Tuesday, December 11, 2012

Intuitions, Philosophy, and Experiment

Herman Cappelen has provocatively argued that philosophers don't generally rely upon intuition in their work and thus that work in experimental philosophy that aims to test people's intuitions about philosophical cases is really beside the point. I have a simple argument against this view.

First: I define "intuition" very broadly. A judgment is "intuitive", in my view, just in case it arises by some cognitive process other than explicit, conscious reasoning. By this definition, snap judgments about the grammaticality of sentences, snap judgments about the distance of objects, snap judgments about the moral wrongness of an action in a hypothetical scenario, and snap folk-psychological judgments are generally going to be intuitive. Intuitive judgments don't have to be snap judgments -- they don't have to be fast -- but the absence of explicit conscious reasoning is clearest when the judgment is quick.

This definition of "intuition" is similar to one Alison Gopnik and I worked with in a 1998 article, and it is much more inclusive than Cappelen's own characterizations. Thus, it's quite possible that intuitions in Cappelen's narrow sense are inessential to philosophy while intuitions in my broader sense are essential. But I don't think that Cappelen and I have merely a terminological dispute. There's a politics of definition. One's terminological choices highlight and marginalize different facets of the world.

My characterization of intuition is also broader than most other philosophers' -- Joel Pust in his Stanford Encyclopedia article on intuition, for example, seems to regard it as straightforward that perceptual judgments should not be called "intuitions" -- but I don't think my preferred definition is entirely quirky. In fact, in a recent study, J.R. Kuntz and J.R.C Kuntz found that professional philosophers were more likely to "agree to a very large extent" with Gopnik's and my definition of intuition than with any of six other definitions proposed by other authors (32% giving it the top rating on a seven-point scale). I think professional psychologists and linguists might also sometimes use "intuition" in something like Alison's and my sense.

If we accept this broad definition of intuition, then it seems hard to deny that, contra Cappelen, philosophy depends essentially on intuition -- as does all cognition. One can't explicitly consciously reason one's way to every one of one's premises, on pain of regress. One must start somewhere, even if only tentatively and subject to later revision.

Cappelen has, in conversation, accepted this consequence of my broad definition of "intuition". The question then becomes what to make of the epistemology of intuition in this sense. And this epistemological question is, I think, largely an empirical one, with several disciplines empirically relevant, including cognitive psychology, experimental philosophy, and study of the historical record. Based on the empirical evidence, what might we expect to be the strengths and weaknesses of explicit reasoning? And, alternatively, what might we expect to be the strengths and weaknesses of intuitive judgment?

Those empirical questions become especially acute when the two paths to judgment appear to deliver conflicting results. When your ordinary-language spontaneous judgments about the applicability of a term to a scenario (or at least your inclinations to judge) conflict with what you would derive from your explicit theory, or when your spontaneous moral judgments (or inclinations) do, what should you conclude? The issue is crucial to philosophy as we all live and perform it, and the answer one gives ought to be informed, if possible, by empirically discoverable facts about the origins and reliability of different types of judgments or inclinations. (This isn't to say that a uniform answer is likely to win the day: Things might vary from time to time, person to person, topic to topic, and depending on specific features of the case.)

It would be strange to suppose that the psychology of philosophy is irrelevant to its epistemology. And yet Cappelen's dismissal of the enterprise of experimental philosophy on grounds of the irrelevance of "intuitions" to philosophy would seem to invite us toward that exactly that dubious supposition.


Bernard said...

"... the answer [how best to resolve various philosophical conflicts] one gives ought to be informed, if possible, by empirically discoverable facts about the origins and reliability of different types of judgments or inclinations." The "if possible" is a significant hedge; e.g. it might turn out that how best to interpret the empirical findings depends on how best to resolve the relevant philosophical conflicts.

Unknown said...

One aspect of your account is that some intuitions can be trained and improved. For instance, when considering whether a sentence is grammatical, we might be warranted in trusting the intuitions of the native speaker over the intuitions of the novice, quite independent of theory.

If intuitions can be trained, then presumably their form is the result of some prior effort of conscious, explicit reasoning (the time spent learning the language, for instance). This seems to put tension on holding some strong distinction between explicit reasoning and intuitions. Intuitions haven't circumvented explicit reasoning, they're just relying on past deployments of explicit reasoning that have been cached in the brain. Whatever epistemological value is gained by explicit conscious reasoning might still benefit intuitions, having themselves been the product of past episodes of explicit reasoning.

