How To Think About Philosophical Methodology

Abstract

Experimental philosophy, at least in its ‘negative’ variety, has standardly been portrayed as presenting a dramatic challenge to traditional philosophical methodology. As such, it has prompted a large variety of counter-arguments and defenses of intuition. In this paper, I argue that many of these objections to experimental philosophy rest on various oversimplifications that both experimental philosophers and their opponents have made regarding intuitions and philosophical methodology. Once these oversimplifications are abandoned, I argue that the experimentalist critique of current philosophical methods becomes somewhat less dramatic, but also much less open to objection.

Experimental philosophy poses a unique challenge to the way we think about philosophical method. As empirical work has uncovered ever more sources of unexpected variation in intuition, it has been tempting to draw the conclusion that intuition is an unreliable source of evidence for philosophical theory building, and that it should therefore be abandoned. At the same time, however, it is unclear how philosophy could proceed in the absence of intuition; how, other than via intuition, can we justify our most basic beliefs about morality, or about logic? Unsurprisingly, many philosophers have resisted the radical experimentalist conclusion that intuition must be rejected, offering a number of convincing arguments which aim to show that the experimentalist findings pose no threat to traditional philosophical methodology.

I argue for a more moderate position on the ‘intuition debates.’ Both sides, I claim, have tended to operate with an oversimplistic picture of philosophical method. They assume that there is a single, unified epistemic source called ‘intuition’; they assume that the central method of philosophy involves the use of intuition to test proposed philosophical analyses, via the so-called method of cases; and they assume that an evaluation of intuition’s suitability for this project involves determining whether intuition successfully generates ‘standard’ epistemic states such as justification or knowledge. All of these assumptions are misleading; worse, they are misleading in a way that disadvantages the experimental philosopher. I argue that, once we adopt a metaphilosophical picture that abandons these assumptions, it becomes quite clear that the criticisms posed by experimental philosophy do have serious implications for philosophical practice. They do not, however, imply that we must give up intuition entirely; nor do they demand any truly radical revisions to our basic philosophical methods.

The Experimentalist Core Argument

The subdiscipline of experimental philosophy centers on the empirical study—usually via survey methodology—of ‘intuitive’ judgments on philosophical cases. The earliest and most well known of these studies, published by Weinberg et al. 2001, provided evidence that intuitive reactions to Gettier cases varied across cultures. By surveying undergraduate students with different cultural backgrounds, WNS (as the trio are often labeled) found that students with East Asian backgrounds were much less likely to provide the ‘standard’ philosophical judgment—instead, they frequently judged that the Gettier subject ‘really knew.’

Other experimental studies quickly followed, revealing numerous unexpected features of non-philosophers’ philosophical judgments. Experimental philosophers uncovered more instances of cross-cultural variation (Nichols et al. 2003; Machery et al. 2004; Ahlenius and Tännsjö 2012); but they also discovered that ‘folk’ intuitions varied as a function of socioeconomic status (Haidt et al. 1993; Nichols et al. 2003), gender (Buckwalter and Stich 2014), emotional state (Wheatley and Haidt 2005; Nichols and Knobe 2007; Schnall et al. 2008; Tobia et al. 2013b), personality (Cokely and Feltz 2009, Schulz et al. 2011), and even the order in which cases are presented (Swain et al. 2008, Schwitzgebel and Cushman 2012, Wiegmann et al. 2012). This variation was deeply worrisome. How, the experimentalists thought, could intuition be a reliable source of evidence given its apparent sensitivity to all these irrelevant factors?

Such worries have been expressed by experimentalists in many different forms, but there seems to be a fairly constant form of argument running throughout. Joachim Horvath (2010) refers to this as the ‘master argument’ of experimental philosophy; Machery and O’Neill (2014) call it the ‘argument from unreliability against the method of cases.’ Exact formulations of the argument of course vary; here is my version. Henceforth, I’ll refer to this formulation as the experimentalists’ ‘Core Argument’ against current philosophical methodology.

• Premise 1: If intuitions vary as a function of truth-irrelevant factors, then intuition is unreliable.

• Premise 2: If intuition is unreliable, then it cannot justify our belief in philosophical propositions.

• Premise 3: If intuition cannot justify belief in philosophical propositions, traditional philosophical methodology must be rejected or radically revised.

• Premise 4: Empirical findings show that intuitions vary as a function of truth-irrelevant factors.

• Conclusion: Traditional philosophical methodology must be rejected or radically revised.

