Concept Revision, Concept Application and the Role of Intuitions in Gettier Cases

Abstract

The aim of the paper is to determine the role of intuitions in Gettier cases. Critics of the Method of Cases argue that arguments developed within this method contain a premise that is justified by its intuitiveness; they also argue that intuitions are unreliable sources of evidence. By contrast, Max Deutsch argues that this critique is unsound since intuitions do not serve as evidence for premises. In Gettier cases, an intuitive premise is justified by other arguments called G-Grounds. I propose a different view on the role of intuition in Gettier cases. I introduce a distinction between concept-revision arguments and concept-application arguments. On the basis of this distinction and Craig's and Spicer's distinction between intuitions of intension and intuitions of extension, I show that (1) intuitions of extension do not serve as evidence for any G-Grounds; (2) intuitions of intension do play an evidential role for all G-Grounds, but (3) in the case of G-Grounds which are used as concept-application arguments, they are a poor source of evidence, while (4) in the case of G-Grounds used as concept-revision arguments, they could be a reliable source of evidence if they are intuitions of intension of a speaker immersed in philosophical discourse.

1. Introduction

The idea that intuitions play an evidential role in philosophy is accepted by many participants in the current metaphilosophical debates (see Bealer 1998; Cappelen 2012: 1). This assumption was articulated by Cappelen as follows (following him, I will call it ‘Centrality’):

Centrality (of Intuitions in Contemporary Philosophy): Contemporary analytic philosophers rely on intuitions as evidence (or as a source of evidence) for philosophical theories. (Cappelen 2012: 3)

Despite its enormous popularity, Centrality was attacked (e.g. Williamson 2007; Deutsch 2010, 2015, 2016; Cappelen 2012; Horvath 2022).

The importance of this dispute is revealed when we consider the Method of Cases. The main idea of this method is that when analysing some concept,¹ e.g. ‘knowledge’, ‘good’, ‘justice’ or ‘beauty’, we propose some theory of this concept and then consider possible counterexamples. If we find a genuine counterexample, then we should reject the theory.

However, proponents of the negative programme of experimental philosophy (hereafter: Negative Xphi) argue that the Method of Cases is unreliable (see e.g. Weinberg et al. 2001; Stich and Tobia 2016; Knobe and Nichols 2017).² They claim that in arguments formulated by this method there is always a premise that is allegedly justified by its intuitiveness (henceforth, I will refer to such a premise as the ‘intuitive premise’). However, as the Negative Xphi studies suggest, intuitions are sensitive to philosophically irrelevant factors (such as culture, gender or order of presentation). If the Negative Xphi critique is sound, then one of the most important philosophical methods turns out to be unreliable.

One of the best-developed defences of the Method of Cases was introduced by Max Deutsch (2010, 2015, 2016). He argues that the Negative Xphi's critique misses the point since it accepts the Centrality claim, which is in fact false. Deutsch analyses paradigmatic examples of the Method of Cases: Gettier's argument against the JTB theory of knowledge and Kripke's argument against descriptivism. He shows that in these examples what justifies the allegedly intuitive premises in the arguments proposed by Kripke and Gettier are not intuitions but other arguments, which he calls K-Grounds and G-Grounds respectively.

The aim of this paper is to determine the role of intuitions in Gettier cases. I adopt Deutsch's main idea that not intuitions, but arguments, which he calls ‘G-Grounds’, justify the clue premise in Gettier's argument. I then examine whether intuitions can be and are used in support of the premises of G-Grounds. I show that in some cases these arguments are in fact justified by intuitions, which sometimes can be a good source of evidence. This possibility appears in arguments that are formulated in favour of revising some concept; intuitions which support these arguments are intuitions of speakers who are immersed in philosophical discourse and who express their expectations about a target concept. The results of the article show that although Deutsch's aim to eliminate intuitions from philosophical justifying practice cannot succeed, the Method of Cases could be interpreted as opening a way to defend traditional philosophical methodology against the Negative Xphi's critique. Regarding the former issue, even if we accept that G-Grounds serve as evidence in Gettier's arguments, they still must be justified by some kind of intuitions; however, regarding the latter issue, in some interpretations G-Grounds could be justified by intuitions which are good source of evidence.

As mentioned, the starting point for my consideration is Deutsch's main idea on the lack of intuition's evidential role in Gettier cases. His account, just as other approaches that deny the Centrality claim, is as controversial as it is influential (for critique of Deutsch's account see Devitt 2015; Colaço and Machery 2017; for its defence see Horvath 2022; for other accounts that deny Centrality see Williamson 2007; Cappelen 2012). My assumption is that it is worthwhile to put the controversies about Deutsch's main claim aside for a moment, and investigate its consequences and possible interpretations in more detail, so in what follows I will not argue in favour of Deutsch's main stance, but analyse those possible consequences instead.

