Metaphilosophy and the Role of Intuitions

Abstract

The practice of appealing to intuitions as evidence has recently been criticized by experimental philosophers. While some traditional philosophers defend intuitions as a trustworthy source of evidence, others try to undermine the challenge this criticism poses to philosophical methodology. This paper argues that some recent attempts to undermine the challenge from experimental philosophy fail. It concludes that the metaphilosophical question whether intuitions play a role in philosophy cannot be decided by analyzing our use of the word ‘intuition’ or related terms, and what philosophers rely on may not be manifest on the surface of what they write. The question what intuitions are and what their role is in philosophy has to be settled within the wider framework of a theory of knowledge, justification, and philosophical methodology.

1 Undermining the Challenge from Experimental Philosophy?

Some philosophers think that intuitions are treated as evidence in contemporary analytic philosophy as well as in the history of philosophy, going back to Plato. In the recent debate about philosophical methodology, Bealer (1998), Goldman and Pust (1998), and Sosa (2007b), among others, provide arguments for the claim that intuitions are justifiably treated so, since they in fact are evidence in philosophy. Advocates of the current movement of experimental philosophy such as Weinberg et al. (2001) or Machery et al. (2004), to mention only two well-known papers, agree that we rely on intuitions, but they consider it a practice which is not conducive to the aims of traditional philosophy. Weinberg, Nichols and Stich’s empirical studies on Gettier Cases suggest that there are significant variations in people’s intuitive responses to these cases, depending on their culture and socioeconomic status.¹ Based on their result, the authors argue that traditional epistemologists do not succeed in establishing genuine epistemic norms because what they really explore are norms local only to their own cultural and socioeconomic group. Machery, Mallon, Nichols and Stich’s empirical studies on Kripke’s Gödel Case reveal significant variations in people’s intuitions about the reference of names. Raising similar concerns as Weinberg, Nichols and Stich, the authors draw the conclusion that ‘philosophers must radically revise their methodology’ and ‘need to get out of their armchairs’.²

This paper does not discuss the experimental philosophers’ studies or their arguments against the use of intuitions as evidence. Rather, it discusses some traditional philosophers’ reactions that take, more or less, the same line of argument. Instead of defending intuitions as evidence, they aim to undermine the challenge experimental philosophy poses to the use of intuitions as evidence by denying that intuitions play the assigned role. Deutsch (2009, 2010) argues that in relevant and frequently discussed cases, intuitions do not play an evidential role. In his book Philosophy without Intuitions, Cappelen (2012) argues that there is no evidence to the effect that we rely on anything which should be labeled ‘intuitions’.³ Before presenting their arguments and responding to them in detail, some more introductory remarks are needed.

Philosophers who think that intuitions play a role as evidence usually also think that reasoning involving thought experiments provides paradigm cases of the use of intuitions as evidence. This paper is concerned with intuitions from thought experiments only, and the following cases will serve as examples. In Frank Jackson’s Mary Case against physicalism, some of us have the intuition that colour scientist Mary learns something new when she leaves her black and white room and sees something coloured for the first time in her life (Jackson 1982). In Kripke’s counterexample to a descriptivist theory of the meaning of names, the Gödel Case, most of us have the intuition that ‘Gödel’ refers to Gödel and not to Schmidt (Kripke 1980). In Alvin Goldman’s Fake Barn Case against the theory of knowledge as justified true belief (the JTB theory), we have the intuition that a person in a particular situation lacks knowledge that P despite having a justified true belief that P (Goldman 1976). In Keith Lehrer’s Truetemp Case against reliabilism, we have the intuition that Mr. Truetemp does not know that the temperature is 104 degrees, despite the reliability of the process which leads to his true belief concerning the temperature (Lehrer 2000). In Stewart Cohen’s Lottery Cases, our intuitions concerning the question whether knowledge should be ascribed to a subject or not vary and are taken to support epistemic contextualism (Cohen 1988).

