolutism? Hey, intuition!

Any scientist can tell you that this is a horrendous practice. Intuition is a very valuable, very human tool. It is a wonderful mechanism for generating conjectures. However, it is completely unreliable as a guide towards truth, and completely unreliable as a means of proof. That is to say, more precisely, that if our goal is to arrive at the truth of a situation, and to demonstrate that that is the truth as well as we can, intuition is nigh worthless. Imagine a mathematician saying he had proven the Riemann Hypothesis because it was intuitively obvious, and then trying to claim the $1 Million prize. This should be the reaction of those who hear the Philosopher claim that something — anything — is intuitively obvious.

This isn’t to say that intuition is worthless, of course — just that it’s worthless as a means of argument. It’s very worthwhile as a guide to the conjectures we wish to argue. But to fail to make that distinction I think is intellectually dishonest, and only promotes poor reasoning skills.

It is more than verbal disagreement and it is, I think, more accurate than what I said! Thanks!

I think that you are right in pointing out that perception rather than intuition might deserve the credit for the belief about the cat. Moreover, the case might also invite the worry that the supposed intuition is actually caused by the belief, and not by an independent process of intuiting. If the belief is there because of perception, then the intuition might be there because of the belief. In that case it seems that, roughly speaking, the intuition will simply inherit whatever reliability we ascribe to the belief. Obviously, this does not make the belief unreliable, but it certainly seems to imply that checking one’s belief against one’s intuition would not really be worthwhile in such cases. So I think our interest should be in the reliability of intuitions that could not be taken to reflect already held belief in this way. The problem, of course, lies in determining when this is the case.

And to dtlocke: you say “I think we are interested in the reliability of intuition used as a belief-forming method. In other words, what we want to know is what percentage of the time a belief-that-is-formed-on-the-basis-of-intuition is true.” I am not sure that this is right. In particular, I do not see why we shouldn’t simply be interested in the reliability of intuitions as such. If an intuition that p is something like “an intellectual seeming that p”, as Bealer suggests, then it seem that intuitions may be considered reliable independent of their impact on our beliefs. For instance, contradictory intuitions about some subject matter, as in the case of the sceptical puzzle, might lead one to suspend belief. Arguably, this should make one interested in the reliability of the intuitions themselves, since this would seem to be a case with no good candidate for the role of “intuition used as a belief-forming method”. However, I am not sure that this is more than a verbal disagreement.

I’m not sure that I need to be positively committed to all that. First, I’m really not committed to any sort of unreliability claim at all, though I do think that, with regard to Brian’s argument here, that it wouldn’t work so well were it offered as a defense of the reliability of the philosophical practice. With the common-sense-type cases included in the mix, then reliability comes along pretty easily. Once we subtract off all the common-sense-type cases, then I think it probably just becomes fairly hard to guestimate the reliability.

As for the quotidian/far-fetched distinction, I agree that we don’t want to have to rest things on our rough sense of it. (Maybe Eric’s student will be able to help us here, though it sounds from Eric’s brief description like they might be using a distinction between types of possibilities, and I have in mind something more psychological than modal.) There’s no reason to think the distinction won’t be amenable to scientific study, though. Here’s one working hypothesis: where our concepts have a prototype structure, “far-fetched” cases are ones that involve a significant tension between different features. One prediction of that hypothesis is that such cases will be more amenable to framing, order effects, etc., because subtle shifts in the weighting of the different features could cause a flip in the prototype’s determination. But that is all speculation on my part way in advance of the necessary empirical work!

I’m not sure that the distinction you highlight is really a distinction between a descriptive and a normative sense of “reliable”. Rather, I think that it is the distinction between a basic sense, which may just be something like “more often true than false” and a more demanding sense, which requires a given practice/method to meet a requirement that goes (far) beyond “more likely to be true than false”. After all, I can also use the basic sense to criticize someone as being unreliable, namely if s/he gets something more often wrong than right. And I can use the more demanding sense in a purely descriptive way, once you give me a precise specification of what the relevant standard actually demands; used in this way, the statement “The umpires judgments are unreliable” seems to be a purely descriptive one.

as to your second point, you seem to be committed, now, to the claim that it is really just a sub-class of intuitions about hypothetical cases that is shown to be unreliable (and hopeless as well) by the results of experimental philosophy, namely the sub-class of intuitions about, as we my call them, “irreducibly far-fetched” cases. I agree that, intuitively, there seems to be a difference between the irreducibly far-fetched cases and the reducibly far-fetched ones, but I’m not sure what this intuition of mine really counts for. For, the distinction seems to be drawn with the sole purpose of isolating and criticizing the specific use of intuitions in philosophy – at least, I can’t see any other motivation for introducing such a distinction…

Actually, I have a similar worry about hope, for the distinction between “hopeful” and “hopeless” practices is new and formerly unrecognized in epistemology as well (which you also emphasize in your paper), and you introduce it with the main motivation of “challenging intuitions empirically without risking skepticism”, as already the title of your paper indicates. I have to admit, however, that you do a great job in trying to convince us that “hope” is not just an ad hoc epistemic category. Nevertheless, you shoulder a pretty heavy argumentative burden (which is, of course, highly admirable), if your build your empirical challenge to the philosophical use of intuitions on (at least) two formerly not recognized yet absolutely crucial theoretical distinctions…

You say, “Because, consider what it means for a practice to be reliable: it means that most of the output intuitions/beliefs of that practice are true.”

