1 Introduction

Philosophy, at its most exciting, shows how the apparently obvious turns out to be bizarre or implausible on closer investigation. This variety of philosophy puts pressure on everyone to rethink their most confidently held beliefs. I will refer to the philosophers who undertake such a project as “revisionary philosophers”. For example, Unger (1980) and Merricks (2001) did revisionary philosophy when they challenged the common belief that there are macroscopic physical things like chairs and tables; and Descartes (1641) and Unger (1975) did revisionary philosophy when they challenged the common belief that we know that we have hands.Footnote 1 Not all philosophy is revisionary, in this sense; nor does it need to be. However, revisionary philosophy seems worth pursuing: we cannot simply assume that the ordinary way of thinking about the world is entirely adequate and thus in no need of revision.

In what follows, I will defend explication, in a Carnapian sense, as a strategy for revisionary philosophy—in particular, for revisionary ontology and radical scepticism. In other words, I will argue that these revisionary philosophers should explicitly commit to using expressions like “Fs exist” (or “There are Fs”) and “S knows that p” differently from how these expressions are used in everyday contexts.

I will begin, in Sect. 2, by motivating this commitment. With the help of examples, I will show that responding to everyday existence and knowledge claims with philosophical objections involves misinterpretation of the everyday claims. In the rest of the paper, I will address two worries that arise for revisionary philosophers, once they admit to departing from the everyday way of using language: the unintelligibility worry and the topic shift worry. The unintelligibility worry is the worry that if the philosophers do not use their crucial expressions the way these expressions are used in everyday contexts, then we cannot understand the philosophical counterparts to the everyday existence and knowledge claims. The topic shift worry is the worry that the subject matter shifts away from what really matters for revisionary philosophy, namely, what exists and what we know in the everyday sense of “exist” and “know”. In Sect. 3, I will address the unintelligibility worry, before focusing on the topic shift worry. As I further construe the topic shift worry, it is the concern that when philosophers use language differently from how it is used in everyday contexts, they cannot challenge or criticize the everyday existence and knowledge claims. And then everyone remains free to believe the content of those claims—which is an undesirable result for revisionary philosophy.

In order to respond to the topic shift worry, the revisionary philosophers who take the explication strategy need to explain how they are still able to criticize the everyday existence and knowledge claims, while they acknowledge that these claims are in some way correct, acceptable, or even true, by their lights. One way to put the attitude of revisionary philosophers who take the explication strategy, towards everyday existence and knowledge claims, would be as follows: “Yes, tables exist; but they don’t really exist”; or: “Sure, you know that smoking causes lung cancer; but you don’t really know that.” But how are we to make sense of these qualified affirmations of the everyday claims; in particular, what can the qualification (the criticism) consist in?

I will first put aside some unsatisfactory options in Sect. 4, building up to my prescriptivist analysis of the “Yes, but not really”, which I will defend in Sect. 5. The prescriptivist analysis is: “You can believe-1 the proposition that you expressed with S (e.g. “Fs exist”, “I know that p”), but you should not believe-2 that proposition.” By “belief-1”, I mean beliefs formed for practical purposes, such as survival and (non-epistemic) well-being. By “belief-2”, I mean beliefs formed for the sake of the intrinsically valuable epistemic achievement. The central idea is that revisionary philosophers who admit to departing from the everyday language can criticize everyday knowledge and existence claims by arguing that while the concepts or, more broadly, linguistic standards embedded in these claims are appropriate for beliefs-1, these concepts or standards are not appropriate for beliefs-2.

2 Why should revisionary philosophers acknowledge departing from the everyday language?

In this section, I will provide the initial motivation for explication as a strategy for revisionary philosophy. I will appeal to competent speakers’ judgment that the philosophers would misinterpret everyday existence and knowledge claims, were they to respond to such claims with their usual philosophical objections. On this basis, I suggest that there is a difference in how the claims of this form are used in the everyday contexts and in the relevant philosophical contexts. The revisionary philosophers need to acknowledge this difference. This acknowledgment, however, leads to the unintelligibility worry and the topic shift worry that will be addressed in the rest of the paper.

Let us first look at a hypothetical case where a claim of the form “Fs exist” or “There are Fs” is made in an everyday context and a philosopher objects with the kind of arguments that are typical in revisionary ontology.

John and the metaphysician. John, an intelligent man without a background philosophy, is hosting a garden party. He has philosopher friends, some of whom are at the party. John asks one: “Why are you always standing up? There are chairs” (or: “There are chairs in the world”; or: “Chairs exist”). The philosopher, a metaphysician who is indeed in the habit of standing often, seizes the opportunity for an interesting conversation and responds: “Actually, I don’t think that there are chairs, or that they exist. If there were chairs, they would be such that anything that only differs from a chair by an atom is also a chair. But in every situation where there would be at least one chair, there would be many millions of objects that only differ from this chair by one atom. So each of those would be a chair. So either there are no chairs at all or there are millions in each situation where there is at least one. I think the first option is more plausible.”Footnote 2 Or perhaps the philosopher would say that existing things should have independent causal powers and chairs do not; or he might give some other philosophical argument against ordinary objects. Assume that he elaborates on the argument well enough for the argument to have the appeal that it normally does in the context of a metaphysics discussion, in so far as the argument’s presentation is concerned.

We can expect, however, that the argument will not have the appeal, in this everyday context, that it normally has in the context of a metaphysics discussion. I predict that John and other competent speakers would get the sense that the metaphysician’s argument was a misplaced objection to what John expressed with the sentence “There are chairs” or “Chairs exist”. Before I explain how this suggests that revisionary philosophers need to acknowledge departing from the everyday way of using language, let us also look at a similar case with a knowledge claim.

John and the sceptic. At the same garden party, John lights up a cigarette and tells the woman standing next to him: “I know that smoking causes lung cancer, but I can’t resist.” The woman happens to be a sceptic – not about the health risks of smoking, but a philosophical sceptic of the radical variety that denies knowledge of hands possession. She objects: “Sorry, but you hardly know that smoking causes lung cancer. You don’t even know that you are not a brain in a vat that is fed perceptual seemings by a mad scientist. And since you don’t know this, you don’t know if the research that you base this claim about the effects of smoking on is real or dreamt up by that brain in the vat.”

