Cappelen between rock and a hard place

In order for Herman Cappelen to argue in his Philosophy Without Intuitions that philosophers have been on the whole mistaken in thinking that we actually use intuitions much at all in our first-order philosophizing, he must attempt the task of characterizing what something must be, in order to be an intuition.¹ I truly sympathize with the frustration he evidently feels at wrangling with that task, because I’ve felt the same in my own project critiquing what I do take to be a fairly common practice in contemporary philosophy that we often gesture at when we speak of intuitions. For the literature on intuitions can be a total mess on even the most basic questions about what intuitions are: beliefs, or sui generis seemings? Special intellective capacity, or continuous with ordinary cognition? Distinctive phenomenology, or not? On so many questions, one finds highly diverse sets of answers on offer. So when taking aim at the entire category of intuitions—either to critique what you think is there, or to argue that we’re mistaken in taking there to be such a category active in philosophy in the first place—one faces a tricky balancing act. Characterize intuitions too specifically, and many philosophers will just say, hey, that’s not what I’m using. Characterize them too generally, and you can find yourself with too little material to work with to paint a clear target for your critique.

Despite my sincere sympathies for Cappelen, I nonetheless object that he has fallen afoul of the first danger, and badly so: he construes intuition so narrowly that I’m not sure that it actually applies to any extant accounts in the literature. Now, of his three “features of the intuitive”, I think he’s right in suggesting that considerations of distinctive phenomenology and conceptual competence are fair game, as they are set forth and discussed at length in the literature. They are both far from universally endorsed, but he is arguing against a disjunctive formulation—that none of his paradigm cases have any of the features—so he rightly errs on the side of inclusiveness. If the total list of features he considers were sufficient and well-motivated, then this crucial step of operationalizing intuitiveness would be a success.

Except, it’s not. The problem concerns the third and last feature, “Rock”. He claims that it is a special kind of epistemic status—“special” recurring frequently in the book, whenever he discusses it—one that is immediate, “glowing”, “privileged”, and highly controversial. It’s a kind of justification so fundamental, unshakeable, indubitable, that an author’s saying pretty much anything at all in defense of p, even expressing any hesitance about whether to endorse p, or about what p’s significance might be, is a clear sign that p lacks Rock. For example, here’s how Stewart Cohen’s discussion of the lottery cases reveals them as totally non-Rock: “The claim that philosophers rely unquestioningly on default justified, immediate reactions to these cases is simply unfounded: the moment the cases are presented, they are questioned, they are not endorsed, and they give rise to puzzlement. This response indicates that the Rock feature is absent.” (p. 165, emphases original). Rock status must be special indeed, to be inconsistent with such things, since I can’t think of any of our cognitions that are like that. Even the cogito can produce puzzlement, after all!

That Rock is so freaky and fragile, and so completely absent in all of the case studies (as he demonstrates compellingly), perhaps should have been a sign not to attribute Rock to intuition theorists in the first place. Cappelen adduces almost no textual evidence to support his claim that “most intuition-theorists” take intuitions to have this feature. One of the only texts he adduces is, in fact, mine:

“In the extant practice of appeal to intuitions as philosophical evidence, one cites one’s application or withholding of a concept from a given case, usually a hypothetical one, in defense of (or in order to attack) a particular philosophical claim. Such citations thus are meant to carry argumentative, evidential weight, but one is not usually required to offer any further argumentation for the intuition itself.[…] Although they are used to provide evidence, one does not, and need not, provide further evidence for them.” (p. 118).