For what it's worth, the neuroscience seems to support such a model. You might be interested in this study:


It suggests that damage to the orbitofrontal cortex (an area implicated in active decision making) results in the brain defaulting to learned behaviors and habits. We might also see intuitions as belonging to the same class of "cached-value" cognitive routines that are deployed in lieu of active decision making; the distinction seems very close to the one you are drawing here. An intuition is a micro-habit, a reflexive judgment, one that doesn't require orbitofrontal input.

This is supposed to all be friendly, but it leaves very little room to distinguish between "intuition" and "explicit reasoning" along epistemological lines, as your definitions would suggest. Intuitions are basically the way your brain compresses a lot of data for easy, rapid deployment; in other words, it's a data structure. That doesn't fundamentally change the rules of warrant, because more often than not the interesting justificatory questions concern the data itself and not the way it is being compressed. A claim stored on a flash drive isn't any more or less likely to be true than one carved in stone.

My point is that approaching the epistemological questions armed with the distinction between intuitions and explicit reasoning isn't going to get you very far. On the other hand, a naturalized epistemology lends itself quite nicely to the neuroscience. Among other things, it suggests an operational definition of expertise. An expert is an individual who has trained their intuitions so as to perform indistinguishably from individuals engaged in active decision making, as measured by orbitofrontal activity.

That has to be easier than defending a dissertation.

Howie Berman said...

Is the use of linguistic convention like Aristotle did a type of reliance on intuition?

Eric Schwitzgebel said...

Thanks for the comments, Bernard and Daniel!

@ Bernard: I agree that the empirical questions and the philosophical ones are interdependent. I also think the line between them is blurry. It's not going to be a straightforward matter of empirical study first, philosophical conclusion second. But neither will it be straightforward in the other direction either, I think -- the usual complicated tangle of things that I love!

Howard Berman said...

Is reliance on linguistic convention as Aristotle did a type of reliance on intuition?

Eric Schwitzgebel said...

@ Daniel: I basically agree with that picture. (Alison and I use the case of expert chess intuitions to make a similar point in our 1998 paper.) I don't think it's going to turn out to be entirely epistemically neutral, since there will be kinds of things we are trained on to the point of intuitive fluency and things we are not so trained on, and that's not going to be independent of their philosophical status; and (as the stone vs. flash drive example suggests) there will be different counterfactual sensitivities. Etc.

I do agree that any generalization of intuitions bad / explicit reasoning good or vice versa is going to be highly problematic for the reasons you cite (among other reasons). To get real juice, I think we'll need to drill down to the specifics.

Eric Schwitzgebel said...

Howie: I think it will depend on the details.

Scott Bakker said...

I think the term keeps us honest, a little 'from - x' to remind us no one *knows* what the hell they're talking about. Does Cappelen consider that maybe it was honesty that motivated its growth? Intuition as admission of ignorance.

It's certainly among the terms scheduled to be dismantled by cognitive neuroscience. Elements like it, vague, with a huge scope of application, generally tend to resolve into multiple complexes. Because of this, I think it pays to remember we are likely all bound to be surprised by whatever neuroscience describes.

When we talk intuition we are 'sourcing as sourceless' certain kinds of information. The thing is, it's all information, hanging sourceless! This is what makes it such an easy target for the kind of critique that Cappelen advertises (in his opening chapter at least). Turn over every rock in consciousness and cognition and you find sourcelessness.

Unknown said...

I think that even if you grant Cappelen his narrower definition of “intuition,” his claims about philosophy fly in the face of actual philosophical practices.

As I understand it, Cappelen defines “intuition” as a proposition that is supposed to be foundational in the same way that perceptual propositions are thought to be foundational. Then he argues that, when philosophers talk about intuitions, they are not talking about foundational propositions that cannot be defeated by further evidence and arguments.

This is a brilliant move: define “intuition” in such a way that it turns out not to refer to anything any philosopher actually does and/or says. But this move fails. It is clear to me that when Jackson claims that Mary learns something new upon her release, when Chalmers claims that zombies are conceivable, when Nozick claims that one wouldn’t want to be plugged into the experience machine, etc., these philosophers do claim to have foundational insight about the hypothetical cases they present. In other words, they don’t take the judgments they make in response to these hypothetical cases to be the sort of judgments that can be defeated by further evidence and arguments. Rather, they take these judgments as indisputable evidence or basic premises in their arguments.

Anonymous said...