Opponents of experimental philosophy have responded to the Core Argument in a number of ways. Several authors, even before the emergence of x-phi, have noted that fully abandoning intuition is simply not an option. Laurence Bonjour (1998), for instance, claims that intuition is involved in any sort of reasoning that goes beyond direct perception.¹ In a similar vein, both Bealer (1992) and Pust (2000, 2001) argue that intuition is involved in even the most basic epistemic claims, such as the claim that we ought not to form beliefs on the basis of unreliable sources of evidence. If this is so, then it looks as though any argument against the use of intuition, including the experimentalists’ argument, will inevitably make use of intuition—thereby falling to self-defeat.

A related complaint is that a rejection of intuition is simply too skeptical—it threatens to overgeneralize. Sosa (1998, 2007) has noted that the epistemic flaws experimental philosophy has revealed in intuition are not notably more severe than the epistemic flaws that infect ordinary perception. But if this is so, then there seems to be an inconsistency in using the Core Argument to reject intuition while simultaneously maintaining the evidential status of perception. Williamson (2007, 2015) has also accused experimentalists of courting skepticism. Williamson, in fact, doubts that there is a single capacity of ‘intuition’ at all—instead, the mental states we call ‘intuitions’ invoke a number of extremely common cognitive processes, such as our ability to apply concepts and to reason with counterfactuals. There is no obvious separation between ‘philosophical intuitions’ and everyday judgment; thus, a rejection of intuition threatens to generalize to radical global skepticism.

Another common reaction to the experimentalist Core Argument is to object that experimentalists have been studying the wrong subjects. Experimentalists standardly use ordinary folk as subjects—commonly undergraduate students, and sometimes even passers-by on the street. But most such subjects have minimal exposure to philosophy. They lack the special skills of professional philosophers—skills like carefully attending to the relevant possibilities, properly filling in the details of an underspecified case, or simply possession of enhanced conceptual competence. Several philosophers, such as Williamson (2011), Devitt (2011), and Ludwig (2007), have therefore argued that the experimental findings are irrelevant. Professional philosophers are experts and will not be subject to the problematic variation that has been observed among laypersons.

A more recent approach to the experimentalist challenge questions the presuppositions of the entire project. Philosophers like Deutsch (2009, 2010, 2015) and Cappelen (2012) have rejected the very idea that intuitions play a major role in philosophical theorizing. Philosophical theorizing, Deutsch and Cappelen argue, does not consist primarily of appeals to intuitions generated by thought experiments to support or undermine proposed philosophical analyses. Instead, philosophers support their views with arguments. It is argumentation, rather than intuition, which provides our primary evidence for or against philosophical theories. But if this is correct, then the experimentalist findings again become irrelevant—experimentalists are critiquing a method that philosophers do not in fact employ.

Finally, within the past few years, the empirical picture which has emerged from experimental philosophy studies has become much less clear. Some of the original, dramatic findings which catalyzed the discipline have failed to replicate—Seyedsayamdost (2015), for instance, failed to reproduce the Weinberg et al. findings, and Lam (2010) similarly provides a non-replication for the Machery et al. (2004) findings regarding intuitions about reference. What’s more, several recent studies have indicated substantial levels of stability in intuition, rather than variability. Sarkissian et al. (2010), for instance, found that free will intuitions varied little across cultures. Wright (2010), meanwhile, demonstrated that several types of epistemic and ethical intuition appear to be insensitive to order of presentation. This seems to conflict with premise 4 of the Core Argument as outlined above; that is to say, it seems to provide evidence against the claim that intuitions are highly variable and sensitive to truth-irrelevant factors.

It will be helpful to make a quick summary of these objections to the experimentalist Core Argument:

• Self-defeat: The Core Argument urges us to abandon intuition—but it must rely on intuition in support of its premises. It therefore falls to self-defeat.

• Skepticism, version 1 (Sosa): The flaws of intuition don’t seem much more serious than the flaws of vision—if we reject intuition, then the same standard will require us to reject vision, too. That’s too skeptical.

• Skepticism, version 2 (Williamson): Intuition isn’t just used in philosophy—it is also used in ordinary cases of concept application, such as everyday knowledge attributions. Rejection of ordinary concept application is too skeptical.

• Expertise: The studies experimental philosophers have performed have used undergraduates as subjects. But undergraduates don’t have as much expertise in philosophy as professional philosophers. We shouldn’t expect the intuitions of professional philosophers to vary.

• No Intuitions: Experimentalists have been wrong to assume that philosophers use intuitions as evidence. So x-phi studies demonstrating flaws in intuition are irrelevant.

• Stability: Recent studies indicate that intuition is in fact stable, rather than variable.

There are, of course, other objections that have been made to the Core Argument and to the project of experimental philosophy more generally. But the above objections, I will claim, have something in common: their efficacy depends on an oversimplified picture of philosophy and its methods. These objections to the Core Argument will be the focus of this paper; I will argue that none of them truly threatens the experimentalist.