The structure of the paper is as follows. In §2, I reconstruct the Negative Xphi's critique and Deutsch's response to it. In §3 I introduce a distinction between concept-application arguments and concept-revision arguments. I show that G-Grounds can be interpreted both as concept-application arguments that are formulated in favour of ascribing knowledge to the protagonist from Gettier's scenario, and as concept-revision arguments that are formulated in favour of adopting a certain concept of knowledge. Then, in §4, I present Craig's (1990) and Spicer's (2008, 2010) distinction between intuitions of intension and intuitions of extension. After introducing these distinctions, in §5 I consider the question whether intuitions (of either of the two types) can justify premises in G-Grounds (interpreted as either concept-application or concept revision arguments). I show that intuitions of extension serve neither as evidence in G-Grounds used as concept-revision arguments, nor as concept-application arguments; however, intuitions of intension in turn serve as evidence in G-Grounds that are used as both kinds of arguments. However, intuitions can be a good source of evidence if they are intuitions of intension of philosophers competent in epistemological discourse, but only in the case of concept-revision arguments. The last case is interesting not only because it makes possible to defend the reliability of Method of Cases against the Negative Xphi's critique, but also contradicts Deutsch's view, since it opens the possibility of some kinds of intuitions serving an evidential role. I conclude by arguing how the proposed account could be adopted for interpretation of other arguments from other philosophical discussions.

2. The Negative Xphi and Deutsch on the reliability of the Method of Cases

Consider the paradigmatic example of the Method of Cases, Gettier's 10-coins case. The proponents of the Negative Xphi claim that in the Method of Cases there is always a premise which is justified by intuition. The content of such a premise concerns whether the case under consideration (e.g. Gettier's 10-coins case) falls under a certain concept (e.g. concept of knowledge). At the same time, empirical studies conducted by experimental philosophers show that intuitions are sensitive to philosophically irrelevant factors, such as culture, socioeconomic status or the subject's gender (see e.g. Turri 2013; Turri et al. 2015; Beebe and Undercoffer 2016; Machery 2017; Watermann et al. 2018). According to proponents of the Negative Xphi critique, these variations make intuitions an unreliable source of evidence (for some analysis of this critique, see Machery 2017).

I propose formulating the Negative Xphi argumentation against the Method of Cases as follows:

1. (N1) Each argument formulated as part of the Method of Cases contains a premise that p, which is justified by the intuitiveness of p.

2. (N2) The intuitiveness of p varies with philosophically irrelevant factors.

3. (N3) If the justification of a premise in a philosophical argument varies with philosophically irrelevant factors, then the premise is unjustified.

4. (NC) Hence: The premises of arguments formulated as part of the Method of Cases are unjustified.

One way to oppose the Negative Xphi is to reject Centrality, and thereby dismiss the first premise (N1) of the Negative Xphi argument. One of the most widely discussed positions of this type was formulated by Deutsch (2010, 2015, 2016). According to him Negative Xphi mistakenly reconstructs the structure of the Method of Cases. Deutsch argues that the Method of Cases is describing a counterexample for one criticized thesis. This counterexample has the form ‘There is an F that is not a G’. In the 10-coins case, F is Smith's justified, true belief that the man who has 10 coins in his pocket will get the job; G is knowledge. Hence, the proper reconstruction of such arguments is as follows:

1. (1) There is an F that is not a G.

2. (2) Hence, not all Fs are Gs.

However, proponents of the Negative Xphi think that premise (1) is justified by its intuitiveness. Hence, according to them, we should add premise (0):

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

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

3. (2) Hence, not all Fs are Gs. (Deutsch 2010: 451)

However, it is controversial that a judgement allegedly justified by its intuitiveness takes the form of a premise (1). The judgement expressed by premise (1) is a general judgement concerning the existence of something, e.g. a justified and true belief, which is not knowledge. According to proponents of Negative Xphi, however, what is intuitive is instead that ‘the particular Smith described by Gettier has a justified true belief that the man who has 10 coins in his pocket will get the job, but he does not know it’. Let us call this kind of judgement – that in some specific case a particular F is not a G – judgement J. Negative Xphi argues that a judgement justified by its intuitiveness states that Smith's belief is not knowledge. So, J takes the following form in the 10-coins case:

(J1): Smith's belief that the man who has 10 coins in his pocket will get the job is not knowledge.

Let us consider the justification of (J1). According to Deutsch (2010: 452–3) intuition that p may be the causal source of the judgement, but it cannot be its justificatory source. The evidence for the judgement that p is, hence, not its intuitiveness but some arguments that are independent of intuition. The Method of Cases consists in presenting a specific thought experiment and arguing that in this thought experiment some F is not G, and hence rejecting the thesis that ‘all F's are G’. In the 10-coins case, Gettier and his commentators present some arguments that Smith has no knowledge that the man who has 10 coins in his pocket will get the job. Deutsch calls the kind of arguments formulated to justify (J1) ‘G-Grounds’.

Deutsch gives three examples of such arguments (2010: 454–5). Below I will have a closer look at one of them, which I will call the Disconnect G-Ground:³

The Disconnect G-Ground:

As Gettier himself says in his original presentation of the 10-coins case, it is clear that Smith does not know that (e) [‘the man who will get the job has 10 coins in his pocket’] is true; for (e) is true in virtue of the number of coins in Smith's pocket, while Smith does not know how many coins are in Smith's pocket and bases his belief in (e) on a count of the coins in Jones's pocket, whom he falsely believes to be the man who will get the job (1963, 122). The thought here, … is … that S's justified true belief that p might fail to be knowledge if there is a disconnection between (i) what causes S to believe p, and (ii) what makes S's belief that p true. In the 10-coins case, it is the number of coins in Jones's pocket that is (partly) causally responsible for Smith's belief that the man who will get the job has 10 coins in his pocket. But that belief is true ‘in virtue of’ the number of coins in Smith's pocket. According to Gettier, this disconnection between what causes Smith to believe and what makes Smith's belief true justifies the judgement that Smith does not know. (Deutsch 2010: 454–5)

The Disconnect G-Ground contains at least two premises, which I will call ‘(G1)’ and ‘(G2)’, and its conclusion should be (J1):

1. (G1) Smith believes that ‘the man who will get the job has 10 coins in his pocket’ on the basis that (i) Smith believes that Jones has 10 coins in his pocket, and (ii) Smith believes that Jones will get the job.