The claim that intuitions are evidence is ambiguous, as has been stated and discussed in the literature (e.g., Williamson 2007, Earlenbaugh and Molyneux 2009). It can either mean that the content of the intuition (the intuited), or that the psychological state (the intuiting) is evidence. Philosophers who think that evidence is propositional and factive refer to the second by speaking about ‘the fact that we have an intuition’ as evidence (e.g., Williamson 2007). Nothing in this paper depends on whether evidence is propositional and factive or not. Note that I am using ‘evidence’ in the sense in which it consists of whatever makes it the case that one has justification for believing a proposition P, as I take it to be used by Deutsch and Capellen.⁴ Moreover, I intend to stay neutral on two controversial issues: first, whether intuitions are a sui generis kind of psychological state or can be reduced to beliefs, judgments, or inclinations to believe, and second, whether intuitions play a distinctive epistemic role in philosophy or not. My arguments are directed against certain strategies used to establish that there are no intuitions and that intuitions do not play an epistemic role in philosophy.

Deutsch argues that the fact that we have an intuition is not often treated as evidence, and especially not in certain frequently quoted cases. His approach can be understood in the spirit of Williamson’s (2004, 2007) work on metaphilosophy and methodology. While Williamson thinks that the contents of our intuitions can be evidence, he argues that appealing to the fact that we have an intuition and to psychological states more generally is a practice we should not pursue.⁵ Williamson fears that the gap between facts about our psychological states with certain contents and the truth of these contents is not easy to close and provokes scepticism: a sceptic might always question an argument from the fact that somebody has an intuition that P to P. Attempts to psychologize the matter of philosophy (e.g., by holding that we are concerned with our concepts only) do not solve the problem, since ultimately, we are interested in the truth of P and not in facts about our psychological states concerning P. We should therefore appeal directly to the contents of our intuitions as evidence. However, intuitions do not provide a special kind or source of evidence, they are ordinary judgments (in some cases they are dispositions to believe), and in the case of thought experiments, they are reliable judgments.⁶ While Williamson criticises the practice of appealing to the fact that we judge a certain way, Deutsch argues that in some frequently discussed cases there is no such practice. Cappelen (2012) takes Williamson’s and Deutsch’s line of argument even further. While Williamson thinks that we can use the contents of our intuitions as evidence because thought experiments provide a reliable method, Cappelen thinks that there is no such method and there is no systematic use of the word ‘intuition’. According to Cappelen, the frequent use of ‘intuition’ has tricked contemporary philosophers into a wrongheaded and confused view of philosophical methodology.

Deutsch and Cappelen think that since intuitions do not play the role assigned to them by experimental (and some traditional) philosophers, the criticism coming from experimental philosophy is deeply misguided and does not pose a challenge to philosophical methodology: there is no need to change the way philosophy has been conducted. This paper argues that Deutsch and Cappelen use strategies which are not conducive to answering the question whether we rely on intuitions in philosophy.

2 Deutsch on Kripke’s Gödel Case

Kripke’s Gödel Case against the descriptivist theory of names has been considered a paradigm case of the use of intuitions as evidence. Deutsch thinks that Kripke appeals directly to the content of his intuition and not to the fact that he has an intuition. Here is, quoted at length, how he describes Kripke’s argument from the Gödel Case:

Kripke offers direct arguments against [the] descriptivist theory of meaning, but he also objects to it indirectly by criticizing the theory of reference it entails. D encapsulates the theory of reference that is a consequence of the descriptivist theory of meaning:

• D: An ordinary proper name, n, as used by a given speaker, S, refers to the object that is the denotation of some/most/all of the definite descriptions S associates with n.