Again, I predict that most competent speakers of English would find this to be a misplaced objection to John. Further, I take the best explanation for this sense of inappropriateness to be that the philosopher, in both scenarios, is misinterpreting what he or she is objecting to, namely “There are Fs” (or “Fs exist”) and “I know that p”. More precisely, the philosopher is interpreting these sentences in a contextually inappropriate way. The philosopher’s way of interpreting the sentences would be appropriate in the philosophy seminar, but not in this everyday context. So I am suggesting that our scenarios are similar to the following case, where a contextually inappropriate interpretation of “tall” gives rise to the sense that the response is misplaced.

The apple. Mary, John’s wife, cannot reach an apple on the tree, and asks John: “John, you are tall. Would you get that apple for me?” John, who has Dutch ancestors, replies: “No, I’m not tall: I’m below average height for a Dutch male.”

Since John is Dutch, it does make some sense for him to compare himself to the Dutch, when he is thinking about whether he is tall or not. However, this is not the relevant interpretation for “tall” in this context. Instead, the relevant interpretation is something like ‘tall enough to get this apple’.

Another example of a case that is similar to John’s conversations with the philosophers is the following one.

The empty bowl. Mary tells John: “The salad bowl is empty. Can you bring another one from the kitchen?” John replies: “It’s not empty: there are still a few bits of something in the bottom.”

Again, John misinterprets Mary, now invoking the strict sense of “empty” instead of the loose one that people have in mind in most everyday conversations. And again, we may expect the reaction that the response was misplaced. My suggestion is that the sense of inappropriateness, in the cases with John and the philosophers, is best explained similarly to the cases of “The apple” and “The empty bowl”.

However, one might wonder whether such linguistic variance is really the best explanation for the sense of inappropriateness, in the cases with John and philosophers. For example, Korman (2015, p. 63) considers and rejects an argument similar to the one that I have provided. His example is the following:

A: There are three cups on the table.

B: No, there are no cups. After all, [Insert argument for eliminativism].

Korman acknowledges that B’s response seems inappropriate, but he rejects the conclusion that “There are cups” means something different in ontology than it does in everyday conversation. He finds that a simpler explanation for the sense of inappropriateness is that bringing up the philosophical arguments, in the everyday conversation, “is disruptive: it does not advance, and it threatens to derail, the primary interests of the parties to the conversation” (Korman 2015, p. 63).

This explanation can be further understood along the lines of the “hermeneutic indifferentism” defended by Eklund (2005). The idea would be that the eliminativist’s response strikes us as inappropriate because the original speaker uttered “There are three cups on the table” indifferently to whether there are cups, trying to get at some other point. However, in the cases with John and the philosophers, John utters “There are chairs” (or “Chairs exist”; and he can even add that “There are chairs in the world” or “Chairs really exist”) and “I know that smoking causes lung cancer”. Does it make sense to assertively utter “There are chairs”, indifferently to whether there are chairs? The speaker may well be indifferent to a part of what he appears to be committed to, by uttering a sentence. This is plausibly the case in the sorts of examples made famous by Donnellan (1966): one can utter “The man drinking a martini looks happy” and “Smith’s murderer looks insane”, indifferently to whether the man is in fact drinking martini or whether the insane-looking man murdered Smith. However, in the cases with John and the philosophers, it is unclear how John could utter that “There are chairs” and “I know that smoking causes lung cancer” indifferently to whether there are chairs and whether he knows that smoking causes lung cancer. If he is indifferent about these things, then what other point is left over for him to convey?

Further, we could ask John whether he uttered “There are chairs” indifferently to whether there are chairs and “I know that smoking causes lung cancer” indifferently to whether he knew that smoking causes lung cancer. I predict that the answer would be negative; and there seems to be no good reason for supposing that speakers are confused about what they are indifferent to. So this explanation for the sense of inappropriateness does not look promising: John is not indifferent to whether there are chairs or whether he knows that smoking causes lung cancer.

One could also propose a different explanation: the sense of inappropriateness does not arise from the speaker’s indifference, but from the social norms governing the situation, as in the following example.

Little girl. At the same garden party, John and Mary’s daughter has been told by a bully that she is ugly. Mary consoles her daughter: “Beauty is in the eye of the beholder. Nobody is objectively beautiful or objectively ugly.” Another one of John’s philosopher friends overhears this and says: “Well, that is not quite right: actually, attractiveness judgment are quite uniform across humans and there is an evolutionary explanation for that…”

Here, the sense of inappropriateness does not arise because Mary is indifferent to whether beauty is in the eye of beholder and is trying to communicate something else. It rather arises because arguing the point is impolite in the situation, unnecessarily hurting the daughter’s feelings. Perhaps in the cases with John and the philosophers as well, the sense of inappropriateness arises from the fact that the philosophers are violating some social norm, such as the norm that one must avoid unnecessary argumentative confrontation at a party.

However, there is an asymmetry between the cases with John and the philosophers, on the one hand, and “Little girl”, on the other. Namely, in the former cases, it seems very plausible to say that the philosophers misinterpret John, whereas in the “Little girl” case, it is not at all plausible to say that the philosopher misinterprets Mary. The lesson that I take from this is that the relevant sense of inappropriateness to be explained, in the cases with John and the philosophers, is something more specific than just a sense of inappropriateness. The explanandum is the sense that the philosophers’ responses are inappropriate because they misinterpret John. And the simplest explanation now seems to be that the philosophers indeed misinterpret John.

3 The explication strategy as a response and the unintelligibility problem

Suppose that the revisionary philosophers cannot find a response to the above considerations, other than conceding that they do depart from the everyday way of using language when they philosophically discuss what exists and what we know. This would not mean the end of the road for revisionary philosophy. The philosophers, I will argue, can admit that in the relevant philosophical contexts, the expressions “S knows that p” and “There are Fs” (or “Fs exist”) are used differently from the everyday contexts. I will call such an admission the “explication strategy”. I will spend most of this paper addressing the main concerns that arise in connection with the explication strategy: the unintelligibility worry and the topic shift worry. Before this, however, I will clarify the sense in which the revisionary philosophers depart from the everyday way of using language, on my view. This clarification will also be relevant for addressing the unintelligibility worry.