The ‘Standard Framework’ of Contemporary Metaphilosophy

It is my view that the objections to the Core Argument discussed above are convincing only because most experimentalists—and most of their opponents, as well—have largely been employing a set of problematic presuppositions regarding matters metaphilosophical. There is, I would claim, a sort of ‘Standard Framework’ within which much of contemporary metaphilosophy operates—one which structures the intuition debates in a way which disadvantages the experimentalists. This framework provides a certain image of philosophy’s goals, methods, and so forth, as well as a general picture of how our methods are to be evaluated. It is, I argue, on the whole an oversimplistic and misguided approach to metaphilosophical issues.

Here, in brief, is how the Standard Framework portrays philosophical inquiry. Philosophers are said to aim at providing analyses of philosophically interesting concepts such as ‘free will,’ ‘consciousness,’ ‘knowledge,’ ‘good.’ A successful analysis elucidates the conditions under which something falls under the target concept—the conditions under which a mental state counts as knowledge, for instance. Proposed analyses are tested via the ‘method of cases’: the philosopher imagines a possible case (generally via a ‘thought experiment’) and compares her intuitions regarding said case to the verdict the analysis generates. If there is a conflict between her intuitions and the analysis, she has found a counterexample and the analysis is rejected—or at least, the intuition provides strong evidence against the analysis. Classic examples of this method in action would include the Gettier cases and standard counterexamples to utilitarianism such as the magistrate and the mob. Intuition, then, is held to be our primary source of justification for judgments like ‘x is a case of F,’ where F is the philosophical category of interest. Philosophical theory building serves to systematize and elucidate the intuitions—and thus, we are justified in endorsing a theory when and only when its classifications comport with our intuitions on all imagined cases. Or at least, nearly all—an intuition may occasionally be rejected as false, but only if we have very strong reasons to suspect error.

There is of course some truth in this picture of philosophy’s basic operating procedure, but it is on the whole a caricature. For the remainder of this paper, I’ll focus on three elements of the Standard Framework which I find especially problematic: the ‘homogeneity assumption,’ the ‘epistemology assumption,’ and the ‘method of cases assumption.’ I’ll argue that abandoning these problematic assumptions disarms a large number of objections to the Core Argument. Nonetheless, the Core Argument will itself need some reworking to purge it of these problematic Standard Framework presuppositions. What emerges will be a modified experimentalist critique of current philosophical methods which is more modest in scope, but also more resistant to objection—and, to my mind, simply more plausible overall.

The Homogeneity Assumption

The picture of philosophical inquiry that the Standard Framework presents centers around a type of mental state called ‘intuition’—this mental state has, as we’ve seen, been at the center of most of the debates over the implications of experimental philosophy. There’s a great deal of disagreement over how this type of mental state should be defined—some argue that intuitions essentially possess a certain phenomenological character, others claim that their distinguishing feature involves a certain link to conceptual competence. Yet others define intuition simply in terms of its automaticity or speed, or its relative opacity to conscious thought.

Throughout these debates over intuition’s character, however, there is a shared tendency to treat ‘intuition’ as though it picked out a reasonably natural, reasonably homogeneous mental category—as opposed to, say, a collection of quite different mental processes. This tendency extends to evaluations of intuition’s epistemic credentials, as well. Recall that the Core Argument, as I’ve presented it, is phrased in terms of the unreliability of intuition as a whole; this is, in fact, how experimentalists tend to state their case. Defenders of traditional philosophical methods, for their part, tend to respond by arguing that intuition (as a whole) is reliable and/or trustworthy. Note for instance the following examples:

• ‘[One project] within experimental philosophy relies on data concerning cognitive diversity to argue that philosophers should not use intuitions as evidence in their theorizing’ (Nadelhoffer and Nahmias 2007, 126).

• ‘Sensitivity to irrelevant factors undermines intuitions’ status as evidence’ (Swain, Alexander and Weinberg 2007, 141).

• ‘I will argue… that we are at the present time unwarranted in drawing any negative conclusions about intuitions from the relevant empirical studies’ (Bengson 2013, 496).

There are exceptions to this tendency, of course—Cappelen (2012), Williamson (2007) in particular are quite skeptical about the usefulness and even the cogency of the term ‘intuition.’ But by and large, metaphilosophy debates the implications of x-phi via an evaluation of the reliability of ‘intuition.’

But even a cursory consideration of the sorts of states labeled ‘intuitive’ suggests that they are anything but homogeneous. We have intuitions about whether an action is morally wrong; about whether modus ponens is valid; about whether Gettier cases count as knowledge; about whether qualia-less duplicates of ourselves are metaphysically possible. Is it really at all likely that these various judgments are produced by a single, unified type of mental capacity? Do philosophers really expect to uncover anything like a single ‘faculty of intuition’ in the brain? Psychological literature only further confirms this suspicion—see Nado (2014) for a sample of empirical evidence indicating the deep heterogeneity of intuition.