To show that D is false, Kripke simply describes counterexamples–cases in which a name, as used by a given speaker, does not refer to the denotation of the definite description(s) the speaker associates with the name. Here is one such case, one of Kripke’s own (Kripke 1972/1980, pp. 83–84): Imagine that Gödel did not prove the incompleteness of arithmetic but that some other man, Schmidt, did. Gödel stole the proof from Schmidt and published it under his own name. But now imagine a speaker who uses ‘Gödel’, but associates just a single description with it, namely ‘the prover of incompleteness’. To whom does this speaker’s uses of ‘Gödel’ refer, Gödel or Schmidt? The answer, Kripke says, is Gödel, not Schmidt. If Kripke is right, D is false. (Deutsch 2009, p. 446)

[...] Kripke, after spinning the tale of Gödel and Schmidt [...] says that, on descriptivism, since the man who discovered the incompleteness of arithmetic is in fact Schmidt, we, when we talk about “Gödel”, are in fact always referring to Schmidt’ (Kripke 1980, p. 83). Immediately following this comment, Kripke says, ‘But it seems to me that we are not. We simply are not’ (Kripke 1980, p. 84). He does not say that it is intuitive that we are not talking about Schmidt; he says straight out, and emphatically, that we are not talking about Schmidt. (Deutsch 2009, p. 451)

In the second quote, Deutsch points out that Kripke does not say that the relevant content is intuitive to us, which would be evidence for the appeal to an intuition. Instead, Kripke appeals to the content of his intuition.

This, however, is a rather weak argument, because clearly Kripke could refer to an intuition using ‘it seems’ (see the second quote). The ‘seeming’ terminology has been used by defendants of rationalist accounts of intuitions such as Bealer (1998), who takes intuitions to be intellectual seemings in contrast to sensory seemings. I thus take it to be very plausible that Kripke intends to appeal to an intuiting. However, it is not my aim to establish this interpretation. The point is that Deutsch does not succeed in showing the contrary.

Deutsch gives two further reasons to support his interpretation of Kripke. First, Kripke argues for his view that ‘Gödel’ does not refer to Schmidt. Second, there is a better reconstruction of Kripke’s argument without reference to an intuition, which corresponds with what Kripke literally says. According to Deutsch, the following is not the correct form of Kripke’s argument (Deutsch 2010, p. 452):

• (0) It is intuitive that there is an F that is not a G.

• (1) So, there is an F that is not a G.

• (2) Hence, not all Fs are Gs.

Instead, the best reconstruction is this:

• (1) There is an F that is not a G.

• (2) Hence, not all Fs are Gs.

I think that both of Deutsch’s arguments in support of his interpretation of Kripke do not show what he aims to show. First, that Kripke gives arguments for his view does not support the claim that he does not rely on his intuiting as evidence. I will get back to this point in the discussion of Cappelen. Second, the best reconstruction of an argument is not necessarily the correct or most faithful reconstruction of an author’s reasoning. Once we have all the relevant evidence for the claim that ‘Gödel’ does not refer to Smith from Kripke’s arguments, we may decide not to use it all in our argument anymore. However, this does not mean that Kripke did not appeal to an intuiting when he first came up with his counterexample to the descriptive theory of names, and hence it does not mean that the appeal to an intuiting is not part of the correct reconstruction of Kripke’s reasoning. Take the following example. In the process of solving a murder, the fact that M was seen near the crime scene in the night of the murder is used as evidence that M killed the victim. However, once M’s bloody glove is found not far from the victim, the first piece of evidence is not important anymore. It is still evidence, but it is not the evidence that ultimately justifies M’s conviction. Asked how they solved the murder, the investigators might not even mention the fact that M was seen near the crime scene. However, this piece of evidence initially made the investigators suspicious of M. Similarly, Kripke might have relied on his intuiting as evidence before gathering more evidence trough various arguments.

If we appeal to intuitings as evidence from thought experiments (either with the term ‘intuition’ or with other terms), it follows that we rely on them as evidence. But we might also rely on intuitings without even appealing to them. According to Williamson (2000), we do not always know what our evidence is, and there are two ways in which a belief can be based on evidence. We can distinguish between explicitly relying on an intuiting and implicitly relying on an intuiting that serves as evidence for our beliefs.