When people ordinarily talk about “explication”, they often mean something like “explanation” or “analysis”. The relevant sense of “explication”, here, is close to Carnap’s use of the term (Carnap 1950a, 1963). According to Carnap, explication means “the transformation of an inexact, prescientific concept, the explicandum, into an exact concept, the explicatum” (Carnap 1950a, p. 1). I take “concept” to mean, roughly, the way in which we use an expression, in thought or in talk; and I take the core idea of the method of explication to be that science or philosophy (or scientific philosophy) may need to depart from the everyday way of using certain crucial expressions. Departing from the everyday language sometimes involves stipulating modified meanings, foreign to natural language, which are to be employed for the purposes of the inquiry. However, departing from the everyday language may also mean using the expression in a way that does not go beyond natural language, but still differs from the way that the expression is used in everyday contexts. For example, we might specify that the word “empty” is to be applied only to completely empty things in the context of our inquiry. We would then depart from the everyday way of using the word, yet we would still stick to the natural language word “empty”, and perhaps we would even use the word in its literal natural language sense. It is this latter variety of explication—explication that involves only departure from the everyday way of using language, but not departure from natural language—that I defend as a strategy for revisionary philosophy.

Let us now turn to the two worries that arise when philosophers depart from the everyday way of using language: the unintelligibility worry and the topic shift worry. The unintelligibility worry is the worry that we no longer understand what the philosophers are talking about, when they leave behind the everyday meanings of the crucial expressions. The topic shift worry is the worry that even if we still understand what the philosophers are talking about, they are no longer talking about what they are supposed to be talking about—since the revisionary philosophers are supposed to challenge beliefs about what there is and what we know in the everyday sense.

Let us begin with the unintelligibility worry. There is a notable tradition of doubting the intelligibility of the “metaphysical” existence claims, if these are to be understood as distinct from everyday existence claims. The tradition goes back to Carnap himself (1950b) and has recently been carried on by Thomasson (2015) and Hofweber (2016). Thomasson writes, for example, that “existence questions, in their normal sense and the only sense that they have, may be answered ‘easily’” (Thomasson 2015, p. 169). Hofweber (2016) describes the outlook of what he calls “esoteric” metaphysics as follows: “On the one hand, it is clear that there are numbers, and mathematics has shown it to be so. On the other hand, this proposal goes, there is the philosophical and metaphysical question of whether there really are numbers” (Hofweber 2016, p. 312). Hofweber rejects such an approach primarily because of the unintelligibility of the supposed philosophical issue of whether there are really numbers.

However, the unintelligibility worry seems to be largely based on the false assumption that departure from the everyday language is also departure from natural language. While there is good reason to believe that revisionary philosophers do depart from the everyday language (as was argued in the previous section), it is rather implausible that they depart from natural language. Otherwise, it would be difficult to explain how the ontologists’ and sceptics’ arguments make sense to students who have not been taught a new language. Students’ rather immediate appreciation of the arguments instead suggests that they quickly pick up on how the natural language is to be used in this new context.

I do not believe, then, that there is much disparity between how well we understand the everyday existence and knowledge claims and their philosophical counterparts. Both remain within the confines of the natural language that we are competent with. This is not to say that we understand either of these claims particularly well, especially when it comes to reflective understanding, i.e. the ability to provide correct and illuminating analyses of the claims, as opposed to the ability to competently use the claims in the appropriate circumstances. I will myself not give an analysis of either kind of existence or knowledge claims (the “everyday” kind or the “philosophical” kind). Instead, I will just highlight the central difference between them: the philosophical existence and knowledge claims are more demanding, in the sense of demanding more from the speaker. The speaker who makes claims of these forms in the relevant philosophical contexts must be ready to address questions and objections that would be misplaced in everyday contexts.

This is not to say that no objections are appropriate for the everyday claims. For example, in response to John’s utterance of “I know that smoking causes lung cancer”, the respondent might have appropriately brought up the possibility that the same gene causes both smoking and lung cancer. The respondent would not have misinterpreted John, in this case. Further, to John’s “You never sit; chairs exist”, the respondent might have objected, without misinterpreting John: “Actually, chairs don’t exist. I’ve tried to sit on some and they have always turned out to be an illusion. I’ve fallen and got badly hurt. I think other people are pretending when they seem to sit on chairs. It must be so uncomfortable.” This response would address the existence of chairs in the relevant sense. It might nevertheless invoke a sense of inappropriateness because the respondent seems to be insane. But it would not be plausible to say that the interlocutor misinterprets John’s claim that chairs exist.

In the relevant philosophical contexts, however, a wider range of questions and objections are appropriate and the speaker must be prepared to address them. In that way, the philosophical sense of the expressions is more demanding: it demands more from the speaker. This is consistent with how van Inwagen, for example, describes what is special about the “ontology room”: the discussants are not to say anything entailing that there are Fs, unless they are willing to answer various tricky questions about Fs. For example, anyone who says in the ontology room that there are paintings is “willing to answer any serious metaphysical question about the properties of paintings”, such as whether a painting could result from an unintended collision of molecules, whether a painting that is modified in a certain way is still the same painting, and so on (van Inwagen 2014, p. 3).

Combining this idea of higher demands on the speaker with Sider’s (2011) ideas, we may further conjecture that the aim of subjecting existence claims to such heightened scrutiny in the ontology room is to develop a way of speaking that reflects the world’s objective structure, or “carves nature at its joints”. The idea would be that a way of speaking that withstands such scrutiny would embody more of the relevant theoretical virtues, like simplicity, elegance, and coherence, than a way of speaking that does not withstand the scrutiny; and embodying the theoretical virtues in turn indicates that the way of speaking corresponds to the objective structure of the world (is joint-carving).Footnote 3

There seem to be no such well-established ideas on what the aim might be of subjecting knowledge claims to similarly heightened scrutiny in the epistemology room (for instance, requiring the speaker to address the brain-in-a-vat arguments). However, the aim here as well is probably the pursuit of some kind of epistemic excellence. The idea might be that a good epistemic agent should be particularly rigorous about testing her beliefs about what she knows, perhaps because this would facilitate a thorough questioning of the foundations of one’s belief system.