The plausible psychological heterogeneity of intuition matters, because it is further plausible that this heterogeneity is accompanied by differing levels of susceptibility to the sorts of biases experimental philosophers have studied. Suppose moral intuitions are produced by psychological processes which recruit emotion, while logical intuitions are not. We would then expect moral intuitions to be more susceptible than logical intuitions to emotional-based biases. This, in turn, suggests that the degrees of reliability displayed by moral intuitions and by logical intuitions aren’t likely to be equal.

Prima facie, this threatens the experimentalist—because it renders the Core Argument problematic. The Core Argument assumes that experimentalist studies license claims about the unreliability of intuition—full stop—and from there claims about the viability of philosophical method. But if intuition is even moderately heterogeneous, this is an extreme oversimplification. Fortunately, the experimentalist’s situation is not quite as dire as it might first appear. In actual fact, the falsity of the homogeneity assumption threatens only the Core Argument as it has been standardly formulated. The homogeneity assumption is in no way crucial to the experimentalist critique; dropping it would simply mean that the experimentalist must replace her sweeping condemnation of reliance on intuition with a number of more limited conclusions about a variety of judgment types. One such conclusion might be, say, that emotional biases appear to contaminate (certain types of) moral judgment; another might be that personality type appears to bias judgments about free will. The degree to which these biases render various subfields of philosophy problematic will then vary from case to case.

The move to more limited, self-contained critiques has a significant advantage, as well—it disarms certain of the objections to the core argument discussed earlier. Most obviously, it enables the experimentalist to almost wholly evade the self-defeat objection. There is, on this picture, no problem whatsoever in employing epistemological intuitions in an argument critiquing (say) moral intuitions. There would, of course, remain a difficulty in employing epistemological intuitions to critique epistemological intuitions—but even there, the experimentalist might be able to evade the issue by drawing her conclusions even more narrowly (perhaps she might uncover evidence that a certain problematic bias affects only intuitions about epistemic luck, for instance).

We also gain a helpful perspective on the more recent wave of findings indicating stability in intuition. These findings, in fact, don’t indicate that intuition is stable—because intuition simply is not a unified, homogeneous, ‘natural kind’ of mental state, to be evaluated with a broad brush. Instead, the findings indicate, of some types of intuition, that they are resistant to some possible types of bias or variation. But given the heterogeneity of intuition, the finding that free will intuitions are relatively constant across cultures tells us little about the cultural variability of epistemic intuitions, or logical intuitions, or ethical intuitions—the last of which quite clearly do vary by culture. Recent stability findings merely demonstrate how complicated a complete picture of the epistemic qualities of ‘intuition’ is likely to turn out to be. Again, this is only problematic if one takes the experimentalist argument to demand a complete rejection of ‘intuition’ as a whole.

Finally, abandoning the homogeneity assumption also gives at least some defense against the worry that the experimentalist’s argument threatens skepticism. Consider Williamson’s claim that the critic of intuition risks global skepticism—if the experimentalist backs away from a wholesale condemnation of intuition, the threat of global skepticism simply evaporates. The reformed experimentalist contents herself with narrower critiques, denouncing only judgments which arise from psychological process types which have been found subject to bias. Of course, we might not immediately possess enough empirical knowledge to determine whether two judgments are produced by the same psychological process type—but that merely shows that experimentalists have more work to do.

There is, however, a residual difficulty in Williamson’s critique that is not resolved by abandoning the homogeneity assumption. The core insight of Williamson’s argument is that there is a substantial overlap between ordinary and philosophical cognition; there is plausibly a great deal of similarity between the psychological process types that underlie philosophical judgments about Gettier cases and those that underlie the ordinary knowledge attributions we all make on a daily basis. Thus, though the experimentalist might not risk global skepticism, she does risk rejecting a substantial number of ‘everyday’ judgments. If she argues that Gettier judgments ought to be rejected because of, e.g., susceptibility to cultural variation, then this would likely require her to also reject, e.g., the judgment that Donald Trump doesn’t know much about foreign policy. Fortunately, as we’ll soon see, there are other modifications to the Core Argument that will avoid this difficulty.

The Epistemology Assumption

The Core Argument, as I have presented it, argues that intuition is unreliable, and that is therefore cannot justify our belief in philosophical propositions. Justification is, of course, one of the core concepts in standard analytic epistemology—it is crucial to most accounts of the nature of knowledge. This reflects a second problematic assumption of the Standard Framework of metaphilosophy: its tendency to employ the concepts of analytic epistemology for the purpose of evaluating philosophical method.