Couldn't one argue that some philosophers still use their intuitions when it comes to something they have consciously considered. In the debate of rich vs sparse content of conscious experience, a lot of the examples do rely on an intuitive judgement as to what side of the debate you fall down on.

It may not be a snap judgement that you were talking about but it is still using intuition for philosophical debate.

Anonymous said...

Eric, I posted this over on the xphi blog, but it's awaiting moderation and I've had trouble with comments not appearing there recently, so I figured I'd post it here too.. -Max

Hi Eric,

I’m glad to see a post about Herman’s book on this blog. I think it is one of those rare books in philosophy that is right about almost everything.

Herman’s argument that philosophers don’t treat intuitions as evidence works just fine if you replace ‘intuitions’ with ‘snap judgments.’ Are you saying it doesn’t? Then which of the many philosophical arguments Herman reviews are such that the philosophers presenting them treat snap judgments as evidence? Or do you have other arguments and philosophers in mind? Which?

I think the only responsible way to reply to Herman’s argument is to argue that he has misunderstood the arguments he reviews, or has wrongly taken them to be representative, or has failed to consider other arguments that are equally representative. Saying that judgments that do not arise by explicit, conscious reasoning are essential to cognition (and so to philosophical cognition) is not a reply to Herman. Nor is saying that the psychology of philosophy is relevant to the epistemology of philosophy. Herman doesn’t deny these things.

He does say that xphi is a big mistake. That’s because he takes many xphi projects to assume, falsely, that philosophers treat intuitions as evidence. Many of them do assume that, don’t they?

At one point in the post you say that the issue of resolving conflicts between one’s explicit theory and one’s spontaneous judgments is “crucial” to philosophy. Is that an empirical claim about how philosophy is actually practiced? If so, what are the specific cases you have in mind? What does the claim mean, anyway? Is it crucial that, in cases of this kind of conflict, and on the way to resolving it, one treats the fact that some of the judgments are spontaneous as evidence for the truth of those judgments? Herman has an argument against the claim understood in that way. I don’t see that you’ve offered a reply.

Anonymous said...

Er, that was me, Max Deutsch, who left that last comment.

Scott Bakker said...

@Max: Thanks for that. It certainly clarifies things for me.

Eric Schwitzgebel said...

Hi Max: Here's the reply I left on the x-phi blog:

Thanks, Max, for your very helpful comment! What you say fits fairly well with how Herman has replied to me in email exchanges.

One apparent take-home message of Cappelen's book is that x-phi is a waste of time, if one is interested in the epistemology of philosophy. And since x-phi (broadly construed!) is our best attempt at empirically understanding the psychology of philosophy and Cappelen does not (that I recall, please correct me if I'm mistaken) propose an alternative method of empirically understanding the psychology of philosophy, then the reader is invited to the conclusion that empirical approaches to the psychology of philosophy are a waste of time if one is interested in the epistemology of philosophy. To my way of thinking that thought is so plainly wrong that we can can use it as a starting point for a reductio.

Consider Cappelen's treatment of trolley problems. He argues that since Thomson and Foot call our shared assessments of those cases into doubt and provide arguments for and against them, those assessments can't be Rock, and so they lack that feature of intuitions. This argument only works on the assumption that the relevant sense of "intuition" is so strong that the act of calling an intuitive judgment into doubt and seeking arguments pro and con undercuts its very status as an intuitive judgment. If one favors my approach to intuition instead, Cappelen's discussion here is mostly beside the point.

It seems to me quite there that there is *some* sense of "intuition" in which the conclusion that you should not push one person in front of a trolley to save five others is (for most people) more intuitively appealing than the conclusion that you should push the person; and it seems also clear to me that that intuitive appeal is not irrelevant to how attractive we find various moral theories. And from those facts, it seems to follow that studying the psychology behind that is not irrelevant to the epistemology of philosophy. Cappelen's terminological choices and argumentative focus masks these truths, I think.

Eric Schwitzgebel said...

Scott, "sourcing the sourceless" is a nice way of thinking about it, I agree.

Eric Schwitzgebel said...

Dan: I agree that a conclusion can still be felt to have an intuitive pull even after one has thought about it explicitly. It would no longer be an intuitive judgment in the sense defined, but we could say that it's "intuitive" in the sense that it's still true that it's the judgment you *would* arrive at other than by conscious explicit reasoning.

Hm, that isn't very clearly put. I might have to think through in a bit more detail how to mark with terminology the distinction between a token-case of a judgment arrived at other than by explicit conscious reasoning and a proposition of the type that would (for you or for some group of people) hypothetically be endorsed as a result of such processes.