2. (G2) The facts that make Smith's belief true are (i*) Smith has 10 coins in his pocket, and (ii*) Smith got the job.

3. (J1): Hence: Smith's belief that the man who has 10 coins in his pocket will get the job is not knowledge.

However, premises (G1) and (G2) do not suffice to conclude that Smith's belief is not knowledge. An enthymematic premise must be added to this argument; hence, we should add a possible enthymematic premise (G3):

1. (G3) For each X, if X knows that p, then some facts that led X to acquire the belief that p are truthmakers for p.

(G1), (G2) and (G3) entails that Smith's belief is not knowledge. Judgement (J1) is then justified not by any intuition but by an argument – the G-Ground, strictly speaking.

It is worth noting that Deutsch not only maintains that the intuitiveness of the fact that Smith's belief is not knowledge does not justify (J1), but he insists that in Gettier's argument there is also no other premise that is justified by its intuitiveness. As he emphasises: ‘Gettier's refutation of the JTB theory appeals to reason and argument at every stage; there is no appeal to intuitions, and, in particular, no appeal to or reliance on intuitions in his arguments for his claims about his thought experiments’ (Deutsch 2015: XIX).

In summary, Deutsch accepts the following theses:

• (D1): No premises in the arguments formulated as part of the Method of Cases is justified by its intuitiveness.

• (D2): (J1) is justified by arguments (G-Grounds) and not by intuitions.

Deutsch rejects the critique formulated by Negative Xphi on grounds of (D1) and (D2). If they are true, then premise (N1) in Negative Xphi's argument is false.

However, there is a possibility that some G-Grounds could be justified by intuitions. Therefore, in order to examine whether Deutsch's claims are entirely true, we have to determine whether all G-Ground's premises are not justified by intuitions.

The aim of the two following sections is to discuss two distinctions: the first one is between two kinds of arguments; the second one is between two kinds of intuitions. These distinctions are crucial, since to determine whether intuitions serve as evidence in G-Grounds, we should take into account all known kinds of intuitions that could play an evidential role in these arguments, as well as all known kinds of G-Grounds in which such a role could be played.

3. Concept-application Arguments and Concept-revision Arguments

There are two kinds of arguments that could be formulated in favour of a claim of the form ‘There is an F that is not a G’ (judgement J): concept-application arguments and concept-revision arguments. In this section I will introduce a distinction between them. I will also show that G-Grounds could serve as both of these kinds of arguments. This remark will be crucial to show that whether Deutsch's theory of the role of intuitions in the Method of Cases is valid or not depends on the interpretation of G-Grounds.

Consider a story that illustrates the difference between concept-application and concept-revision arguments.

Imagine a football match. The ball rolls towards the goal. When it is near the goal line, the defender kicks it out of play. Two people watching this match are discussing whether a goal has just been scored. Both know that according to the rules set by FIFA a goal is scored when the whole of the ball passes over the goal line with its entire circumference. In this case, they do not discuss the rules set by FIFA; instead, they discuss whether the ball has crossed the goal line or not. The disagreement in this regard may result, for example, from an eyesight defect, but not from a disagreement regarding the rules of football. One of these people could argue that the ball has crossed the goal line because the assistant referee who was close to the goal said so, and hence the goal is scored. I will call arguments of this type concept-application arguments.

Concept-application arguments assume that we assess whether a given situation falls under some concept (in our example, the concept of ‘goal’), having previously accepted some specific conditions of using this concept (the entire circumference of the ball must cross the goal line). In concept-application arguments, we try to show that a situation being scrutinized meets the accepted conditions for using a certain concept.

However, if this situation is being considered by FIFA members, they may wonder whether the rules of the game should be changed. They can agree and be certain that the ball has not crossed the goal line; however, they can discuss what the rules for scoring a goal should be. One such argument may, for example, relate to the fact that referees are less likely to be mistaken if the ball touches the goal line than if the entire circumference crosses it. For this reason, some FIFA members may claim that the goal in the match was scored even though it did not cross the goal line. The argument in this case concerns the rules of football. I will call arguments of this type concept-revision arguments.

In concept-revision arguments, we do not accept any definition for a given concept as a common-ground. The details of the particular situation are not under scrutiny. We assume that we know enough about the situation that we are trying to assess to decide whether it would fall under any concept if it were defined in (reasonable) terms. In concept-revision arguments, we try to show what conditions should be imposed on the target concept.

Note that these kinds of arguments differ in their role but not in what kind of premises they use. It is possible that one argument could be interpreted both as a concept-application argument and as a concept-revision argument. The argument that ‘the goal wasn't scored because the ball crossed the goal line after the end of the match’ could be used as concept-application argument if we assume that a goal is scored when the ball crosses the goal line during the match. However, such an argument could also be used as a concept-revision argument if we argue that we should accept that the goal could be scored only during the match.