Most participants in the ‘intuition debates’ seem to be concerned with whether or not intuition is a source of justification, a source of evidence, a source of knowledge, and so forth. Experimentalist conclusions are frequently stated in such terms; the tendency is even stronger among defenders of intuition. A few characteristic examples:

• ‘Experimental evidence seems to point to the unsuitability of intuitions to serve as evidence at all’ (Alexander and Weinberg 2007, 63).

• ‘Experimental philosophy challenges the usefulness of [appealing to intuition] in achieving justified beliefs’ (Alexander et al. 2010, 298).

• ‘There is inadequate reason to move away from the intuitively attractive view that we have a faculty of intuition, in many ways akin to our faculties of perception and memory and introspection, that gives us reason for belief, and with it, often enough, gives us knowledge’ (Sosa 2006, 634).

One can easily see how this way of framing the debate leads, for instance, to a conception of experimentalists as ‘intuition skeptics’—after all, they often seem to commit themselves to the idea that intuitions do not generate the ‘epistemic goods’ (such as justification, evidence, or warrant) required for knowledge.

Of course, it is also common for experimentalists to simply state their conclusions in terms of the unreliability of intuition. But this sort of conclusion is typically left underspecified; experimentalists state that intuitions are ‘unreliable,’ without further elaboration. It’s then immediately inferred that philosophical methods are in trouble. But exactly how unreliable are experimentalists claiming intuition to be? Not just any amount of unreliability will secure the experimentalist conclusion. A method of philosophical inquiry which achieves 99.99999999% reliability, after all, displays some level of unreliability—but certainly not enough to merit any serious worries about its use. This is, of course, exactly Sosa’s point when he reminds us that perception displays unreliability as well; we employ moderately unreliable evidential sources in nearly all forms of ordinary belief formation.

Given that experimentalists don’t flesh out their claims of unreliability, and given the ubiquity of standard analytic epistemology concepts within the intuition debates, the default interpretation of their unreliability claims would seem to be this: the experimentalist intends to argue that intuition is sufficiently unreliable to disqualify it as a source of evidence/justification/knowledge. This is, at least, the interpretation that many opponents of x-phi have seemed to default to.

If this is really the experimentalists’ intended conclusion, though, then experimentalists are open to multiple charges of extreme skepticism. First, we have Sosa’s point about perception. Sosa rightly notes that the epistemic flaws of intuition don’t seem to be significantly worse than those of perception: perception, after all, is subject to numerous illusions and expectation-based biases, as well as widespread (though moderate) imprecision and error. So if intuition is to be disqualified as a source of evidence, justification, or knowledge, then by parity of reasoning perception ought to be so disqualified as well. This, of course, essentially functions as a reductio ad absurdum of the experimentalist position.

Secondly, we have Williamson’s similar, but separate, argument that experimentalists risk global skepticism. We’ve already seen that metaphilosophers have tended to neglect the heterogeneity of intuition; but Williamson’s argument reminds us that they have also tended to neglect the substantial overlap between philosophical intuition and everyday types of reasoning. As noted earlier, even rejecting the homogeneity assumption does not fully avoid the skeptical worries that arise from this overlap; a rejection of the psychological mechanism underlying (say) Gettier judgments is quite likely to bring down substantial portions of our everyday knowledge attributions as well.

But all of this relies on a tacit assumption that the epistemological standards at issue are the ones that govern ordinary, everyday belief formation—in other words, the standards pertaining to knowledge. Plausibly, however, the epistemological standards relevant to a specialized, professional field of inquiry like academic philosophy are rather more stringent than those that govern everyday belief. In other words, ordinary levels of justification simply aren’t enough. Intuition might be sufficiently reliable for ordinary belief formation, but not reliable enough for philosophy.

Consider an analogue: scientific inquiry. Scientists do not in fact consider ordinary, uncontrolled perception to be reliable enough for the purposes of many scientific tasks, such as data gathering conducted during the course of an experiment. This is evident from the bevy of methodological practices scientists have put in place to control for the epistemic deficiencies of perception: practices like double-blind data gathering, the use of multiple observers/coders, videotaping, the use of specialized implements like eye-tracking software, and even reliance on basic measurement tools like the ruler and scale. None of these procedures are obligatory during everyday instances of perceptually based belief formation. In other words, though perception obviously qualifies as a source of evidence in ordinary belief formation, its use in scientific inquiry is restricted—when it comes to data gathering in experimental contexts, scientists rely only on instances of perception which have been closely controlled by methodological procedures which greatly reduce the risk of bias and error.

The sorts of methodological critiques and replies that seem reasonable within the sciences are, moreover, quite dissimilar to those that populate our own intuition debates. Early proponents of double-blind, placebo-controlled trials presumably did not argue that perception fails to grant evidence, justification, or knowledge; and it would have been absurd to react to their proposals by countering that their arguments threatened skepticism. Instead, the appropriate argument was simply that then–current methods of scientific inquiry needed adjustment. Nothing at all was thereby implied about the suitability of uncontrolled perception for ordinary belief formation. Obviously, it is perfectly possible to argue that the methods currently employed in the sciences are problematic without thereby risking any form of skepticism—so why should things stand any differently for the use of intuitions in philosophy?