Eric Schwitzgebel said...

Moti, here's my reply (also on the x-phi blog):

Thanks, Moti, for that comment. I'm somewhat inclined to agree with you, but I also think it depends in part on how strongly one reads "Rock". If it's a violation of "Rock" to think that such claims don't profit from any argumentative support and are supposed to be swallowed bald by everyone, I don't think Chalmers and friends would go that far. But that interpretation of Rock would make Cappelen's characterization of intuition very problematically narrow indeed.

G. Randolph Mayes said...

Eric, I don't find that I disagree with anything you say, but I'm not sure I understand why you like that definition. On the face of it, a definition of X as "Everything that is not Y" doesn't seem like a positive step toward identifying a concept with explanatory value. Everything you say about intuition seems compatible with Herbert Simon's definition of intuition as recognition. I think someone like Johnson-Laird, would want to say that when we reason consciously our mental models provide intuitions about validity, evidential strength and explanatory power. So even though you characterize your definition as broad, I think maybe it is not broad enough.

Eric Schwitzgebel said...

I'll agree with you about the last point, Randy. It's not as cleanly separable as I sketch it here (which is my complaint about overplaying the System 1 / System 2 dichotomy, in other contexts). But I disagree about your first point: Anything but X can be a very useful concept, even if it doesn't pick out a unified class. "Not a human being", for example, is a useful class that is negatively defined and diverse of membership but potentially useful.

G. Randolph Mayes said...
This comment has been removed by the author.
G. Randolph Mayes said...

That's a surprising disagreement! I'm not sure I buy it. I didn't deny any usefulness to definitions like that, just explanatory usefulness, and it seems to me that this is the appropriate measure of value to focus on here. After all, the best definition of intuition is not the most intuitively correct one, nor the one that secures a conclusion we've decided to defend. Rather, it's the one that satisfies the aims of a fruitful scientific theory. Dead right about overplaying the two system dichotomy, though. I think even Kahneman would agree with that.

Eric Schwitzgebel said...

You've probably thought quite a bit more than I have about the explanatory usefulness or not of negative definitions, so I'd be interested to hear more about your views on that. Here's one possible explanation that uses "not human": Q.: Why was Jim not permitted to legally marry Obleo? A.: Because Obleo is not human. (Obleo is Jim's motorcycle.)

G. Randolph Mayes said...

(Delete previous comments because of confusing typos, sorry.)

Well, it's something I need to think more about, now. But you're at least right that we can easily construct systems of rules in which negative definitions have explanatory value. [General method: Let X = "All things that are not Y." Then create a rule of the form If X (or ~X), then Z (or ~Z).] And since we can construct rules like that it follows that there can be (and in fact are) cognitive agents whose behavior can be partly explained in that way.

To perhaps strengthen your point, we can apply this kind of explanation directly to the natural world. Let X be "Any creature that is not the mother." So, Why can't Jim comfort the baby? Because Jim is X.

But notice in this case the explanatory value of the answer depends on the assumption that there are properties necessary and sufficient for affording comfort to the baby, and ~X uniquely possesses those properties. So it's really ~X, that provides us with the causal model, not the vast majority of things in the universe that are X. (The same can be said of your Obleo example.)

So I will guess that this is the case with intuition as well. If X = intuition = "judgments not resulting from reasoning," you'll get explanations for which the relevant causal properties are those of reasoning, but not of intuition.

This is a very nice distraction from grading finals. Thanks, Eric.

Eric Schwitzgebel said...

You might be right about that, Randy. I confess, though, to a certain suspicion that there's not a sharp difference between negative and positive properties.

G. Randolph Mayes said...

Well, I think more than one of my naturalism students would be ashamed of me if I argued with you on that point.

Unknown said...

I haven't been through the comments yet, so forgive me if any of this is repetitive or has been addressed. You open by stating you have a simple argument against his view, and then lengthily if eloquently describe your own view on intuition.

This was somewhat misleading to me as a reader, as he's making a claim using his definition and you're telling us your definition. This makes your simple argument a semantic one, and not really against his. For example, if I were to make the claim that Apples are delicious whilst specifying HoneyCrisp within the text, you could tell people you had an argument against the taste of Granny Smith apples - making valid points that don't really oppose the gist of my work, just the headline of it.

