One more take on the role of intuition in philosophy

by Massimo Pigliucci

www.pitt.edu/~jdnorton

Julia — over at her Measure of Doubt blog — has reopened our discussion concerning the use and misuse of intuition in philosophy, which originally started in an episode of the Rationally Speaking podcast (the first two rounds of our discussion after that can be found here and here). Since I chided her for cherry picking examples in her critique, this time she relies on a (as yet unpublished, but available as an online preview) paper by J.R. and J.R.C. Kuntz, to appear in the Review of Philosophical Psychology. (Wow, a whole journal devoted to that? At any rate, here is the link to the article, since Julia provided her readers only with access to the abstract. You will need academic library access if you don't want to purchase the paper.)

I will first stake out my own view about the role of intuitions in philosophy, then briefly comment on Julia’s post, and finally go into some details in the Kuntz’s paper, which is characterized by modest results and a number of methodological flaws (but which nonetheless does manage to make a couple of good points to which philosophers should pay attention).

As the Kuntz’s make clear, intuitions can play two distinct roles in philosophy (or, for that matter, in anything else, including science): one in what philosophers of science call the context of discovery — providing the starting point from which logic (or empirical evidence, in the case of science) take off — and one in the context of justification, i.e., as data used to test hypotheses. The former is not only perfectly justified, but in fact inevitable. Galileo’s and Einstein’s thought experiments began as intuitions, and so does much valuable research in science and scholarship in philosophy. Hilary Putnam’s famous thought experiment about a twin earth can be used as a good exercise to make explicit, and if necessary to criticize, reject or modify, our thinking about language and its referents. However, intuitions cannot and should not be used as data to test hypotheses or draw conclusions, because they are not epistemically reliable — hence the silliness of David Chalmers’ zombies, which tells us precisely nothing about consciousness, as I have argued before.

There is a second important distinction that needs to be made when talking about the role of intuitions in any specialized field, since failing to acknowledge its import makes for some unnecessary confusion in both Julia’s essay and the Kuntz’s paper. I am referring to the distinction between the intuitions of professionals and those of lay “folks” (to use the Kuntz’s terminology). Pace much (again, in my opinion, hyped) clamor about experimental philosophy (an oxymoron actually referring to psychological research on concepts relevant to philosophy), it matters exactly zero what folk intuitions about free will, twin earths, Chinese rooms and the like are, just as it matters zero what folk intuitions about logic, mathematics, physics or the game of chess are to the practices in those fields. As for the intuitions of experts, there is plenty of cognitive science literature (developed from studying chess players, math teachers and nurses, among others) showing that intuitions in one’s domain of expertise become increasingly reliable the longer one has been practicing in that domain. (Interestingly, it takes several thousands of hours of practice to develop intuitive skills equivalent to those of a good chess player, and thousands more to achieve equivalency with a grand master.)

Now back to Julia’s essay. Despite my warning about cherry picking, she does it again right at the beginning, mentioning Chalmers’ zombies and a pretty risible claim by G.E. Moore about the desirability of a clean over a dirty planet even if there were nobody around to enjoy the landscape. By the same token, I could easily compile a large list of blunders by scientists and declare the whole enterprise in deep trouble (actually, someone has already done the compiling, see this delightful book about science’s greatest mistakes).

Julia then cites one of the major findings of the paper, that 51% of the philosophers sampled by the Kuntzs (see below for methodological issues) think that intuitions are useful in philosophical justification. She however waits until the very end of her essay to mention that a whopping 83% of respondents thought that intuitions are useful in discovery (not justification) and does not report at all that 70% of philosophers think that intuitions are not essential to justification. Julia, however, does acknowledge that philosophers of science seem to be particularly skeptical of the role of intuition in the context of justification. Kudos to ourselves, my dear colleagues.

Okay, to the meat of the matter now: the Kuntz’s paper itself. Other than the findings already mentioned, the most visible result is that philosophers seem to prefer two of the seven accounts of intuitions that were provided to them by the researchers, while they largely dislike two more. He are the Kuntz’s seven conceptions of intuitions, as presented in their survey:

1) Judgment that is not made on the basis of some kind of observable and explicit reasoning process.

2) An intellectual happening whereby it seems that something is the case without arising from reasoning, or sensorial perceiving, or remembering.