He selects this passage from Weinberg (2007), p. 320, but I can’t help but wish he had paid more attention to my next sentence: “However, they are not generally taken to be incorrigible or indubitable…. We may choose not to endorse an intuition if the balance of evidence speaks against it, or if one comes to think that the intuition was not formed in a sufficiently truth-conducive way” (pp. 320–321). So on my characterization, their status as a justifier-in-no-need-of-justification is only prima facie, and we should expect intuitions to sometimes be challenged, doubted, counter-argued, shored up, or weighed in the balance. And so it should be pretty obvious that if Rock’s absence is indicated by any questioning, puzzlement, being argued about, or even being discussed at all (p. 121), then my characterization of the appeal to intuitions bears no resemblance to Rock.²

What my characterization does closely resemble, however, is a much more ordinary epistemic status that Cappelen contrasts with Rock: p’s starting in the “common ground”, such that one can legitimately deploy p as a premise without explicitly adducing further evidence for p.³ Cappelen clearly thinks that CG is an ordinary, unmysterious kind of status, and he is surely correct. In fact, it seems to me almost exactly the sort of uncontroversial epistemic status that should be a first hypothesis about what someone means when they write about propositions that are “used to provide evidence”, even though “one does not, and need not, provide further evidence for them”! I say “almost”, because Cappelen (as perhaps befits a philosopher of language) cashes CG out in terms of one’s discursive setting, e.g., “taken for granted among conversational participants” (p. 119), and not in epistemic terms. So we need to add one small, innocuous strengthening to get more precisely the status at play in this particular debate about methodology, to distinguish evidential from non-evidential CG. Non-evidential CG includes being granted arguendo, or assumed hypothetically to establish entailments. In such cases, whatever it is that makes p CG does not of itself provide any evidence for p; it need not incline any participants to endorsing p; and it is not undermined by any introduction of evidence against p. (Consider assuming p in order ultimately to disprove it in a reductio.) I’ll say a proposition p is Evidential Common Ground when all parties take themselves already to have evidence for p, in a manner substantial enough to license p prima facie as a justifier without need of further offered justification.

Looking back at my text⁴ with that in mind, there’s clearly no reason to take it to concern an epistemic status more demanding than ECG. Although some intuition theorists have also defended some other interesting epistemic statuses for intuition, such as empirical incorrigibility, these are controversial minority views. And even those are still a far cry from Rock. In fact, I don’t think any intuition theorist endorses anything even remotely like Rock. Even if I am wrong about that, and someone somewhere does defend something like Rock, it is still clearly not a commitment widespread among intuition theorists.⁵

So Rock, on Cappelen’s understanding—it’s his notion, so we should defer to his understanding—is so spectacularly demanding, that it is unsurprising that the case studies lack it altogether. For that very reason, it is of no consequence that they do so. What Cappelen argues does not exist, was something that we had no reason to think that any intuition theorist believes to exist; at a minimum, it is something it would be plainly in error to attribute to intuition theorists collectively. This suffices to show that Cappelen’s attempted refutation of the argument from philosophical practice is in vain: his mis-operationalization of intuitiveness sets radically too high a bar, and so we cannot infer anything of interest from the fact that nothing clears that bar. In coming to understand Cappelen’s mistakes here, though, we can point towards how we might better examine questions of the where—and whether—of intuitions.

My diagnosis of Cappelen’s Rock-headed error: a toxic interaction between an inappropriate methodology and a substantive misunderstanding. First, his method is to survey a choice set of paradigm cases of alleged intuition usage, while requiring that we be able to tell from what is manifested plainly in those texts themselves whether the intuition features are present. Rock has this methodological virtue—it is enough for there to be any significant discourse in a text either in support of or hesitant about p to indicate that p is not Rock. But many epistemic qualities are not like that. Just to illustrate: some claims in philosophy texts may be ECG as widely shared experiences requiring no special education, whereas some others are ECG due to special training presumed shared with one’s readers, like logic or the history of philosophy. Neither of those epistemic qualities will typically be detectable in the text itself.⁶ Determining whether a claim had such a quality must recruit substantial extra-textual information, such as knowing what’s in a standard PhD curriculum. They would thus not be discernible using PWI’s methods. Those methods impose potentially misleading restrictions on what sorts of epistemic features can be considered.