Call one’s belief in p explicitly evidence-based if it is influenced by prior beliefs about the evidence for p [...] Call one’s belief in p implicitly evidence-based if it is appropriately causally sensitive to the evidence for p [...] the causal sensitivity of the belief in p to the evidence for p need not be mediated by further beliefs about the evidence for p. (Williamson 2000, pp. 191–192)

Hence, according to such a view, we could rely on an intuiting as evidence without explicitly appealing to it or even without actually believing that it is part of our evidence.

3 Cappelen on the Role of Intuitions in Epistemology

Cappelen takes Deutsch’s line of arguing even further. He thinks that we have been engaging in a use of the word ‘intuition’ which some philosophers have falsely interpreted as referring to a particular mental state:

Speaking loosely, here is how I think of the situation we philosophers are in with respect to the word ‘intuitive’. I call this ‘the verbal virus theory of “intuition” proliferation’. Philosophers’ use of ‘intuition’ is a kind of intellectual/verbal virus (or tick) that started spreading about thirty to forty years ago. It is a bad habit and we should abandon it. (Cappelen 2012, p. 50)

Cappelen claims that there is neither a systematic use of the word ‘intuition’ (in ordinary language or in philosophy) nor does our practice of writing philosophy papers and books provide any evidence to the effect that there is a particular mental state the term would refer to. While the first part of Philosophy without Intuitions is concerned with the use of the word ‘intuition’ and related words, the second part is an extensive and careful examination of cases in which philosophers are commonly thought of as appealing to intuitions.

I will focus on the second part. The reasoning here goes, roughly, like this. If intuitions were to play a role as evidence in philosophy, there would be traces of an ‘effective’ or ‘non-idle’ use of them in texts written by philosophers. By an ‘effective’ or ‘non-idle’ use of intuitions Cappelen seems to have in mind a somehow detectable effect on philosophical texts (Cappelen 2012, pp. 115–116). The question of whether intuitions play a role in philosophical practice has thus to be answered empirically in the following sense:

Those interested in the role of intuitions in contemporary philosophical practice should be deeply engaged in an empirical study of that practice and be concerned with how to detect the presence of an appeal to the intuitive. Given the amorphous and shifty understanding of the intuitive in the philosophical tradition, it is not helpful to just look at a text and ask: Is there an appeal to intuitions in this text? Given the many understandings of ‘intuitions’ found among philosophers, a debate over this question, without further precisification, is worse than pointless. (Cappelen 2012, p. 130)

In an effort to make the question of whether there is appeal to the intuitive more precise, Cappelen aims to determine features which are easier to detect in philosophical texts and which are supposed to operationalize the appeal to intuitions. He is looking for the following (Cappelen 2012, pp. 112–113):

F1: Seem true/special phenomenology. An intuitive judgment has a characteristic phenomenology.

F2: Rock. An intuitive judgment serves as a kind of rock bottom justificatory point in philosophical argumentation. It justifies, but needs no justification.

F3: Based solely on conceptual competence. A correct judgment counts as intuitive only if it is justified solely by the subject’s conceptual or linguistic competence.

However, Cappelen does not find any of F1–F3 in the work of well-established and recognized philosophers. He concludes we have evidence that intuitions do not play a role in their work and in good philosophy more generally.