The aim of this section was to make progress with the unintelligibility worry that arises when revisionary philosophers admit to using expressions like “S knows that p” and “Fs exist” differently from how the expressions are used in everyday contexts. First, I noted that departing from the everyday language need not mean departing from natural language. Indeed, looking at how students can appreciate the ontologists’ and sceptics’ arguments without being introduced to a new language, it seems that revisionary philosophy does not depart from natural language. Then, I looked into the central difference between the everyday existence and knowledge claims and their philosophical counterparts: the appropriateness of a wider range of objections, in the philosophical case. I take this to be linguistic variance, rather than something like a social norm variance between the contexts. I speculated that the aim of the more demanding standard in the relevant philosophical contexts is the pursuit of epistemic excellence: limning the structure of reality, in the case of revisionary ontology, and especially thorough revision of one’s belief system, in the case of radical scepticism. I will now turn to the topic shift worry.

4 The “topic shift” worry; the “Yes, but not really” response; and some bad ways of analysing the “Yes, but not really”

Again, the topic shift worry is the worry that when philosophers depart from the everyday way of using their crucial expressions, they are no longer talking about what they are supposed to be talking about. There is a well-known historical precedent for this general worry regarding the method of explication: Strawson’s criticism of Carnap. Strawson criticized the method under the assumption that it would be applied to the analysis of ordinary concepts and found that “to offer formal explanations of key terms of scientific theories to one who seeks philosophical illumination of essential concepts of non-scientific discourse, is to do something utterly irrelevant” (Strawson 1963, p. 505). The aim of revisionary philosophers is not to analyse the ordinary concepts of knowledge or existence, so one cannot accuse them of changing the topic on those grounds. However, a similar worry still arises.

In so far as revisionary philosophers are supposed to be challenging the everyday existence and knowledge claims and belief in those claims, they seem to change the topic when they take the explication strategy. If the philosophers use the relevant expressions differently from how the expressions are used in everyday contexts, then it is unclear whether the philosophers can have any objection to the everyday existence and knowledge claims. Of course, there could still be other objections to the everyday claims to consider, for example, that a third factor might cause both smoking and lung cancer. But it is hard to see how there could be relevant objections forthcoming from revisionary philosophers. And if no objections to the everyday claims are forthcoming from the revisionary philosophers, then it seems that everyone remains free to believe the content of those everyday claims, by the revisionary philosophers’ lights; and thus their revisionary ambitions are frustrated.

One option for the revisionary philosophers, at this point, is to denounce the aim of challenging everyday existence and knowledge claims and belief in those claims—which is to give up on doing revisionary philosophy. Some ontologists have indeed been going in that direction (van Inwagen 2014; Horgan and Potrč 2008). The epistemic significance of these non-revisionary projects, however, remains mysterious. Others have vaguely suggested that they are still somehow challenging ordinary talk and thought from the ontology room: that although the everyday existence claims are in some shallow sense invulnerable to the philosophical objections, the ontology room crowd still inquires into what is really the case and has the final word. Yablo describes this sort of position on ontology as follows: “The goal of philosophical ontology is to determine what really exists. Leave out the ‘really’ and there’s no philosophy; the ordinary judgment that there exists a city called Chicago stands unopposed” (Yablo 1998, p. 258). But how can changing the topic to what really exists (or what we really know) make the ordinary judgment stand opposed? More insight is needed into the work done by the “really”.

The problem, then, is: the revisionary philosophers apparently need to admit that they have no objection to what people express when they assert that “There are Fs” (or “Fs exist”) and “S knows that p”, in everyday contexts. Presumably, the revisionary philosophers should then also have no objection to people believing the content of those assertions. However, if they want to remain revisionary philosophers, they need to repudiate or challenge such beliefs, somehow. It will not do to simply say that the content of those beliefs is not really the case: that while there are indeed Fs and Fs even exist, there aren’t really Fs and Fs do not really exist; and while we indeed know that smoking causes lung cancer, we don’t really know. We need an analysis of what this “really” means and how it puts revisionary philosophers in a position to challenge the everyday existence and knowledge claims and belief in those claims. In the rest of the paper, I will seek an answer to this question.

In this section, I will look at three unsatisfactory analyses of the revisionary philosophers’ “really”: (1) “really” as a degree modifier; (2) “really” in the sense of “literally” or “strictly”; and (3) a Siderian prescriptivist reading. Then, in Sect. 5, I will outline an alternative, also prescriptivist reading of the revisionary philosophers’ “Yes, but not really”: “You can believe-1 the proposition that you expressed with S (e.g. “There are chairs”), but you should not believe-2 that proposition.”

4.1 “Really” as a degree modifier

Let us first look at a fairly obviously unsuitable analysis of the “really”, to get a better handle on the kind of account needed. On this analysis, the “really” is a degree modifier. The idea is that existence and knowledge, like tallness and smartness, come in degrees; and revisionary philosophers are in the business of revealing that chairs have some existence, but not a lot of it (and in that sense, “Chairs exist, but they don’t really exist”), and John has some knowledge that smoking causes lung cancer, but not as much as he could possibly have (and in that sense, “John knows that smoking causes lung cancer, but he doesn’t really know that”). Neither the idea that existence and knowledge come in degrees, nor the further idea that revisionary philosophers discuss what we know or what exists to an especially high degree, is particularly popular; however, for revisionary ontology, such an idea has been proposed by McDaniel (2013).Footnote 4

We may leave aside the question of how plausible this is as a semantic account of “know” and “exist”: it is in any case not the needed explanation of the revisionary philosophers’ “really”. We need an account that explains how the philosophers can repudiate what John expresses when he utters the sentences “There are chairs” and “I know that smoking causes lung cancer”. But the degree modifier account of the “really” does not explain that. For example, suppose that John’s wife Mary is conversing with Lucy, who has never seen John. Lucy says: “I heard that John is tall”; and Mary responds: “Yes, he’s tall, but he’s not really tall”. Mary is using “really” as a degree modifier; and she is not in any way qualifying her affirmation of what Lucy says. In other words, she is not challenging, criticizing, or repudiating Lucy’s claim, but merely adding more information about John. So this is not the analysis of the “Yes, but not really” that we are looking for: the analysis that would illuminate how the revisionary philosophers are able to criticize the everyday existence and knowledge claims.