Now, as Cappelen insists, there must be something more than ECG at stake with intuitions—oodles of propositions are ECG, in philosophy and elsewhere, far beyond what intuition theorists take to examplify intuitiveness. I diagnose that Cappelen commits a substantive error in spelling out that ‘something more’ in fundamentally the wrong kind of way. Consider one of his earliest glosses on the special epistemic status: “Intuitive judgments about cases are treated by philosophers as having a kind of epistemic status, which is different in kind from that of typical judgments based on perceptual input, memory, testimony, or those reached by inference from such judgments.” What comes vividly into view here is that Cappelen has mistaken intuition’s epistemic ‘something more’ as a special kind of epistemic status, when it is better understood as a special basis for what is an ordinary kind of epistemic status. For one doesn’t start theorizing in terms of intuitions because one thinks some special unassailable, unquestionable, unpuzzlable epistemic status is needed for philosophizing. Again, I just don’t know anywhere in the intuitions literature where one sees that. What one does see are claims to the effect that philosophers have a set of propositions taken to be ECG, playing important argumentative roles, but for which those standard explications of ECG status (perceptual input, etc.) nonetheless seem inadequate. They don’t need a “special”, “privileged”, “glowing” status—they just need some appropriate means of becoming ECG in the first place. A better gloss would be: “Intuitive judgments about cases are treated by philosophers as already possessing substantial justification, but where that justification is not itself wholly based on perceptual input, memory, testimony, or inference from such judgments.” This is still rough—the “wholly” is meant, for example, to cover the possibility that the case judgment may partially involve such sources but still go substantially beyond them—but it illustrates the right direction to go in, away from special statuses and towards justificatory bases.

One juncture where Cappelen could have gone down that more fruitful path is p. 112, with Rock put forward along with non-inferentiality and non-experientiality. These latter features are not only much better attested to in the literature than Rock, they also point towards the issues that actually drive that literature, when they exclude various standard bases for being ECG. On the one hand, we take these case verdicts to be ECG. On the other hand, it seems that none of the standard ways we have for articulating what makes something ECG can do the job for them. Our first-order confidence is in tension with our second-order ignorance.

That tension provides a rich starting point for metaphilosophical inquiry. The manifest inadequacy of Rock should teach us not to try to start by extracting the features posited for intuitions themselves from the literature. Any notion of intuition shared across intuition theorists is better thought of as an epistemic functional notion: whatever it is (if there is anything) that does the apparently needed work of legitimating such claims as ECG, given that standard sorts of justificatory bases are inadequate. The key epistemic quality of the intuitive is thus similar to such qualities as being ECG due to shared philosophical training: we cannot expect to detect them while zooming in tightly on the texts themselves. Cappelen’s textual strategies are thus inappropriate to what turns out to be the real job at hand.

A proper investigation of philosophical methodology should throw Rock out the window, and focus instead on clarifying that functional role that intuitions-whatever-they-may-be are supposed to fill. An intuition critic like me might argue that nothing at this time fills that role successfully. An intuition meta-critic like Cappelen could argue instead that that epistemic functional role itself is at worst fairly small and insignificant. That should still be a meaningful goal to pursue, but not using the textual strategies of PWI. Instead, to chart out the extent of the epistemic functional role alleged for intuition, we would need to go carefully through the case studies, considering their argumentative goals and resources, both explicit and presupposed, and show that for each paradigm case verdict p, one of three possible intuition escape clauses can be found:

• (i) p is ECG, but in a way that is unproblematically explicable; or

• (ii) p is not ECG, because it is not taken as justified at all; or

• (iii) p is not ECG, because it is argued for directly in the text in a way that requires no appeal, even tacitly, to the functional epistemic role of the intuitive.

Cappelen mostly doesn’t do much to explore escape clause (i) in PWI—he is trying to show that nothing is Rock, and that doesn’t require anything like sorting out different ways that a claim can be ECG. For example, he almost considers the question in his discussion of Goldman’s fake barns, and how the case verdicts there (taken to be CG) might be working differently than claims like “Barack Obama knows that he was born in America” (pp. 172–173). But he only argues that the intuition features are not present in those cases, without actually considering the important way in which they really are different from that claim about Obama: it is much easier to articulate ordinary sorts of bases for the Obama claim. E.g., basically everyone who was not a foundling knows at least what country they were born in, and Obama was not a foundling.