I would suggest that, in general, ‘professional inquirers’ like scientists and philosophers hold themselves to higher epistemic standards than ordinary folk do during their everyday cognitive activities. And indeed, this is as it should be. It takes a great deal of time, education, and resources to gather data in a way that respects all the various methodological expectations that scientific researchers hold themselves to. The ordinary man on the street can’t very well be expected to employ anything like double blinding or multiple observers during ordinary epistemic activity. That would be absurdly overdemanding—most ordinary folk simply don’t have the time or ability to achieve scientific rigor when determining, say, whether or not to believe that New Yorkers are less friendly than Londoners, or whether or not to believe that it tends to be humid after rainy days.

Professional inquirers, by contrast, are paid to spend 40 + hours per week on inquiry—usually, in-depth inquiry into a very narrow range of phenomena. The man on the street simply does not have forty spare hours weekly to devote to careful, rigorous investigation of (say) the emergence of social cognition in infants—the developmental psychologist, however, does. She also typically has access to a laboratory, assistants, funding, and so forth; the man on the street, quite obviously, lacks this as well. It is not overdemanding, then, to expect extraordinarily high degrees of epistemic quality from a professional inquirer. She is fully able to devote substantial amounts of time and energy to reducing the effects of bias and error that arise from the fallible evidential sources upon which she relies.

These observations hold double for communities of inquirers—for instance, for the scientific community as a whole. In most cases, an individual researcher will never fully settle any open question in her field. The greatest scientific questions often remain open for years, decades, or more—questions regarding, for instance, the mechanisms by which life initially emerged from inorganic matter, or regarding the nature of dark matter. But this is no obstacle for the scientific community; unlike the man on the street, or even the individual researcher, the scientific community faces no limits on available time. And, of course, the resources that the scientific community is able to devote to inquiry for any given hypothesis far exceed those available to an individual. By combining resources in this way, it becomes possible to achieve levels of rigor and error reduction that are unavailable to most individuals during everyday cognitive activity.

The moral, then, is as follows. Intuition quite plausibly resembles uncontrolled perceptual observation, in the sense that it is subject to various biases, imprecisions, and errors. Nonetheless, like perception, intuition is admissible as a source of evidence in everyday epistemic activity—and as a source of justification and knowledge. But philosophers, given their role as professional inquirers, ought to hold themselves to a higher standard. For after all, like all communities of professional inquirers, we possess far greater spare resources in terms of available time, cognitive energy, and the like than does the average man on the street. Like the scientist, then, our goal should be to reduce the impact of the various epistemic flaws our evidential sources are prone to—and to do so to a far greater degree than would be required of an ordinary inquirer.

The ordinary man on the street is not obligated to concern himself overmuch with the various biases of intuition during his ordinary epistemic activities, such as everyday instances of knowledge-attribution. The philosopher working in a professional context, however, is obligated to address those biases—and has thus far failed to do so. Current philosophical methods do not involve standardized procedures to reduce the effect of potential bias—we have no analogue to, say, double blinding. Thus, though intuition is in fact a source of knowledge and justification, current philosophical reliance on it is problematic. On this way of looking at things, skeptical worries are wholly beside the point—experimentalists’ critiques of current philosophical method need not imply skepticism.

The Method of Cases Assumption

Thus far, we’ve seen that several of the most prominent objections to the experimentalist Core Argument can be defused by giving up two unneeded components of the Standard Framework of metaphilosophy—the homogeneity assumption and the epistemology assumption. However, two troublesome objections have not yet been fully addressed. The first of these is the ‘expertise defense’—the claim that philosophers will likely be resistant to the biases the experimentalists’ subjects displayed, due to the enhanced expertise that comes along with philosophical training. The second remaining objection is the claim that philosophers do not, after all, rely on intuitions in their philosophical theorizing. If intuitions play no role in philosophy, then empirical evidence for their unreliability provides no threat to current philosophical methods.

There is, in fact, an element of truth in the ‘no intuitions’ objection. Many participants in the ‘intuition debates’ appear to operate under the assumption that the primary method of philosophy involves the testing of philosophical analyses against intuitive reactions to thought experiments—what is often called the ‘method of cases.’ On this sort of view, intuition provides something like the ‘data’ that a successful philosophical theory needs to capture and systematize. But the proponents of the ‘no intuitions’ objection are correct to object to this characterization of our methods—the ‘method of cases’ simply is not the primary method of philosophy. As Cappelen and Deutsch note, philosophers quite frequently provide arguments in favor of their philosophical positions, rather than merely claiming that those positions best fit intuition. Moreover, many of the argumentative moves that philosophers make don’t even remotely rely on anything like philosophical reactions to thought experiments. Case in point: not a single one of the objections to experimental philosophy we’ve discussed relies on anything like the ‘method of cases.’