4.2 “Really” in the sense of “literally” or “strictly”

Sometimes we use sentences of the form “p, but not really p” to say that something is true only if interpreted metaphorically, but not if interpreted literally. Yablo suggests this analysis of “really” for ontology, saying that “’really’ is a device for shrugging off pretences” (Yablo 1998, p. 258). So the revisionary philosophers might explain to John what their “(not) really” means, as follows: “Well, it’s like that time when you left your first wife and told her not to worry: there are plenty of fish in the sea. Your wife, of course, understood that you meant that there are plenty of nice fellows around. But since she was particularly passionate about the environment, she found it necessary to point out that there aren’t really (literally) plenty of fish in the sea; that the fish population is actually declining at an alarming rate, due to unsustainable fishing practices.”

An initial problem with this analysis is that metaphorical use of language is generally recognized as such by the speakers, on reflection; but there is no sense of pretence or figurativeness in the uses of “Fs exist” and “I know that p” in our everyday examples. Yablo voices this concern: “I’m not sure what it would be to take ‘there is a city of Chicago’ more literally than I already do” (Yablo 1998, p. 259). This worry can be addressed, however, by replacing “literally” with “strictly” and contrasting it with “loosely speaking”, instead of “metaphorically speaking”. So the philosophers could rather explain the matter like this: “Remember when you and Mary bought your first flat screen TV and Mary said, in awe: ‘Wow, that is a flat screen.’ You responded: ‘Well, yes, but not really: nothing is flat, strictly speaking.’” There is no suggestion here that we engage in pretence when we normally say about things that they are flat.

This is how Unger (1975) treats everyday knowledge claims in his defence of scepticism: as only loosely true. According to an account along these lines, knowledge and existence do not come in degrees; they are absolute properties. People either know that p or not and things either exist or they do not. Further, when we ascribe knowledge and existence in everyday contexts, we use sentences like “Fs exist” and “S knows that p” loosely. What we actually express is that Fs are close enough to existing, for current purposes; or that S is close enough to knowing that p, for current purposes.

As a semantic theory of “know” and “exist”, this is more plausible for the former than the latter; but even in the case of “know”, contextualism is a strong competitor to such literalism. What matters here, however, is not whether this is a good semantic theory of “know” and “exist”. The problem is instead that, in any case, this proposal does not address the question that we are asking. The proposal does not provide the needed analysis of the “Yes, but not really”, because it does not show how the philosophers can reasonably repudiate what John expresses when he says that “Chairs exist” and “I know that smoking causes lung cancer”. It might seem like it is a repudiation of a sentence when we say that it is only true on a loose interpretation. However, what needs to be repudiated is the proposition that John expressed with the sentence, not the sentence itself. If there is a sense in which the proposition expressed is affected by the criticism here, then it is not clear to me what that sense is.

So these familiar ways in which the “Yes, but not really” response is used in ordinary discourse cannot do the job, or at least it is not clear how they could do the job. (The job, again, is explaining how the philosophers remain able to challenge the everyday existence and knowledge claims and belief in those claims.) The analysis will have to get more creative. To finish the section, I will discuss and reject the proposal that the “Yes, but not really” means “p is true, but you should not believe that p”. In the next section, I will outline a more promising proposal that builds on this one.

4.3 p is true, but you should not believe that p

In order to understand the revisionary philosophers’ qualified affirmation of the everyday knowledge and existence claims, we need to understand how the relevant propositions are affirmed and how that affirmation is qualified. According to the proposal considered next, the affirmation consists in acknowledging that the propositions that the speakers express when they utter the sentences “There are chairs” and “I know that smoking causes lung cancer”, in everyday contexts, are true. And the qualification—quite a qualification indeed—is that these propositions should nevertheless not be believed. The underlying idea is that truth is not enough for good beliefs: the beliefs (or more precisely, the propositions believed) also need to be cast in the right concepts. Beliefs, then, are couched in a certain language, and they can be criticized or commended on account of that language. This underlying idea is due to Sider (2011). Sider also has a view on what the right concepts are: the right concepts are those that “carve nature at its joints”, i.e. reflect the objective structure of reality. As Sider puts it, “beliefs aim to conform to the world” and “if belief and the world are both structured, belief aims not just at truth, but also at the right structure” (Sider 2011, p. 62).

On this account, then, the revisionary philosophers would be challenging the everyday existence and knowledge claims by repudiating the concepts, or more generally, the language embedded in these claims. I said above that this will be a more “creative” interpretation of the “Yes, but not really” response, suggesting that this is not a familiar way in which the phrase is used in ordinary discourse. However, we might sometimes use responses of the form “p, but not really p” like this in ordinary discourse as well. For seeing how this can work, it is helpful to draw on Plunkett and Sundell’s (2013) framework for analysing apparent object-level disputes as covert metalinguistic negotiations. For example, consider the following dispute.

John: “This painting of soup cans is not art: it has no emotional impact.”

Mary: “Yes, it is art. It is a clever attack on the consumerist society.”

Although this appears to be an object-level dispute on whether the painting is art or not, it can be plausibly analysed as a metalinguistic negotiation about how the term “art” should be used. The parties make use of different criteria for “art”; yet, they are not speaking past one another, but negotiating about how this term, with an important social function, should be employed in talk and thought. Now, it is easy to imagine the following variant of this dispute.

John: “This painting of soup cans is not art: it has no emotional impact.”

Mary: “Yes, it is art. It is displayed in “art exhibitions”, it is referred to as “art” in encyclopaedias, and so on.”

John: “Well, in a sense, you are right that it is art. But it is not really art. More precisely, it is true that this painting is art, given how the word “art” is actually used. But this is not how the word “art” should be used. And if we did not use “art” in this way, we would not believe that this is art. So it is true that it is art, but you should not believe that it is art – because you should not use this concept of art.”