We can use this distinction to analyse G-Grounds, which could be interpreted both as concept-application arguments as well as concept-revision arguments. If some G-Ground is an argument in which a specific concept of knowledge is initially accepted and it is proved whether or not Smith's epistemic state meets its conditions, it is a concept-application argument. In Smith's case, G-Grounds could serve as concept-application arguments in favour of the thesis about the particular epistemic state of a particular person. I will refer to G-Grounds that are used as concept-application arguments as ‘application G-Grounds’.

G-Grounds, however, can also be treated as arguments against a certain definition of knowledge but not as arguments against the attribution of knowledge to Smith. They may be formulated for or against a particular theory of knowledge. They could justify a particular stance on what knowledge really is. This interpretation of G-Grounds leads to, as I will call them, revision G-Grounds, i.e. G-Grounds which are used as concept-revision arguments.

The distinction between concept-revision and concept-application arguments may look similar to the distinction between the method of conceptual analysis and conceptual engineering (see Cappelen 2018). The conceptual engineering is contrasted with conceptual analysis mainly in terms of the purpose and the method of investigating philosophical concepts. While conceptual analysis aims to discover the truths about concepts, so the central method is precisely an analysis, the central concept in the metaphilosophy adopted by conceptual engineers is an explication. Explication of a given concept takes into account its colloquial usage to some extent. However, explication's purpose is to present a given concept within a different and a more precise conceptual framework, and thus also in a different theoretical context. For this reason, within the explication of a given concept, its meaning is regulated or ameliorated.

Concept-revision arguments seem perfectly accurate for explication-centred metaphilosophy, such as conceptual engineering. Let me note that despite their growing popularity, conceptual engineering and the method of explication aroused a lot of controversy (see e.g. Strawson 1963; Deutsch 2020; but see also Carnap 1962; Koch 2021). Putting the discussion on conceptual engineering aside, let me stress that revision arguments might be incorporated to conceptual analysis as well, once one accepts that at least sometimes within this method some ameliorative arguments might be formulated (for arguments in favour of this view see e.g. Díaz-León 2020). That means that even if one is sceptical towards the project of conceptual engineering, concept-revision arguments might work within the framework of conceptual analysis as well.

4. Intuitions of extension and intuitions of intension

In the last section, I have shown that there are two kinds of G-Grounds. In this section I present the second important, but largely forgotten distinction between intuitions of intension and intuitions of extension introduced by Craig (1990) and developed by Spicer (2008, 2010), which will be useful in exploring the question of whether intuitions justify the premises of G-Grounds.

Intuitions of extension are intuitions about the epistemic state of an agent in a particular situation. We can get to know someone's epistemic intuitions of extension simply by asking questions like ‘Does John [in a given case] know that X?’.

Another type is intuitions of intension, which differ from intuitions of extension in terms of their content. Epistemic intuitions of intension are about general properties of knowledge. They are related more to the very concept of knowledge than to the epistemic state of a particular subject. They are manifested in answers to questions like ‘Can a person know something without believing it?’ or ‘Is knowledge certain or not?’.

It might be tempting to reduce intuitions of intension to intuitions of extension, since one could argue that the general proposition about knowledge could be such that knowledge is instantiated by the set of particular subject's beliefs, or that the content of any intuition of extension entails the content of a particular intuition of intension (and, at least sometimes, vice versa). However, although the evaluation of the adequacy of the intuition of extension/intension distinction is outside the scope of this paper, let me show some rationales for accepting it.

Some reasons come from the studies on folk epistemologies and epistemic intuitions, which are conducted within two different research paradigms. Folk epistemologies are studied mostly by developmental psychologists. They usually ask participants for their opinions on general properties of knowledge. Intuitions of intension are therefore indicators for a clue variable for this research, namely the content of folk epistemology. On the other hand epistemic intuitions are studied by experimental philosophy, and indicators for having a specific epistemic intuition is an expression of a proposition that is a content of intuition of extension (see Hardy-Vallée and Dubreuil 2010). As empirical results show, folk epistemologies and epistemic intuitions are sensitive to different factors and in different manners. For example gender differences regarding folk epistemologies are much bigger than regarding epistemic intuitions, if the latter are sensitive to gender at all (compare Baxter Magolda 2001; Karabenick and Moosa 2005; Buckwalter and Stich 2015; Seyedsayamdost 2015). It seems therefore that accepting intuitions of extension and intuitions of intension are partly independent, and that they at least sometimes are incompatible.⁴

5. The Evidential Role of Intuitions in G-Grounds

Given the two distinctions between concept-revision and concept-application arguments, and between intuitions of intension and intuitions of extension, we can return to the main question: do intuitions play an evidential role in G-Grounds?

In the following sections, I adopt Deutsch's view (D2) and explore whether each kind of intuition can support premises in each kind of G-Ground. I show that even if (D2) is true, Deutsch's account of the role of intuitions in the Gettier cases is only partially valid. This is because (D1) is false. I argue that intuitions of extension do not serve an evidential role in either application or revision G-Grounds (§5.1), which is in line with (D1). However, I will show that intuitions of intension do play an evidential role in application G-Grounds (§5.2), which contradicts (D1). Moreover, intuitions of intension are not a good source of evidence for application G-Grounds, so although Deutsch is right about (D2), Negative Xphi's critique is valid in such a case. In §5.3, I demonstrate that intuitions of intension also serve as evidence in revision G-Grounds (Table 1). However, intuitions of intension could be a good source of evidence in this case. This last result is striking, since while it contradicts Deutsch's claim (D1), it still enables to achieve one of Deutsch's motivations – it provides a way to defend reliability of the Method of Cases from the Negative Xphi's critique. It can be done by rejecting the premise (N3) from the Negative Xphi argument, since, as I will show, the reliability of intuitions of intension in concept-revision arguments is independent of their stability, but comes from both the immersion in philosophical discourse of their holders and from the dialectic of concept-revision arguments which is normative and not purely descriptive in its nature.