Secondly, I certainly don't define intuition the way you do. What you describe as being intuition seems to me to be quite simply the path of least resistance when it comes to the electrochemical magic of thought. Snap judgments use the most well-trodden paths of definitions when using communication, being rejected only as further details are revealed.

Intuition on the other hand, is knowing without conscious thought, not deciding without conscious thought. You learn from what you already know, it is your inner tuition.

Your notice of 'professional philosophers' stuck out to me also, I didn't know that was a valid career path. I'll have to look into it.

Eric Schwitzgebel said...

Grizwald: Thanks for the thoughtful comment!

On your first point about whether the argument is simple: You've nicely expressed how that claim might be misleading. Maybe I shouldn't have put it that way. Still, it's kind of a simple thought: (1.) Think of "intuition" very broadly. (2.) Conclude that it's necessary for all cognition and thus philosophy too. Maybe the sub-argument for (1) is complex, though.

I'm not sure I understand your concern about characterizing intuition the way I do. Could you amplify a bit?

"Professional philosopher": I guess I mainly meant to refer to philosophy professors. I can see how one might think there's an important difference, though.

Unknown said...

Yeah, sorry I just stumbled back on this comment. Forgot to notify of replies.

How I define intuition is separate from snap judgments. Take the Eureka guy, Archimedes. For me, intuition was the little voice that told him - quit thinking about it, you could use a nice bath.

patrick said...


I think you are being a little too quick with Cappelen's argument. (FWIW I think his critique of x-phi is based on a silly mistake--but not the sort you are pointing out--and I agree with you that empirical work into philosophical reasoning is needed.)

First, intuition as “snap” judgment is vague. What counts as “snap”? Say you specified snap judgments in terms of specific cognitive processes in a scientifically respectable way. There is still a big problem, in that you have not yet shown they are actually at use in philosophy in an epistemically significant way (not just that they aren't epistemically relevant, as in the worry you raise, but that they aren't relevant in actual philosophical practice).

Consider the fact that philosophers never actually test to make sure people are giving us “snap” judgments when we ask about cases (in the sense of specific cognitive processes that such an account would have to fill in). For instance, when we ask about (say) Geitter cases do we discount any responses that are not “snap” judgments? Do we ask the participants to not think about it too hard or for too long, lest they produce conscious and explicit judgments about the case which we are not interested in when we query for intuitions? We have never tested for such specific types of judgments, and traditional pen-and-paper x-phi doesn't get at them, either (this is where neuroscientific work on the processes comes into play, I agree.) But this is, in sum, a criticism Cappelen routinely makes in the book—you have an account of intuition, sure, and maybe it is perfectly psychologically accurate as well. But you have not shown those processes in particular are at work in philosophical methodology, and at work when people talk about and employ “intuitions.”

Remember also that his critique of intuition in analytic philosophical methodology is also broader than the failure of any one account of intuition (a point which some comments above seem to have missed), for it still stands that there are a huge variety of uses of “intuition” in philosophy that are inconsistent with one another (as he argues in the first half of the book), so even if one account in particular turns out correct—your own, say—there still remains the problem that many philosophers who have used intuitions as evidence have been using some other types of states than what your account picks out. So if you are right, the majority of people who have theorized about intuitions are still mistaken about what they have taken those states to be, which has significant consequences for their role as evidence in those works.

Eric Schwitzgebel said...

Thanks for the thoughtful comment, Patrick! What you say fits pretty well with what Cappelen has said to me in email, too, and what one can get from the arguments in his book.

Probably, my repeated use of the word "snap" in the post was misleading. Intuitive judgments, on my account, don't have to be snap; it's just *clearer* that they're intuitive when they're snap. Not that you're necessarily portraying me otherwise; just to be clear.

It is true that my account of intuition is underdeveloped. I feel pretty confident that there's a reasonable sense of "intuitive"/"counterintuitive" on which pushing the fat man in a trolley problem is "counterintuitive" and on which knowledge is "intuitively" not present in an Gettier case, etc., and that the harmony of a theory with this intuitiveness both is (sociologically) and ought to be (normatively) a mark in its favor. But Cappelen and Max Deutsch (and maybe you) might not find that claim as plausible as I do. And I confess that to fully back it up I need more depth in my account of the psychology and epistemology of intuition.

The guiding thought of the post, though, was that Cappelen's characterization of intuitions is considerably narrower than anything in the ballpark of what I would accept. On their face, his arguments will at least need to be adapted to work against someone with a very broad sense of what counts as "intuitive".