In the latter dispute, John and Mary agree that the painting counts as art in ordinary English. They may or may not disagree about how the word “art” should be used. In any case, regardless of whether there is a disagreement, there is a sense in which John repudiates Mary’s claim that the painting is art, when he says that it is not really art: John criticizes the way of using language embedded in Mary’s claim. If Mary herself does not endorse this way of using language, then John’s criticism is not a criticism of Mary; but it still counts, it seems, as a criticism of the proposition expressed by Mary.

We can, then, imagine “p, but not really p” used like this in ordinary discourse, to mean that something is true, given the language being used, but this language should not be used and so the proposition cast in this language should not be believed. In contrast to the previous cases, there is repudiation of the relevant proposition in this use of “really”, although this repudiation is not about the truth of the proposition, but the way of using language embedded in the proposition. (This all assumes certain things about beliefs and propositions, but I suppose that these assumptions are uncontroversial enough. For example, this assumes that propositions are something like concepts combined in a certain way, rather than the things in the world that the sentences are about; and that beliefs are attitudes towards propositions.)

Sider’s point that truth is not enough for good beliefs, that good beliefs should also be true in a good language, is plausible and has been picked up by others as well. For example, Plunkett acknowledges that we should take note of the “different dimensions of what makes thought successful in the first place: namely, that one dimension of success involves truth or correctness, and another involves using the right concepts” (Plunkett 2015, p. 866). However, Sider himself at least sometimes appears to regard as relevant only one desideratum for the right concepts, namely the desideratum of joint-carving. This is why revisionary philosophers cannot take his account over without modification.

Sider’s view seems to be that all of our beliefs aim to conform to the world and therefore should ideally be cast in concepts that reflect the structure of the world, i.e. in joint-carving concepts. It is unclear how serious Sider is about this claim. He sometimes only suggests that in the context of inquiry, joint-carvingness is to be pursued. Perhaps not all contexts of forming beliefs (where “forming” beliefs includes casting them in certain concepts) are contexts of inquiry. In any case, regardless of whether Sider really thinks that all beliefs should be cast in joint-carving concepts, because all beliefs have the built-in aim of mirroring the world, let us get clearer on this idea and what is wrong with it.

The idea, then, would be that the rigorous questioning of claims like “There are chairs”, in the ontology room, reveals paradoxes and tough problems, such as the problem of the many (Unger 1980) and the special composition question (van Inwagen 1990). Siderian ontologists would draw the conclusion that the ordinary conceptual scheme is not theoretically virtuous and therefore it is presumably not joint-carving. Further, those who take seriously Sider’s suggestion that all beliefs ought to be cast in joint-carving concepts (whether Sider himself takes it seriously or not) would recommend not framing beliefs in a language that precludes the rigorous questioning characteristic of the ontology room and hence allows the inconsistencies and other theoretical shortcomings revealed in the course of that questioning. Now, what is the problem with this suggestion?

Briefly, the problem is that we do not just form and express beliefs in order to mirror the world and thereby flourish as epistemic agents. We also, and perhaps even mostly, form and express beliefs about the world—we try to represent the world accurately—in order to obtain practical aims, such as survival and (not purely epistemic) well-being. Further, for such practically oriented beliefs and the concepts they are cast in, the rigorous questioning characteristic of the ontology room and the joint-carvingness that we may expect to come along with it are not at all obvious desiderata. Suppose that revisionary ontologists indeed reveal that the ordinary conceptual scheme is lacking in theoretical virtue and therefore also in joint-carvingness. Even so, this theoretically problematic conceptual scheme might well be unproblematic and even advantageous when it comes to forming practically oriented beliefs. For example, the (theoretically) “bad” concepts might serve cognitive processing efficiency: forming representations of our environment quickly and reacting appropriately.Footnote 5 This is not to say that joint-carvingness is entirely unnecessary for practical beliefs. Concepts with highly disjunctive application conditions (e.g. ‘horse or quark or triangle’), for example, might be unconducive to cognitive processing as well as inferior in joint-carvingness. However, the proposal that we should pursue perfect joint-carvingness, without regard to any other desiderata (such as cognitive processing efficiency) in all belief formation seems to be based on an impractical epistemic ideal.

A similar proposal may be made to explain the sceptic’s qualified affirmation of everyday knowledge claims; and similarly, this proposal fails because of the impractical epistemic ideal that underlies it. Plunkett and Sundell in fact suggest that their analysis could be applied to cases of cross-contextual disagreement about knowledge claims, if contextualism about “know” is true: “the speaker in the high-standards context faults the speaker in the low-standards context, not for literally expressing a false proposition, but for employing overly lax epistemic standards” (Plunkett and Sundell 2013, p. 28). Applying this to our example: the sceptic would be faulting John for employing overly lax epistemic standards when John says, and believes, that he knows that smoking causes lung cancer; at the same time, the sceptic would be acknowledging that John’s self-ascription of knowledge is true, given the lax standards he is employing.

What are the standards that the sceptic is recommending, then? Apparently, the idea is that we should use expressions like “S knows that p” in the sceptic’s way, whenever we form and express beliefs about what someone knows. That is, we should never say or believe that we know something, unless we are willing to respond to arguments like the brain-in-a-vat argument. The rationale for the universal application of this high standard might be that the comparatively lax standards come packed with a set of problematic attitudes and dispositions, such as settling on hypotheses too easily and not asking enough in the way of justification for one’s beliefs—where not asking enough means not subjecting one’s beliefs to the kind of radical questioning characteristic of the sceptic’s context. An ideal epistemic agent, one might think, should not have such intellectually lazy dispositions.

At first, this seems like a plausible account of what the sceptic might mean with the “Yes, but not really”, namely: “What you express with ‘I know that smoking causes lung cancer’ is true, but you should not believe this, because you should not employ such lax standards for knowledge.” And just as the previously considered epistemic ideal—that beliefs should conform perfectly to reality—made some sense, the epistemic ideal underlying this analysis makes some sense as well. However, implementing the sceptic’s standards for all knowledge ascriptions would be highly impractical. This is because of the role of knowledge ascriptions in everyday thought and talk.