Nonetheless, the claim that philosophers never employ intuitions takes things rather too far. The method of cases is not the primary method of philosophy, but it certainly is used. Intuitive reactions to cases are often held to provide counterexamples to philosophical analyses, and it is implausible to hold that such moves are invariably accompanied by argumentation in support of the judgment. Cappelen and Deutsch have pointed out that arguments were indeed present in some of the most well-known presentations of thought experiments in philosophy, but these lofty cases are hardly representative of the average philosophy text. My suspicion is that a survey of less prestigious journal articles would reveal that ‘bare’ intuition is quite frequently used in the absence of argument.

Of course, conducting such a textual survey is beyond the scope of the current paper; so let’s simply grant, for the sake of argument, the claim that all thought experiment case judgments are supported by argumentation. Even this does not suffice to show that intuition is irrelevant to philosophical theorizing. First, as others have pointed out (see, e.g., Brogaard 2014; Chalmers 2014; Weinberg 2014), intuition is plausibly needed to support the premises of the arguments Cappelen and Deutsch have pointed to.² Second, even in cases where argumentation does accompany a thought experiment, the role of the argumentation is plausibly to explain the truth of the intuited proposition, not to provide evidence for it.³ The primary evidence for the proposition is standardly taken to be the intuition. Consider, for instance, the post-Gettier literature. Though Deutsch (2010) persuasively argues that argumentation is indeed present in Gettier’s original 1963 paper, responses to Gettier’s paper don’t typically focus on these arguments. Indeed, the arguments are rarely even mentioned. Instead, much of the post-Gettier literature simply accepts that Gettier’s judgment on the cases was obviously the correct one; they then go on to give their own explanations of (rather than evidence for) the fact that Gettier’s protagonist Smith does not know. On Deutsch’s interpretation, the post-Gettier literature would appear to be a series of additional arguments in support of denying knowledge to Smith; but it is much, much more plausible to think that these philosophers, and Gettier himself, were offering explanations of an intuited proposition’s truth rather than arguments in its support.

Third, even if the role of argumentation is primarily evidential, rather than explanatory, intuition plausibly still plays a role in the ‘context of discovery’—in other words, one generally first intuits the truth of a proposition p and then later formulates arguments in support of p. It may well be that, after formulating an argument for p, the argument then provides the primary justification for p; but even if this is so, the threat of bias remains. Consider the fact that, for any given philosophical problem, there tend to be good arguments for multiple different, incompatible solutions. Of course, there cannot be sound arguments that favor incompatible conclusions; but there might be, for instance, multiple valid or inductively strong arguments employing different but equally subjectively plausible sets of premises. Or, perhaps, multiple abductive inferences which employ differing background assumptions to reach incompatible ‘best’ explanations.

Consider, for instance, a familiar enough case: one philosopher argues from the value of human life to the impermissibility of abortion, another from a woman’s right over her own body to its permissibility. Many of us feel the force of both arguments, and the plausibility of both sets of initial premises (or at least can, in certain moods)—and it may well be a brute intuition regarding a preferred conclusion that leads us to ultimately favor one argument above the other. Indeed, the weight of, e.g., an intuition regarding a conclusion might even lead one to strengthen one’s credence in premises which support it, or weaken one’s credence in premises which support its negation. Perhaps, for instance, the strong intuitive pull of abortion’s permissibility has led me to lower my credence in the claim that, e.g., an early term fetus counts as a human life. In more complex cases, the intuitiveness of a given conclusion might lead one to fail to consider possible arguments for incompatible conclusions—a person with a strong Christian upbringing might easily overlook, for example, the apparent paradoxes surrounding omnipotence.

The intuitiveness of a proposition P, then, may well lead philosophers to develop arguments in support of P rather than its alternatives. In other words, philosophers might (like everyone else) be subject to confirmation bias: that is to say, to the tendency to search for and interpret new evidence in a way that supports the views one already holds. Imagine we discover that the intuitiveness of some proposition P is subject to, e.g., cultural biases. Now, suppose philosopher McX finds P intuitive, and for that reason searches for (and finds) good arguments in favor of P. The trouble is this: had she been raised in another culture, she would have found not-P intuitive, and she would have searched for (and probably found) good arguments in favor of not-P. Assuming we abandon the epistemology assumption, we might even admit that McX’s argument justifies her belief in P—but we might still maintain that her methodological approach is problematic for philosophical purposes due to the uncontrolled biases operative in the context of discovery.⁴