Table 1. Summary of the results presented in sections §5.1–§5.3.

5.1. Intuitions of extension in application and revision G-Grounds

To determine whether G-Grounds could be justified by intuitions of extension, we have to determine whether any premise in such arguments expresses the possible content of intuitions of extension. On the example of the Disconnect G-Ground (see §4), in this section I will argue that there is no possible candidate for a premise that could express an intuition of extension.

The content of an intuition of extension could be a proposition which states whether or not a particular epistemic state falls under the ‘knowledge’ concept. What is crucial is that the content of an intuition of extension is about knowledge ascription and not about knowledge itself.

Now let us check whether in my reconstruction of the Disconnect G-Ground there is a premise that could be the content of an intuition of extension. The argument has three premises: (G1), (G2) and (G3) (see §2).

Neither (G1) nor (G2) is about knowledge ascription. Neither of them is about whether some entity falls under some concept. While intuitions of extension reflect how we categorize entities, (G1) and (G2) are about the relations between some facts in the world. (G1) states that there is a causal relation between Smith's belief and both (i) and (ii). (G1) is therefore about the causes of Smith's belief. (G2) states that between Smith's belief and both (i*) and (ii*) there is another relation, namely a relation which holds between true beliefs and their truthmakers. Hence, (G1) and (G2) do not express any intuitions of extension.

Since (G1) and (G2) cannot be the content of an intuition of extension, let us think about a proposed enthymematic premise (G3).

(G3) clearly is an example of the content of an intuition of intension. It is not about ascribing knowledge to particular subject, as this would make it an intuition of extension; it is rather a proposition about some general property of knowledge, as is characteristic of intuitions of intension. This property is a non-accidental connection between what makes some belief true and how this belief is acquired. Since (G3) is not about ascribing knowledge to someone, it could not be the content of any intuition of extension.

However, it is possible that (G3) is not the only enthymematic premise candidate that could make the Disconnect G-Ground work. There could be another enthymematic premise candidate that could be the content of some possible intuition of extension.

Consider an obvious candidate:

• (G3*) Smith does not know that ‘the man who has 10 coins in his pocket will get the job’.

There is, however, no connection between (G3*) and the two other premises. Conclusion (J1) results directly from (G3*), hence that argument does not need either of the other two explicitly expressed premises to be conclusive. Moreover, the content of this premise is just the same as the content of the conclusion. If we accept that (G3*) is the enthymematic premise of the Disconnect G-Ground, this argument turns out to be circular.

Consider then another possibility:

• (G3**) If Smith knows that ‘the man who has 10 coins in his pocket will get the job’, then the facts that led Smith to acquire the belief that ‘the man who has 10 coins in his pocket will get the job’ are truthmakers for the proposition that ‘the man who has 10 coins in his pocket will get the job’.

Note that the content of (G3**) is not the same as conclusion (J1). It is also not the case that conclusion (J1) follows directly from (G3**) without the other premises. Therefore, upon the acceptance of premise (G3**), there is no danger of a vicious circle, nor can this solution be accused of not taking into account the other premises of the Disconnect G-Ground.

Even though (G3**), (G1) and (G2) entail (J1), (G3**) cannot be seen as premise that is justified by an intuition of extension. This is because the content of (G3**) is not actually the content of an intuition of extension. Intuitions of extension are about the epistemic state of an agent in a particular situation; in contrast, intuitions of intension concern properties of knowledge itself. If a proposition states a condition for knowledge, then such a proposition is about some property of knowledge, and thereby such a proposition expresses an intuition of intension. The proposition expressed by (G3**) is a conditional proposition and is about the same property of knowledge as (G3). It is not about whether Smith has knowledge or not: it is about the conditions under which it is concluded whether or not Smith has knowledge. Hence, (G3**) is the content of a special case of an intuition of intension.

(G3**) expresses a singular and not a general proposition, which may suggest that it concerns a particular case in which we assess whether a given entity has knowledge. However, the difference between intuitions of intension and intuitions of extension does not apply to whether the content of an intuition is a particular or a general proposition. Intuitions of intension can be about specific particulars, nevertheless they remain intuitions of intension. An intuition of extension determines whether a given subject has knowledge or not. The content of an intuition of intension states in what case we can talk about knowledge. The content of an intuition of intension, however, can be both singular and general; it can therefore be both a proposition with the following content:

1. (I1) For each X, if X is an animal, then X is a dog,

as well as a proposition:

1. (I2) If Saba is an animal, then Saba is a dog.

On the other hand, the only proposition that can have analogous content in the case of intuitions of extension is the following:

1. (E1) Saba is a dog.