I cannot do justice to Cappelen’s extensive discussion of cases, but I nevertheless hope to demonstrate that the appeal to intuitions cannot be operationalized in the way Cappelen thinks.⁷ Going trough F1–F3, I will focus on what I consider the main arguments Cappelen gives concerning the following three examples from epistemology: Cohen’s Lottery Cases, Lehrer’s Truetemp Case, and Goldman’s Fake Barn Case.⁸

3.1 F1: Seem true/special phenomenology

Cappelen claims that in the respective cases, he ‘cannot, even with the best of will, discern a special feeling’.⁹ He acknowledges that he can only speak for himself and mentions that he has only ‘anecdotal evidence that some intuition-theorists have these special feelings when they contemplate certain propositions’ (Cappelen 2012, pp. 117–118).¹⁰ However, the respective authors of the three cases we are looking at do not mention any phenomenology or special feeling, e.g., Cohen:

At no point does Cohen implicitly or explicitly indicate that he thinks the phenomenology of [his judgments is] argumentatively significant. [M]aybe some people have distinctive phenomenology when they contemplate one or both [cases], but that phenomenology plays no effective argumentative role in this discussion. (Cappelen 2012, p. 166)

Clearly it does not follow from the fact that Cohen or others do not mention any phenomenology that there is no such phenomenology or that it does not play a role. First, Cohen might experience it but simply be unaware of the role it plays. Second, he might be aware of it but still not mention it. To compare: I may believe that my sensory seemings cause or justify certain beliefs, but it does not follow that I necessarily mention them when I use my beliefs in argumentation. For instance, if I believed that the sensory seeming that there is a chair caused and justified my respective belief that there is a chair, this does not mean that I would necessarily mention the sensory seeming. With no defeater being around, I might move straight to the belief that there is a chair. Hence, there might be a special phenomenology coming along with seemings which cause and possibly justify certain beliefs even if we do not mention them as such. Again, my aim is not to establish that there is such a phenomenology and that it plays a causal or justificatory role, but it seems obvious that simply looking at what philosophers say or write does not establish anything about the existence or non-existence of seeming states or about their role in the process of belief formation or justification.

3.2 F2: Rock

Cappelen’s main argument against F2 seems to be that rather than providing justification for a certain view, the cases cause puzzlement. With respect to Cohen’s cases, Cappelen says:

The claim that philosophers rely unquestioningly on default unjustified, 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 feature indicates that the Rock feature is absent. (Cappelen 2012, p. 165)

In the same line with Deutsch, Cappelen moreover takes the fact that some authors proceed to give arguments after delivering the case to be evidence against the intuition having Rock status:

If in a context C, evidence and arguments are given for p and those arguments [and evidence play] a significant argumentative role in C, that is evidence that p is not Rock relative to C. (Cappelen 2012, p. 121)

To begin with, is not clear to me where the idea that philosophers rely unquestioningly on unjustified, immediate reactions comes from [originally, the feature is stated differently: the intuitive judgment needs no justification, see Cappelen (2012, p. 112) and F2 above]. More importantly, it is not controversial and certainly no surprise that cases like Cohen’s, Lehrer’s, and Goldman’s are often a source of puzzlement. As counterexamples to philosophical theories, they are puzzling because they provide evidence against some of our established beliefs. But it is also not controversial that our intuitions have to be questioned. Even contemporary rationalists such as Bealer (1998) think that intuitions are fallible and hence some of them mislead us. It would thus be unreasonable even for rationalists to unquestioningly endorse their content. One explanation of why some philosophers give arguments in addition to a thought experiment could be that these arguments serve to determine whether an intuitive judgment really is one that provides justification while not being in need of justification. Another explanation could be that in the case of a counterexample to a strong and widely-accepted theory, we might want to gather as much evidence as possible before we change the theory or give it up entirely, which might mean to provide some additional arguments. Koksvik (2013) has argued that an intuition can provide foundational justification and yet be the result of conscious reasoning.¹¹ Koksvik asks us to imagine that he does not grasp de Morgan’s laws, and that we give an argument which shows that de Morgan’s laws hold (Koksvik 2013, p. 712). It is possible that at the end of the argument it seems to him that de Morgan’s laws are true. Koksvik concludes that