3) A propositional attitude that is held with some degree of conviction, and solely on the basis of one’s understanding of the proposition in question, not on the basis of some belief.

4) An intellectual act whereby one is thinking occurrently [sic] of the abstract proposition that p and, merely on the basis of understanding it, believes that p.

5) An intellectual state made up of (1) the consideration whether p and (2) positive phenomenological qualities that count as evidence for p; together constituting prima facie reason to believe that p.

6) The formation of a belief by unclouded mental attention to its contents, in a way that is so easy and yielding a belief that is so definite as to leave no room for doubt regarding its veracity.

7) An intellectual happening that serves as evidence for the situation at hand’s instantiation of some concept.

A majority of respondents liked (1) and (2) and dislike (6) and (7) which, despite some quibbling on the part of the Kuntz’s, actually indicates that philosophers tend to have somewhat consistent views of what intuitions are. However, the problem here is that it is not at all clear how, at a conceptually deep level, these ideas of intuition are in fact different from each other (they are certainly not identical, but several could be taken to be overlapping, and indeed even almost identical once the complex and sometimes obfuscatory language is stripped away).

So far, the results of the study aren’t that bad for philosophers: a majority, especially of philosophers of science, thinks that intuitions are useful in discovery more than justification, and there is some agreement about what intuitions actually are (though I’d be curious to see what one would find by asking scientists about the type of intuitions that inform their discovery phase). The real problems with the paper are methodological.

To begin with, this was a voluntary online survey. Bad idea. The sample (which isn’t that large to begin with, only 282 people) is very likely to be self-selected in terms of interest in the topic and other criteria (such as level of online activity, which is still relatively low in the humanities, especially among older faculty) and areas of expertise within philosophy (the Kuntz’s do address the latter problem, but claim that their sample is not likely to be biased based, unfortunately, on yet another online survey, and one conducted by a philosopher — Chalmers — not a social scientist!).

Moreover, we discover that an (unreported) number of participants had not actually finished their PhD, raising the question of the extent to which this is in fact a survey of professionals (see my comment above about how many thousands of hours are necessary to develop expert intuitions). To complicate things further, some sub-areas of Philosophy included in the analysis were barely represented at all (e.g., aesthetic, n=3; postmodernism, n=2; education, n=3; feminism, n=1; philosophy of literature, n=1; philosophy of mathematics, n=5; Philosophy of religion, n=4; law, n=3; and so on).

The descriptive statistics of the survey are not problematic (they simply report the rankings of the various options by participants, broken down by categories such as gender, geographical area, subfield of interest, etc.). But then we get to the bivariate correlations among variables (measured as Spearman’s non parametric rho), such as the relationship between the importance attributed to intuition and the preference for one or another of the seven types of intuitions from the menu described above. While all correlations discussed by the Kuntz’s are statistically significant (as much as I personally put little faith in p-value based statistics), the effect sizes seem pretty tiny. A good number of the coefficients are in the 0.15-0.20 range, indicating very weak correlations. It would have been nice to know what the percentage of explained variance of one variable by the other was, the so-called coefficient of determination, but that’s not available because the Kuntz’s didn’t do a parametric analysis, which is necessary to estimate these coefficients.

Then again, didn’t they? Toward the end of the results section they mention an analysis of variance, which is a parametric procedure. So, then, were the data approximately normally distributed, so that ANOVAs were possible? But if so, why use Spearman’s rho for the correlations, since non parametric statistics are known to be less powerful and ought not to be deployed if the data even approximately satisfy parametric assumptions? And why don’t the authors report the results of the ANOVAs in a table, instead of simply briefly summarizing their results verbally?

In the end, I don’t think we’ve learned much from the Kuntz’s paper about philosophers and intuitions, other than being reminded of the valuable conceptual point that everybody interested in either using or studying intuitions ought to make very clear the distinction between the context of discovery and the context of justification. And the paper certainly lends very little support to Julia’s sweeping critique of "philosophers" for using an epistemically unreliable method.

Here, I am with Henri Poincaré (quoted by the Kuntz’s) when he wrote back in 1908 that “It is by logic that we prove, but by intuition that we discover.” Which is true also for formal logic and math. Substitute “empirical evidence’ for “logic” in the quote, and you get science.