In fact, it may turn out that some intuitions of intension could be acquired as a result of considering intuitions of extension. For example, I could acquire (I2) as a result of considering (E1), which is a related biological intuition of extension of the form ‘Saba is a dog’, and other similar intuitions of extension like ‘Marshall is a dog’, ‘Marshall is an animal’, ‘Danny is a dog’, ‘Danny is an animal’ etc. However, this does not have to be the case. I could have intuition (I2) without considering whether Saba is a dog, but, for example, I could infer it from the general proposition (I1). Regardless of whether I accept (G3**) because I inferred it from a singular proposition (like (E1)) or from a general proposition (like I1), the content of (G3**) is the content of an intuition of intension but not of an intuition of extension.

In sum, the content of (G1) or (G2) or (G3) does not correspond to the content of an intuition of extension. Moreover I have also checked whether there are any plausible enthymematic premise candidates in the scrutinized G-Ground other than (G3), and I have shown that plausible candidates either lead to a vicious circle (as (G3*)) or expresses intuition of intension (as (G3**)). In view of the above considerations, intuitions of extension cannot justify any premise of the Disconnect G-Ground.

This result concerns both application and revision G-Grounds. Regardless of whether the Disconnect G-Ground is formulated in order to prove something about Smith's epistemic state or about knowledge in general, the premises of this argument remain the same. If none of the premises of such a G-Ground could be justified by an intuition of extension, it does not matter which interpretation of this argument we adopt. Hence, an intuition of extension cannot serve an evidential role in the case of application or revision G-Grounds.

5.2. Intuitions of intension in application G-Grounds

In this section, I show that intuitions of intension could justify the premises in application G-Grounds; however, they cannot be a good source of evidence because they are dependent on philosophically irrelevant factors. This conclusion is coherent with the Negative Xphi's critique and shows that application G-Grounds are unreliable sources of evidence.

Note that (G3) could be the content of an intuition of intension as it expresses a certain property of knowledge. As I already mentioned in the last section, this property can be expressed as a ‘non-accidental connection between what makes it true and how it was acquired’.

While considering application G-Grounds, in order to assess whether the epistemic state of the agent in a given scenario falls under the concept of knowledge, we must first adopt a specific concept of knowledge. In order to assess whether the Disconnect G-Ground is valid, it is only necessary to check whether or not the property expressed by (G3) is included in some specific concept of knowledge (hereafter, I will refer to the adopted concept of knowledge as K).

Consider now in virtue of what we could accept K, and how (G3) could be justified. The first possibility is that K is someone's private psychological concept. So, if a property described in (G3) is included in K, we can justify (G3) by the intuitiveness of (G3). However, such a justification could be accepted only by a person possessing K. In this case, the justification of (G3) is its intuitiveness, while (G3) is the content of an intuition of intension.

On the other hand, if K is a concept adopted under some convention, then why we adopt the given convention should be justified since it could be adopted for many debatable reasons. Such a choice can be made without any justification – just like the starting point for the analysis of the concept of knowledge; however, in that case the strength of a G-Ground would be negligible. Whether a G-Ground is valid would depend only on an arbitrarily chosen convention. If K is adopted without any justification just as a starting point for consideration, it would be bizarre to use such a concept in application G-Grounds. It is hard to imagine any reason why we would check whether a particular situation meets the arbitrarily adopted concept K, and then, in virtue of that, conclude that the given situation is a case of knowledge. More plausible is that someone adopts some concept for some reasons. Even if a concept used in the analysis of ‘knowledge’ is adopted under a convention, we could argue in favour of such a convention. One might indicate that we should adopt K because it is simple or is consistent with some other theory. If, however, we argue why we should adopt some concept of knowledge, and then on that basis we argue that Smith does not know that p, this would no longer be an application G-Ground: it would be a revision G-Ground, since revision G-Grounds differ from application G-Grounds in that the former are arguments for adopting some properties of the analysed concept of knowledge, but unlike application G-Grounds they are not arguments for ascribing a particular epistemic state to someone.

Thus, within application G-Grounds a certain premise may be justified by an intuition of intension. In the case of the Disconnect G-Ground, this premise is (G3). However, (G3) is justified only if its intuitiveness corresponds to the properties of the psychological concept of knowledge that the person evaluating the G-Ground possesses. These intuitions however vary with philosophically insignificant factors such as culture (see, e.g., Nisbett et al. 2001; Hofer 2008), gender (see, e.g., Baxter Magolda 2001; Karabenick and Moosa 2005) or even childhood experiences (see Bartsch 2002), which means that they are not a good source of evidence according to the Negative Xphi critique.

The following conclusions can be drawn from this section: intuitions of intension may justify some of the premises of application G-Grounds, but such intuitions are a poor source of evidence. These results support the Negative Xphi critique, but only in the case of application G-Grounds. To determine whether Negative Xphi's critique is sound or whether adopting (D2) makes it possible to defend the reliability of Gettier's argumentation, we also have to explore the role of intuitions of intension in revision G-Grounds.

5.3. Intuitions of intension in revision G-Grounds

Finally, we are able to consider the last option. The previous section shows that (G3) could be the content of an intuition of intension. Moreover, we know that the conclusion of application and revision G-Grounds is the same: it is (J1). This means that intuitions of intension could be a source of evidence for one premise in a revision G-Ground, and therefore Deutsch is wrong about the role of intuitions in Gettier cases since there are some kinds of intuitions that support G-Grounds. However, in the next section I will argue that some kinds of intuition could be a good source of evidence in the case of revision G-Grounds. This would mean that although Deutsch is wrong about (D2), adopting (D1) makes it possible to defend Method of Cases against the Negative Xphi's critique.