[...] while a proposition which is the conclusion of an argument does not have foundational justification if it is justified only because it follows from premises that are also justified, it is no bar to its having foundational justification in virtue of being the content of an intuition that what allows, prompts, or causes the intuition to arise is a conscious reasoning process. (Koksvik 2013, p. 713)

There could, moreover, be not purely epistemic reasons why an author may give arguments. First, the arguments could serve a heuristic purpose, i.e., they could help us to get to the truth without actually playing a justificatory role. Usually, one would think that intuitions are used for heuristic reasons, and that the arguments in support of their contents bear the epistemic burden. However, a view according to which our intuitions have Rock status and our arguments merely serve to rationalize them is not completely implausible. Second, an author might give arguments merely for conversational or rhetorical reasons. While it is obvious to the author that the intuition has Rock status, it is not to their audience. In order to convince the audience, it might be easier to give arguments for the truth of the content of the intuition than for the claim that the intuition has Rock status. Third, arguments might be required in order to get a paper published. Surely there are certain presentational requirements on philosophical texts which favour certain methods over others. Maybe it is not sufficient in academic philosophy to simply give the thought experiment without support from arguments. In all three cases just mentioned, intuitions could still provide the relevant justification and Rock could still be true. Again, it turns out that simply looking at the texts philosophers produce does not tell us anything about the epistemology of intuitions.

3.3 F3: Based solely on conceptual competence

Cappelen claims that given the intense controversy over the question whether there is conceptual justification, what concepts are, what conceptual competence is, what is required for conceptual justification, and which propositions are conceptually justified, the idea that philosophers can effectively rely on this kind of justification is prima facie implausible. Cappelen is looking for the following features in philosophical writing: whether the author mentions that or explicitly discusses how her judgment is based on conceptual competence, whether the judgment is of a necessary truth, if there are similarities to paradigm cases of conceptual truths, and if F1 or F2 are present (Cappelen 2012, pp. 124–128). The presence of any of these features would count as evidence for the presence of F3. However, Cappelen does not find any of them in any of the cases he is looking at.

Here is what Cappelen says about Lehrer:

What is clear from simply reading the text is that Lehrer doesn’t say that [giving conceptual justification] is what he is trying to do, he never tells us that he has no interest in an answer that relies on information that is not required for concept possession. Had he imposed such a constraint on his theorizing we should expect him to tell us that and to tell us what he takes concepts to be, what he takes concept possession to be, what he takes the relevant kind of justification to be. (Cappelen 2012, p. 168)

In contrast to Lehrer, Goldman did indeed express a view concerning the role of intuitions in philosophy, albeit well after his discussion of the Fake Barn Case. In his paper ‘Philosophical Intuitions: Their Target, Their Source, and Their Epistemic Status’, Goldman explicitly defends the view that our judgments in thought experiments are based solely on conceptual competence (Goldman 2007). Cappelen claims that Goldman does not relate his theory of concepts and concept possession to any particular argument in his paper, and that we would expect him to do so if it was important to his arguments (Cappelen 2012, p. 173).¹² However, in the first paragraph of the paper discussing the case, Goldman does note that the theory he will suggest seeks ‘to explicate the concept of knowledge’ (Goldman 1976, p. 771). Given that the Fake Barn Case and similar cases play a significant role in Goldman’s paper, it seems plausible to think that they are intended to help to explicate our concept of knowledge. But even if there was no explicit reference to Goldman’s metaphilosophical view, it would hardly be more charitable to assume that Goldman changed his view over time than to assume that in 1976, he more or less held the view he explicitly defended in 2007.