He makes a similar move when considering Thomson’s violinist, and the verdict that it is impermissible to force a person to remain hooked to the violinist. Since Rock is so weird and controversial, he just shifts the burden of proof on the intuition theorist to show that Rock is needed for Thomson to be justified in that verdict (pp. 154–155). When he does offer a quick sketch of what Thomson’s unstated justification for that verdict might be—basically, that it would be very unpleasant for a person to be confined in that way—it doesn’t go well. He recognizes an objection (p. 156n13) that this would obviously be a poor way for Thomson to justify that verdict, since if she could justify the verdict that way, she could use that argument to end the entire abortion debate directly, without needing to consider all the strange hypotheticals as she does. (Another way of seeing the same point: no one in this debate would think that that kind of argument could be in CG to settle this question, since pro-lifers will not acknowledge that as a good argument. That is why Thomson needs a case-based argument in the first place.) Cappelen shrugs off the objection, and says, well, then “Thomson’s argumentative strategy is problematic.” It is worth noting just how odd Cappelen’s move here is. I would think that if we’ve shown that Thomson’s argument doesn’t rely on intuitions only so long as we interpret her as making a lousy argument, then probably we should modus tollens and recognize that what we’ve actually shown is that Thomson’s argument does rely on intuitions. Anyhow, the important moral here is that escape clause (i) is mostly not considered by Cappelen, and it seems that in many cases here it would be hard to pursue and still do justice to the authors’ clear argumentative goals.

Now, I think that escape clause (i) could apply in some of the cases that involve commonplace linguistic phenomena. Burge and Perry may be the best candidates for this; Cohen’s lottery cases, too, except that author apparently did not take himself to be appealing to such linguistic observations, but rather on his own personal take on the case and how it struck him.⁷ But it is hard to see how to even begin to tell such a story for cases like the fake barns and runaway trolleys and zombies. Philosophers often have good reason to be interested in highly novel cases, far beyond anything we might have antecedently observed.⁸ They thus are going to be very hard to tell a standard sort of validating story about.

Escape clause (ii)—that p is not taken as justified—will apply when the cases are not asserted at all, and there are a number of places where Cappelen denies that the famous case verdicts are endorsed by their authors, such as the lottery cases or the trolley cases. I’m pretty sure that he is simply wrong about those cases. He rightly observes that those texts are primarily looking to put forward puzzles to be solved. But if we take those authors not to be endorsing their proferred verdicts, we will miss out the way in which they think, first, that they really have identified a puzzle needing to be solved, and second, that they have figured out some clever solution as required. In terms of challenging Rock, there is no difference between being puzzled by a case verdict, and using it to raise questions (i.e., wondering about its implications, while taking it to be true), and being puzzled at it, and questioning the verdict itself (i.e., wondering if it’s true at all or what could justify it). All flavors of puzzlement preclude Rock. But escape clause (ii) can apply only with the latter sort of puzzlement. And there do not seem to be much of that, if any, in the case studies. At a minimum, it will not be enough to show it simply by noting the existence of questioning, discussion, or puzzlement regarding p.

One might hope that clause (iii) can be demonstrated in nearly as straightforward a way as rejections of Rock: find some argumentation for p in the text, any argumentation at all, and that escape hatch pops right open. But in fact it is nowhere near that easy. Of course, arguments for p often do preclude p’s being ECG (or there wouldn’t be an escape clause (iii) at all). Importantly, there are also other common kinds of argumentation for p that need to work with some p’s ECG status starting already in place. I’ll discuss just two such kinds here.