I think this picks out one, descriptive, sense of “reliability” but there is another, normative, sense. Consider when I say that baseball umpires’ judgments of foul balls / homeruns are not reliable enough so MLB should employ instant replay. I don’t mean that they’re not getting the call right most of the time. I’m saying that they’re not getting the call right enough of the times, especially in borderline cases. That is, they’re not meeting some standard I believe to be appropriate. When we talk about a practice being reliable in everyday usage, I believe, we often actually have the normative sense in mind.

So this is the normative sense of “reliability”: as meeting some standard. I agree with Jonathan that descriptive reliability really is rather cheap. But I don’t think normative reliability is so cheap. Perhaps this is because there are other important notions underlying the normativity that are doing the epistemological heavy lifting.

(I must confess ignorance of not having read Jonathan’s paper. So perhaps what I call normative reliability is the same as his “hopefulness”. Nevertheless, I think the normative notion is equally deserving of the term, as evidenced by our everyday usage.)

Joachim, I don’t think that Brian’s argument can be run so easily against the “practices” version, because that version would deprive him of two of the main resources he has drawn on. First, and most centrally (and along the lines of your comment #6), it deprives him of the appeal to the vast range of ordinary cases that serve to make intuition’s track-record look so good. Such cases just do not play much role in philosophical debate; claims like “torturing kittens for fun is wrong” tend to show up in the literature not to help make the case for one theory of normative ethics over another, but rather as examples of obvious moral truths, or of the sorts of claims that one has to make some sense of in one’s metaethics, etc.

Second, it also deprives him of the kind of move he makes in his comment #7, too — because those sorts of claims don’t seem to play much of a role in philosophical practice, either. (Maybe the ET case could in the right circumstances, say if one is looking to attack a species-centric account of ethics, perhaps in a debate over eating meat.) Yes, such cases could be multiplied endlessly, but as a matter of fact, doing so doesn’t seem to be something that can do any philosophical work for us. Once one gets the recipe for making variations of the “torturing X for fun” cases, they don’t seem to count any more for or against a philosophical theory than any one version of them could. Brian’s cases are also different from, say, the cases that Eric mentions in #4, in that it seems like they just collapse down to the quotidian versions anyway — one can divide through by the far-fetchedness, and see them just as gussied-up versions of ordinary cases, substituting “ET” for “kittens”. But there’s no equivalent ordinary cases for Block or Searle.

In general, really obvious & commonsensical cases (whether quotidian or far-fetched) can’t do much work for us in our practices, because mostly we’re mooting among the highly restricted & elite set of theories that philosophers find worthy of consideration and defense. And these theories will most often be ones that one cannot choose between on obvious & commonsensical grounds. Most theories of normative ethics get all these cases right, or have really good resources for accommodating the cases of that sort that they don’t seem to get right. So all the cases that Brian needs to swell the ranks of the highly probably intuitions, are also cases which are unlikely to actually ever get used by anyone in our practices.

(Maybe Gettier is a counterexample (and thus a meta-counterexample!) here, if it has the status of not-previously-noticed obvious case. I am not inclined to grant it that status, but even so, it seems to me unusual in this regard. Contrast it with, say, the Gypsy Lawyer case, or the various Grabits, or the cases in Jennifer Lackey’s learning from words arguments….)

I also have to disagree with your claim that “if his argument really shows that intuitions as used in philosophy are reliable, then I think that to nevertheless criticize them as “hopeless” seems to lose a lot of its epistemological bite.” I go to some length to argue that hopefulness is not just a bit of epistemological lagniappe, but really a very central characteristic of trustworthy evidential practices. It should definitely not be put in the same pile with certainty, which I agree is something that’s merely nice to get when you can, but not an epistemic necessity.

Part of the problem is that it’s just not right to say that mere reliability does very much on its own to lift philosophical intuitionizing above most human epistemic practices. I suspect that most such practices will count as reliable too, along much the same lines as those offered here by Brian. Most astrological predictions come true, after all (because they are sufficiently open-ended that they are unlikely to come out false). I suspect that most uses of things like the sortes vigilianae were reliable, too, since much like intuition, they were in part a way to channel one’s own common sense. Most epistemic practices in the history of the species have been carried out in a manner continuous with common sense, and thus will get the benefit of the same kind of track-record argument. Reliability really is rather cheap, which is part of why we should look to other notions to do our epistemological heavy lifting.