What could be the reason why philosophers do not mention their metaphilosophical views or do not relate them to the discussion of first-order philosophical problems? I take the most charitable interpretation to be that they do not do so because it is not necessary to express one’s metaphilosophical view in order to contribute to the discussion of first-order philosophical problems. Philosophers agree on many forms of arguments and strategies, but they have different views on why exactly these strategies and arguments are successful or unsuccessful. Philosophers who think that in philosophy we are ultimately concerned with concepts do not exclusively engage in concept talk and do not exclusively rely on work engaging in this talk when discussing first-order problems. Rather, these philosophers give a certain interpretation of what we have access to and thus what we really mean when we make claims about knowledge: what we really mean when we say, e.g., ‘knowledge is not JTB’, is that our concept ‘knowledge’ is such that knowledge is not JTB. Besides the fact that it would be unnecessary and unnatural to add ‘the concept of’ to every subject matter under investigation, it would not be smart to impose one’s metaphilosophical views on the reader when discussing a first-order problem.

Cappelen seems to be aware of the compatibility of first-order philosophical discussions with various metaphilosophical views when he expresses the following:

The point to emphasize here is that those who disagree with Goldman’s metaphilosophical views should not for that reason alone dismiss his other contributions to epistemology and philosophy more generally. What he does in his non-metaphilosophical papers is open to many metaphilosophical interpretations. Those who reject Centrality (and Goldman’s views of conceptual analysis) can still find those papers persuasive and important. (Cappelen 2012, p. 173, fn. 23)

In the discussion of Goldman, Cappelen’s strategy has changed accordingly, and he seems concerned with establishing that his own metaphilosophy, including the view that there is no conceptual justification, is compatible with Goldman’s presentation of the Fake Barn Case. A deviation from Cappelen’s original aim becomes obvious when we look at how he describes one of Cohen’s Lottery Cases, quoted at length:

Cohen starts by asking a fairly simple question:

• Suppose S holds a ticket in a fair lottery with n tickets, where the probability n-l/n of S losing is very high. Does S know that his ticket will lose? ([ Cohen 1988],  p. 92)

He immediately gives us a hedged answer: ‘Although (if n is suitably large) S has good reasons to believe he will lose, it does not seem right to say that S knows he will lose. This remains true for arbitrarily large n’ (p. 92, Cappelen’s italics). Note that ‘seem’ is used as a hedging term and so this is not a full out endorsement of the answer. Cohen goes on to discuss a related case, where he gives another hedged answer in reply to a question:

• Now, suppose S learns from Jones, the person running the lottery, that Jones intends to fix the lottery so S will lose. Does S, then, know that he will lose? Better still, suppose S reads in the paper that another ticket has won. In both of these cases we are inclined to say that S does know that he loses. [(Cohen 1988, p. 92), Cappelen’s italics]

[...] Cohen immediately goes on to point out that the answer he is inclined to give to the second question is in tension with the answer he is inclined to give to the first question:

• In the first case, it seemed, contrary to fallibilist assumptions, that as long as there is a chance that S wins, no matter how small, he does not know that he loses. But the other two cases indicate otherwise. There we said that S can know, on the basis of his reasons, that he will lose. But surely his reasons do not entail that he loses. Generally reliable sources lie, have their intentions thwarted, make mistakes, etc. The probability that S loses conditional on these reasons is less than 1. [(Cohen 1988, p. 92), Cappelen’s italics]

(Cappelen 2012, p. 164)

To describe the case, Cappelen uses language appropriate to his view introduced in previous chapters of the book, according to which ‘intuitive’ is often used as a hedging term. Cohen, however, uses ‘it seems’ and ‘inclination to believe’, and even ‘intuition’ (Cohen 1988, p. 92). Clearly this terminology could be seen as evidence for a meta-theory which involves intuitions as evidence. In a footnote, Cappelen admits that his description is biased:

One option mentioned [in the first part of the book] was to interpret ‘intuitive’ as ‘pre-theoretic commitment’. I take that to be a charitable way to reinterpret the use of intuition-talk in [the relevant passage in Cohen]. (Cappelen 2012, p. 167, fn19)