Let’s now turn to the final remaining objection to the Core Argument: the ‘expertise defense.’ Here, too, giving up on a simplified view of philosophical method (that is, the method of cases model) gives us a powerful response to the objection. No one has empirically demonstrated that philosophers possess expertise that will enable them to resist the errors and biases to which folk intuition is prone (indeed, there is some evidence in the opposite direction—see, for instance, Cokely and Feltz (2011), Schwitzgebel and Cushman (2012); Tobia et al. 2013a, b). Instead, proponents of the expertise defense claim that we are entitled to a default assumption of philosophical expertise. After all, we assume that, e.g., professional scientists, professional chess players, and so forth have expertise, even in the absence of empirical studies to that effect.

If we conceive of philosophical expertise as ‘expertise in intuition,’ then the defense does seem to provide an obstacle to the experimentalists’ critique. But why should the experimentalist accept a characterization of philosophical expertise in these terms? We’ve already seen that the homogeneity assumption is problematic: intuition isn’t a single mental capacity, and so there will not be a single form of ‘expertise in intuition,’ but instead many. Now that we have also questioned the method of cases assumption, ‘expertise in intuition’ seems even more inapt—appeal to intuition is not the primary method of philosophical inquiry, but merely one method among many.⁵

Philosophical method is incredibly rich and varied, relying on many different types of cognitive processes—some ‘intuitive,’ some not. Do we really get a ‘default presumption of expertise’ for all of these? Surely we can assume some form of expertise—but are we entitled to assume every type of expertise that would be needed to address the various biases uncovered by experimentalists? Note that scientists don’t assume that they are immune to observational biases—though they clearly possess some forms of specialized expertise, they don’t assume that they will automatically resist, e.g., confirmation bias simply as a byproduct of their scientific education. Instead, they assume that they too will be subject to such biases; and consequently, they take explicit steps (such as blinding) to mitigate the problem. We philosophers, by contrast, have no such practice.

Conclusion

Let’s summarize. Many of the extant objections to the Core Argument of experimental philosophy rely, in one way or another, on the three problematic assumptions discussed in the previous section. The self-defeat objection fails once we abandon the idea that intuition should be critiqued as a whole; the use of epistemological intuitions in an experimentalist argument is self-undermining only if that argument urges rejection of epistemological intuitions that issue from the very same psychological process. Findings indicating stability in intuition are also harmless to the experimentalist, provided that we recognize that it is wholly to be expected that different types of ‘intuitive’ cognition will display different levels of bias and unreliability.

Both versions of the skepticism objection, as we’ve seen, fail to threaten an experimentalist who is careful to launch multiple, narrow critiques rather than a sweeping condemnation of all intuitive cognition. Further, skeptical worries become even less troubling once we abandon the assumption that critiquing philosophers’ use of intuitions requires denying that intuitions constitute a source of knowledge.

We’ve granted that the ‘no intuitions’ challenge has an element of truth to it—but we have also noted that we can abandon the ‘method of cases’ assumption while still maintaining that the biases of intuition render even argument-based methods problematic. And finally, the expertise defense only appears plausible given the idea that we can make a default assumption regarding the existence of some general-purpose ‘expertise in intuition.’ Given the diversity of both types of intuition and types of philosophical method, this is deeply implausible.

The experimentalist core argument, of course, will need to be modified to comport with the departures we’ve made from the ‘Standard Framework’ of metaphilosophy. Here, then, is my suggestion for an amended Core Argument:

• (1) If intuitive judgments produced by a given psychological process are found to vary as a function of truth-irrelevant factors, then this is evidence that the psychological process displays at least some degree of bias and unreliability.

• (2) If a psychological process displays bias or unreliability, then professionals who make use of that process in at least some of their methods ought to attempt to devise corrective procedures for mitigating that bias or unreliability, or else turn to other sources of evidence.

• (3) Philosophers are professionals who make use of many different ‘intuitive’ psychological processes in some (though not all) of their methods.

• (4) Empirical findings show that several psychological processes used by philosophers produce intuitions that vary as a function of truth-irrelevant factors.

• (5) Philosophers ought to attempt to devise corrective procedures for mitigating the bias and unreliability of the affected psychological processes, or else turn to other sources of evidence.

The revised argument is, of course, much less dramatic—perhaps it is rather less exciting, as well. But it is also extraordinarily plausible. It is, essentially, simply a claim that philosophers ought to try to improve the accuracy of their methods whenever potential flaws are uncovered. Professionals in other fields do this all the time—most notably and clearly in the experimental sciences. It’s hardly a controversial claim that scientists ought to adhere to blinding procedures and other bias reduction methods—why should the analogous claim be in any way controversial in philosophy?