Let me stress that the defence of reliability of Gettier's argument, presented below is an interpretation of Gettier's argument. I do not claim that this is its only possible interpretation. I do not claim also that Gettier himself, if asked, would agree that I correctly described his method. I insist however that this interpretation is plausible, and that it might be useful if one wants to defend the reliability of the Method of Cases.

The justification for the view that some intuitions of intension are a good source of evidence in Gettier's argument requires granting philosophers the specific status of ‘experts’. However, it is crucial that I am not saying that philosophers have some competences that are better developed in philosophers due to their philosophical training, as other proponents of the so-called expertise defence postulate (see e.g. Kauppinen 2007; Williamson 2016). I am also not saying that philosophers make more accurate verdicts about some cases.

In order to grasp the kind of expertise I'm referring to, we have to look closer at the role of intuitions of intension in concept-revision arguments. According to the view I am defending, while using a concept-revision argument we could appeal to our intuitions of intension that expresses our expectations towards this concept. For example, if I have an expectation towards the concept C that C should be such-and-such, I could express an intuition of intension that C is such-and-such. Note that it is not argued in such a case that C is such-and-such. It is argued that C should be such-and-such. Therefore, such an argument is normative and not descriptive in its nature.⁵ Moreover, note that our expectations concerning C could be more or less accurate in particular contexts. For example an expectation towards a philosophical concept of knowledge that knowledge should capture only funny beliefs is less accurate from the philosophical perspective than an expectation that knowledge should be certain. Similarly, an expectation that during playing hide-and-seek all children ought to learn something important about teamwork is less accurate from the perspective of children engaged in the game, than, for example, an expectation that the game should be as exciting as possible. On the other hand, from the perspective of parents, the former expectation towards the rules of hide-and-seek game might be more accurate, just as the expectation regarding the funny beliefs might be accurate from a perspective of someone who, let's say, construes an unusual world in a science fiction novel.

The status of ‘experts’ that I ascribe to philosophers bases on the fact that their intuitions of intension regarding philosophical concepts are more accurate in the context of philosophical debate. This follows from their both explicit and tacit knowledge about the purpose of the particular debate and arguments formulated within this debate in the past. This kind of expertise comes from philosophers’ immersion in a particular kind of activity, and it is the same kind of expertise as children from my neighbourhood have regarding the rules of their games, musicians have regarding the concepts or properties of musical activity, or mathematicians have regarding mathematical concepts. All of these groups create a part of reality and since their expectations towards this part of reality partly constitute it, they are specific kinds of experts regarding it.

Now, consider the 10-coin case. We may argue that Smith does not know that e because his epistemic state does not meet our expectation towards the concept of knowledge, according to which if someone knows that p there must be a non-accidental connection between what makes p true and how belief that p was acquired. This expectation was actually mentioned by Gettier in his argument (see Deutsch's presentation of the Disconnect G-Ground in §2). It seems plausible that he expressed this stance since it is intuitive that knowledge should meet the mentioned expectation. Gettier was well informed about, for example, aims of philosophical analysis of knowledge, doubts raised in the past etc, and his expectations are formed partly on the basis of this awareness. The argument was formulated within an epistemological dispute, and therefore Gettier's expectation expressed by his intuition of intension, forms a reliable source of evidence.

It is important to note that philosophical concepts, even if homonymous with everyday concepts, could be slightly different from them because they serve specific purposes defined by philosophers during analyses. An example of such a purpose may be to fit concept ‘A’ into the established theoretical framework containing ‘A’. For example, concept of knowledge should be compatible with concept of justification established within some theory. Such expectations towards the consistency of theoretical framework are negligible for most people but they are crucial for both philosophers and scientists. There are however also some expectations which are not shared by philosophers and scientists, but specific for philosophers and were formed during the history of, for example, epistemology. For instance, according to Craig (1990: 6) how philosophical concept of knowledge differs from the folk one is that the former should face a sceptical challenge in some way just because of philosophers’ expectations, while the latter does not. Nowadays, it is important that such a concept should be immune to Gettier cases. Since Gettier (1963) wrote his famous paper and others wrote some answers to it, it is also clear that philosophers require that knowledge cannot be true just by accident.

Hence, unlike the folk use of the concept of knowledge, the philosophical use must consider the tasks that have been imposed on the concept of knowledge over the course of the history of philosophical inquiry. To understand which concept is at issue in the Method of Cases, and therefore to properly interpret thought experiments, it is crucial to understand the theoretical context in which a particular thought experiment is investigated (see Cohnitz and Häggqvist 2016; Sękowski 2022). Therefore, since philosophical concept of knowledge emerges from the philosophical practice, people familiar with that practice acquire linguistic competence to use the term ‘knowledge’ that is in line with the expectations that epistemology imposes on it.