First, there is plausibly a strong connection between the notions of knowledge and rational action (Hawthorne and Stanley 2008, Fantl and McGrath 2002). Hawthorne and Stanley point to examples from everyday discourse, such as “You shouldn’t have gone down this street, since you did not know that the restaurant was here” (Hawthorne and Stanley 2008, p. 571). They propose that “[w]here one’s choice is p-dependent, it is appropriate to treat the proposition that p as a reason for acting iff you know that p” (Hawthorne and Stanley 2008, p. 578). And they suggest that such a connection with action might be “one of the most fundamental roles for knowledge” (Hawthorne and Stanley 2008, p. 574). Especially the part of the principle that establishes knowing that p as a necessary condition for treating p as a reason to act is relevant here. Granted, this necessary condition seems dubious, as it stands: it seems that we may have to and often do treat p as a reason to act, although we do not know, but only suspect or tend to believe that p.Footnote 6 However, a more relaxed version of the principle remains plausible, namely that one should generally strive to obtain knowledge that p before treating p as a reason for action; or perhaps that one is prima facie subject to criticism when one has treated p as a reason for acting without knowing that p.

Suppose that something like this more relaxed principle does capture an important role for everyday knowledge ascriptions. And suppose we followed the sceptic’s advice never to say or believe that we know something, unless we can rule out the brain-in-a-vat hypothesis and similar scenarios. Leaving the described association of knowledge with action in place, we should then generally strive to rule out the brain-in-a-vat hypothesis, in order to establish that we know that smoking causes lung cancer and thus be permitted to treat the proposition that smoking causes lung cancer as a reason to quit smoking. Further, if we have imprisoned someone because we think that they are guilty, without having ruled out the brain-in-a-vat hypothesis, then we are prima facie subject to criticism for treating the proposition that the person is guilty as a reason for acting, without ruling out the brain-in-a-vat hypothesis. These examples suggest that implementing the sceptic’s standard for knowledge across the board would make the concept perform one of its central functions worse in everyday contexts.

Further, other functions of everyday knowledge claims are also threatened by overly demanding standards. For example, we might ascribe knowledge to indicate someone’s status as a good informant on a matter, e.g. in “Geoff will know whether the 1977 BMW R1100 comes standard with disk breaks; he knows everything about that bike” (Chrisman 2007, p. 242). It seems that we will not be able to identify reliable informants via knowledge ascriptions, if we first need to deal with brains in vats and similar scenarios, to figure out whether anyone knows anything about anything.

In sum, the problem with this analysis of the “Yes, but not really”, as “p is true, but you should not believe that p”, is that impractical epistemic ideals underlie both the ontologist’s and the sceptic’s versions of it. These ideals may seem reasonable when we form and express beliefs in the comfort of the metaphysics and epistemology rooms, but the standards motivated by these ideals will have undesired consequences in the everyday contexts of forming and expressing beliefs in pursuit of practical goals. In the next and final section, I will outline my proposed modification to the proposal just considered. I will keep the prescriptivist element, but do away with the impractical epistemic ideals. I suggest that we should distinguish between practically oriented beliefs-1 and theoretically oriented beliefs-2 and analyse the “Yes, but not really” as follows: “You can believe-1 the proposition that you expressed with S (e.g. “Chairs exist” or “I know that smoking causes lung cancer”), but you should not believe-2 that proposition.”

5 The recommended prescriptivist analysis

I have argued that the epistemic ideals underlying the previous proposal and the linguistic choices motivated by such ideals are unreasonable, when it comes to forming beliefs for practical purposes. However, we do not always form beliefs for such practical purposes; and when we form beliefs for theoretical purposes, then the ontologist’s and sceptic’s epistemic ideals and linguistic choices seem quite sensible. How should we modify the prescriptivist analysis of the “Yes, but not really”, in the light of this recognition?

I suggest distinguishing between different belief-like attitudes, which I will call, as neutrally as possible, belief-1 and belief-2. Beliefs-1 are formed to serve the non-cognitive aims of the belief-former; beliefs-2 are formed particularly for the sake of achieving epistemic excellence. Given this distinction, “Yes, but not really”, as the revisionary philosopher’s response to an everyday utterance of S (e.g. “Chairs exist”, “I know that smoking causes lung cancer”), should be understood as follows: “You can believe-1 the proposition that you expressed with S, but you should not believe-2 that proposition.” The speaker is thus permitted to form beliefs-1 that presume the everyday linguistic standards for “exist” or “know”, but is advised against forming beliefs-2 that presume these standards. The criticism of the everyday existence and knowledge claims that remains available for the revisionary philosophers, on this account, is that these claims are couched in a language that should be avoided when forming beliefs-2. In the rest of the section, I will flesh out the proposal a bit more and defend it against objections.

First, one might feel uneasy about the distinction among beliefs that this proposal assumes. The distinction between the two roles of belief is appealing enough, but it is not so clear whether there are kinds of belief corresponding to these roles. One may insist that we have just one set of beliefs that we employ for both theoretical and practical purposes. Now, as I have suggested in the previous section, these theoretical and practical purposes may clash when it comes to choosing the best concepts (or linguistic standards) for belief formation. Theoretically “bad” concepts, such as inconsistent or disjunctive concepts or concepts with incomplete application conditions, might come in handy when we represent the world for practical purposes. If we indeed had just one set of beliefs for both roles, then we would seem to be required to make compromises between the conflicting theoretical and practical desiderata when we develop the language that we use to form those beliefs. However, we do not seem to be forced to make such compromises. Instead, we seem to be free to keep our theoretical beliefs distinct from our practical beliefs and to engage in two relatively independent projects of linguistic engineering.

A further, minor consideration in favour of the distinction is that there is a significant difference between the action dispositions associated with the two kinds of beliefs. For the theoretical beliefs, the action dispositions are limited to a special class of what might be characterized as “research actions”: assertions (mostly in professional contexts), downloading papers, reading papers, going to conferences, etc. The action dispositions associated with the practical beliefs are much more varied.Footnote 7 In connection with the last point, it is worth emphasizing that the distinction between practically and theoretically oriented beliefs is not that the former are used in practical reasoning about actions and the latter are not. As just noted, the theoretically oriented beliefs come with dispositions to undertake a number of research actions. Instead, the basis of the distinction is that the theoretically oriented beliefs are ultimately formed for theoretical purposes (like achieving intrinsically worthwhile epistemic excellence), while the practically oriented beliefs are ultimately formed for practical purposes (like survival and non-epistemic well-being).