This gives rise to the suspicion that all Cappelen has shown is that the cases can be interpreted and described in the light of his own metaphilosophical view. This, however, defies his declared project, which is to empirically test whether there are intuitions involved in these cases. But it demonstrates an important point: that we can usually apply different metaphilosophical views to first order investigations. And it also explains why Cappelen finds it difficult to detect an appeal to the intuitive in philosophical texts:

[...] those who claim philosophers rely extensively on intuitions are not making it easy to support or criticise their view. All the key features they claim intuitions have are difficult to effectively detect in a particular text or philosophical exchange. [...] Of course, this will also seem to be bad news for the kind of project I’ve set myself. (Cappelen 2012, pp. 128–129)

Cappelen nevertheless confidently adds that ‘[...] it will become very clear that philosophy is not a domain where anything like the intuitive plays an important role’.¹³ He concludes that

[...] careful reading of the text reveals no evidence that the propositions in question have these features [i.e., F1-F3], and so we should stick with the simpler deflationary account. (Cappelen 2012, p. 173)

While we can’t expect any evidence in favour of or against philosophy without intuitions from a philosophical text, simplicity may still speak for such a view. Surely, a metaphilosophy without intuitions is likely to be ontologically more sparse and thus in some sense simpler than a theory which includes such states and a particular role for them to play. However, simplicity is only one criterion out of many, and other theories might have other advantages. If we think that intuitions are, e.g., seeming states, and that this view has relevant epistemic advantages which outweigh the criterion of simplicity, Cappelen has done nothing to convince us.

4 Determining the Role of Intuitions

My discussion of Deutsch’s and Cappelen’s arguments concerning the role of intuitions in philosophy raises the more general question of how to decide whether we rely on intuitions or not. At first sight, an obvious question to ask is whether we use the term ‘intuition’ or similar terms in the premises of our arguments. This can easily be answered by looking for the word ‘intuition’ or similar terms in philosophy papers. The result may be that in contemporary analytic philosophy, the word is used quite frequently, but not so in the history of philosophy. We have seen that nothing follows from this.

Cappelen thinks that we should look at our practice of writing philosophy papers. He is right that it is not helpful to simply look at a text in order to answer the question whether there is an appeal to intuitions, but the solution is not to define more specific features that have been ascribed to the intuitive and search for those, since whether F1–F3 can be found in a philosophical text is as much theory-dependent as the answer to the original question. Any of the cases he discusses can possibly be interpreted in the light of his theory of intuitions (or, better, of the lack of them).

Whether we rely on intuitions in philosophy cannot be determined by empirically investigating philosophical practice. Philosophers who disagree about the ontology and epistemic role of intuitions disagree deeply about epistemic and methodological matters. Their respective view on what intuitions are is tightly linked to what they think the nature of philosophy is and to how they answer key methodological questions. Bealer (1998) thinks that philosophy is autonomous in that central questions in philosophy can in principle be answered without relying substantively on the sciences. What distinguishes philosophical questions from others is their universality, generality, and necessity. Intuitions play a fundamental role in philosophy in that they provide a justificatory procedure for the answers to these kinds of questions. Sosa’s (2007b) view on intuitions, for instance, is an integral part of his general virtue epistemology, according to which our intuitions and perceptions are manifestations of our epistemic competence. For Williamson (2007), the ontology and epidemic role of intuitions follows from his thesis that philosophy is not in principle different from the sciences or from ordinary thinking. Philosophy therefore does not need a particular justificatory procedure such as intuitions, and intuitions are just ordinary judgments.

Whether we rely on intuitions in philosophy has thus to be decided within the framework of a theory of knowledge, justification and philosophical methodology. Each of the views concerning intuitions and their role in philosophy mentioned above need to be defended and have been defended in the literature. In the process of deciding which theory of intuitions is correct, simplicity will be only one criterion. Accordingly, the challenge from experimental philosophy has to be and has been addressed on the level of these theories (see e.g., Williamson 2011; Sosa 2007a).