Note that the difference between philosophical and everyday concepts was described above as “slight”. Although there is a lot of such epistemological platitudes which are insignificant or even strange for lay people, a possible and plausible expectation is that the concept of knowledge ought to be, for example, coherent to some extent with the ordinary usage of the word ‘knowledge’. It is nicely expressed by one recent philosophical strand focused on the method of conceptual analyses: the so-called Canberra Plan represented by, for example, Chalmers, Jackson or Lewis (see Braddon-Mitchell and Nola 2009). According to the proponents of the Canberra Plan, the philosophical investigation should start with some platitudes, and then try to find out which entities satisfy most of these platitudes. It is assumed that, especially in the case of technical terms, not all folk beliefs ought to be included in the list that should be satisfied by the final definition of some term (see: Kornblith 2016). However, since a theoretical concept, such as the philosophical concept of ‘knowledge’, has to be defined partly with regard to the set of the most obvious platitudes about knowledge which are shared by philosophers, scientists and lay people, the philosophical concept of knowledge should not concern something entirely different from the folk concept of knowledge. The philosophical concept of knowledge cannot refer to some colour or a kind of cutlery as this would be extremely incompatible with philosophical expectations towards ‘knowledge’, which in this case coincide with common usage of the term ‘knowledge’.

The role of appealing to intuitions of intensions could be then as such, that it justifies a concept-revision argument, which is convincing if the expectation that is expressed by intuition justifying the argument is shared or at least looks plausible. There remains a question, however. What if two philosophers do not share the same intuition of intension or share two incompatible intuitions of intension?

It is a common situation in philosophical practice. Suppose that A would argue that the proper theory of concept of ‘X’ is so-and-so, because of i1, which is A's intuition of intension about X. However, B argues that A's theory of X should be refuted because of i2, which is another intuition of intension about X. The plausible prediction, coherent with Deutsch's picture of philosophical methodology, is that A and B would probably look for an argument which could help to decide which one is right. This is Deutsch's solution to the ‘relocation-problem’, according to which at some level all arguments rest on intuitions (2015: 122–7). He argues that even if G-Ground, or any other philosophical argument seems to rest on some intuition, we should be aware that there is always some possible argument which could justify that intuition. As mentioned, the explicit formulation of Gettier's argument contains a phrase which refers to the possible intuition of intension that justify Disconnect G-Ground and therefore we are able to claim that this argument serves as evidence in Gettier argument. However, as I mentioned in §3, Gettier's argument lacks any clue of some possible argument which could justify the intuition of intension that is a premise in Disconnect G-Ground. Therefore, even if an intuition of intension expressed by (G3) could be justified by some argument, it does not change the fact that (G3) is accepted, and it makes the Disconnect G-Ground sound, because it is intuitive.

Let me return to the example of A and B discussing the theory of X. It is now clear that the argument which might tip the scales in B's favour also would rest on some intuition of intension. Note that in such a case both A and B should abandon a hitherto accepted definition of ‘X’ and accept another. It is because B's argument based on intuitions of intension must be a concept-revision argument which, naturally, leads to conceptual revision. Of course, later, someone else – C, who is immersed in epistemological discourse, could propose another concept-revision argument based on intuition of intension i3. If A and B share i3, then the definition of X would be revised again. This illustrates how philosophers could approximate a definition of ‘X’ which meets all their relevant expectations.

Variety of philosophers’ intuitions of intension is not a problem when it comes to revision-concept arguments then. The reliability of intuitions of intension in a case of justifying concept-revision arguments does not come from stability of these intuitions, but from the facts that they express expectations of someone who takes part in a particular practice of negotiation of the target concept's meaning, and that this person is immersed in this practice. Epistemic intuitions of intension do vary across epistemologists. However, if a philosopher is a mindful and minimally competent participant of a particular dispute on a given epistemological concept, and if she is aware of at least the most influential expectations imposed on the scrutinized concept in the past within this dispute, then her intuitions of intension towards the meaning of the target concept are reliable, in a sense that they might be reasonably considered and discussed within this dispute.

In sum, even if we agree with Deutsch that the evidence for (J1) is arguments instead of intuitions, these arguments are supported by their own intuitiveness. This conclusion contradicts Deutsch's thesis (D1). However, in the case of revision G-Grounds, intuitions of intension could be a good source of evidence. This is the case if these are intuitions of intension of a speaker immersed in the context of a particular philosophical dispute. However, this conclusion contradicts one of Deutsch's views, and it also opens a way to defend the Method of Cases against the Negative Xphi's critique. Although intuitions of intension do serve an evidential role in the Gettier cases, they could be reliable.

6. Conclusion

In this paper I have adopted Deutsch's idea, that the intuitive judgements in Gettier cases are justified by G-Grounds. I have showed that although, pace Deutsch, in the Method of Cases intuitions could play justificatory role, there is a way to preserve his main motivation, i.e. to show a defence of the Method of Cases against the Negative Xphi's critique. This possibility opens if we consider the role of intuitions of intension in concept-revision arguments, since intuitions of intension of subjects immersed in epistemological discourse could form a reliable justificatory source for arguments that aim to revise philosophical concepts.

As mentioned before, the presented picture of the role of intuitions in Gettier case is a kind of interpretation of Gettier argument. Probably a controversial interpretation for some. I even agree that Gettier, if asked, might disagree with it. However I argue that Gettier's argument can be interpreted in the proposed way. Moreover, this interpretation seems worthwhile since, for example, it opens a possibility to defend the Method of Cases. In sum, my argumentation here could be considered in a similar way to revision-aimed-arguments; it might be counterintuitive to claim that Gettier intended to revise concepts. However, perhaps we should accept such an unintuitive interpretation if we wish for the reconstruction of his practice to meet some specific expectations.⁶