First, intuition is often compared to perception in that both are defeasible and corrigible, and also sufficiently epistemically independent from our theories that they can legitimately provide justification for those theories.⁹ Thus, as noted above, one might sometimes have an intuitive ECG proposition and nonetheless have reason to fear that various sorts of defeaters may arise for it. Accordingly, one may provide some argumentation as preemptive countermeasures against those defeaters. Thomson’s appeal to the informally-polled agreement of her colleagues regarding the trolley cases (p. 159) should be understood in this way: she is wisely anticipating the possible defeater that her take on these cases is idiosyncratic, or driven illegitimately by her own theoretical considerations. Much disputation about phenomenal zombies is also like that, concerning the appropriate circumstances for intuiting p’s possibility really revealing p’s being metaphysically possible; this also explains why discussions of the case itself often entwine with metaphilosophical arguments about it (p. 183). The overall shape of these arguments require taking something as providing substantial justification for p, and then the argumentative work serves to keep that justification undefeated and in play. We do not, in such cases, have an escape from needing that intuitive functional epistemic role, precisely because that something is still required.

The second ECG-friendly form of argumentation are textual cues for novel applications a source of evidence. Sometimes when you want someone to point their instruments in a new direction, you have to give them guidance as to what to look for. Compare to how you may have to describe what sorts of structures to look for when gazing through one’s microscope at a new specimen, or instruct someone viewing a Necker cube how to manipulate their visual attention to ‘flip’ the percept’s orientation. “Do this, so you can see that”—obviously the epistemic role of the microscope or of one’s perceptual capacities is not avoided by this kind of argumentation, but amplified by it.

Because philosophical thought-experiments are often both highly novel and rather convoluted, we should expect widespread cuing in philosophical practice. The Truetemp case is a nice example.¹⁰ When Lehrer writes that Truetemp “has no idea why the thought occurred to him or that such thoughts are almost always correct. He does not, consequently, know that the temperature is 104 degrees when that thought occurs to him,” this can look like an argument from premises to a conclusion, if one is not tracking the overall dialectical situation correctly. It is better understood as Lehrer helping draw our attention to what he takes to be aspects of the case that will produce the relevant cognitive response in us. He is steering us towards what he takes to be the proper viewing conditions for the case, but not offering considerations that bypass any such viewing.

Similarly, Cappelen writes that Lehrer “goes on to modify the scenario” in various ways, including making it such that “even the doctor who introduced the device into Mr. Truetemp is ignorant of its effects and reliability. Why would Lehrer go on to develop ways of reinforcing the conclusion, if he wasn’t in the business of giving arguments for it?” (p. 170). This degree of argumentation surely precludes Truetemp’s being Rock, but of course we see now that that doesn’t matter. We should yield to the rhetorical question—Lehrer is indeed offering considerations on behalf of his desired verdict—but once we interpret those arguments in the right way, we crucially need not see him as arguing from his internalism to that result. In fact, it’s of crucial importance that we not read him that way, on pain of rendering his argument viciously circular.¹¹ What Lehrer is clearly not doing, with this reinforcement, is offering some extra bit of evidence for the claim that Truetemp doesn’t know, as one might do when trying to strengthen the premises in an argument for an inferred conclusion. He is, rather, doing further work to prompt the intended intuition in the reader. So long as he can elicit in you some case of highly reliable belief formation whose opacity causes you to take it not to be knowledge, then he has the illustrative example he seeks. Lehrer is (as Cappelen correctly suggests) arguing for the desired verdict in the Truetemp case, but fortunately not (as Cappelen incorrectly takes it) in a way that would invalidate his goal of using it in his argument against the externalists. As a result, his argument presupposes the existence of the whatever-it-is that is the basis of the justification for such case verdicts.

Cappelen’s program is thus stuck between the patently incorrect characterization of intuitions in terms of Rock, and the methodological hard place of sorting out carefully when any of the three escape clauses might apply to the case studies. I don’t think that either will ultimately work, but the hard place approach at least holds out the prospect of some success. Moreover, whatever is uncovered in such an investigation would help us learn more about the epistemic functional role of the intuitive. I will close by noting that, in the meantime, the objectives of many experimental philosophers interested in intuitions¹² are untouched by PWI’s arguments. Both ‘negative program’ and ‘positive program’ experimental philosophers care about exactly this question of what might be the whatever-it-is that may, or may not, successfully license verdicts like those in the case studies as ECG. That’s still a very open question, so far as I can tell, and far from being a “big mistake”, such psychological work is proving a useful tool in exploring the legitimate extent of this presupposed source of justification.