However, one might still think that the distinction is not as sharp as I am suggesting. In particular, one may doubt whether for every project of belief formation or every individual belief, it is determinate whether the belief forming in question is ultimately done for practical purposes or for theoretical ones. For example, AI research seems to be a single project of belief formation that aims both at theoretical understanding of the human mind and at developing technologies for serving non-cognitive human aims. Also, normative ethics may be conceived as both practically and theoretically oriented. My response is that the position outlined here does not require a clean division of all beliefs into the practically oriented and the theoretically oriented; nor does it require that for any area of human activity, one can say whether practical or theoretical beliefs are formed in that area. My proposal only concerns the beliefs that are formed and expressed in the relevant everyday contexts and those that are formed and expressed in the relevant philosophical contexts (in the relevant ontology and epistemology rooms). And regarding these two kinds of contexts, the distinction seems to be just as sharp as I have suggested.

Now that I have elaborated on the belief-1/belief-2 distinction, how does this distinction help us make sense of the revisionary philosophers’ qualified affirmation of the everyday existence and knowledge claims? I have already mentioned that the relevant criteria for linguistic choices are somewhat different for beliefs-1 and beliefs-2. The impractical epistemic ideals considered in the previous section, the ideals underlying the revisionary philosophers’ linguistic choices, make much more sense in the context of forming beliefs-2 than in the context of forming beliefs-1. For example, the ideal that beliefs should conform perfectly to the world makes good sense for beliefs-2. Consequently, the derivative ideals of a perfectly joint-carving conceptual scheme and a perfectly theoretically virtuous—simple, coherent, elegant—conceptual scheme also make good sense when we take this conceptual scheme to be meant for forming beliefs-2. Likewise, the sceptic’s impractical epistemic ideal—of always employing the high standard for knowledge, such that the sceptic’s objections to knowledge claims are appropriate—makes sense in the context of forming beliefs-2. The sceptic’s way of employing knowledge ascriptions is not the only thinkable way, in the context of forming beliefs-2; but it is a sensible option.

In this light, it would make good sense for the revisionary philosophers to maintain that they are targeting beliefs-2 and not beliefs-1. Then, they could explain their “Yes, but not really” to John as follows: John is entitled to a certain belief-like attitude (belief-1) towards the proposition that he expresses with “Chairs exist” or “I know that smoking causes lung cancer” (or at least the philosophers themselves have no objection to him having this attitude); but he should withhold another belief-like attitude (belief-2) towards the same proposition.

Now, how does this proposal satisfy the central desideratum for the analysis of the “Yes, but not really”: how does it explain the sense in which the revisionary philosophers who take the explication strategy can still repudiate everyday knowledge and existence claims? After all, when people make everyday knowledge and existence claims, they express their beliefs-1, not their beliefs-2—so how can the philosophers repudiate what they express by saying that it is not suitable for belief-2? Indeed, I am assuming that when the revisionary philosophers argue that the language that John employs when he makes his everyday existence and knowledge claims is not suitable for belief-2, then this counts as a criticism of the propositions he expresses, even if John does not believe-2 the propositions. The philosophers would not be criticizing John, but they would be criticizing the propositions that he expresses.

For an analogy, suppose that in the context of pursuing epistemic excellence for its own sake, we should only use “empty” for absolutely empty containers (perhaps because this is a more joint-carving use of the term). Then, when Mary says in an everyday context that “The bowl is empty”, a philosopher could respond: “Yes, but it is not really empty”. He would be criticizing the proposition that Mary expressed, by insisting that the linguistic standard employed therein is not suitable for forming beliefs in pursuit of epistemic excellence for its own sake. This is not a criticism of Mary herself (who might not be at fault in any way), but a critical observation about the linguistic standards at work in the construction of the proposition: these standards are only suitable for forming beliefs-1, but not for beliefs-2. The role of revisionary philosophers, accordingly, would not be faulting the folk or even faulting the beliefs that they hold, but flagging that the language that their claims are couched in has a limited area of application, namely beliefs-1; and good beliefs-2 should be framed using different linguistic standards. This is a subtle, mild kind of repudiation, but I think it can still be called repudiation, and that this helps understand revisionary philosophers’ ability to challenge everyday existence and knowledge claims and belief in those claims.Footnote 8

6 Conclusion

I defended explication as a strategy for revisionary philosophy, arguing that revisionary ontologists and radical sceptics should admit to departing from the everyday way of using expressions like “Fs exist” and “S knows that p”. They should make this admission, unless they can explain away the sense of misinterpretation that arises when we imagine revisionary philosophers’ arguments being posed as objections to everyday knowledge and existence claims. I addressed two main worries with the explication strategy, namely the unintelligibility worry and the topic shift worry, focusing on the latter.

We can make progress with the unintelligibility worry by recognizing that the revisionary philosophers may still plausibly use natural language, while they do not use everyday language. Further, I suggested that the central distinction between the linguistic standards in the everyday contexts and the relevant philosophical contexts, for both sceptics and ontologists, is that the philosophical contexts allow posing questions and objections that would be inappropriate in the everyday contexts. I also suggested that in both cases, the aim of this more demanding standard in the philosophical context is the pursuit of epistemic excellence. In the case of revisionary ontology, the pursuit of epistemic excellence takes the form of pursuing a theoretically virtuous and therefore presumably joint-carving language. In the case of radical scepticism, the pursuit of epistemic excellence means fostering virtuous intellectual habits, such as examining the foundations of one’s belief system.

I specified the topic shift worry as the need to explain how the revisionary philosophers can still challenge the everyday existence and knowledge claims and belief in those claims, when they take the explication strategy. The revisionary philosophers would want to say something like: “Yes, chairs exist, but they don’t really exist” or “Sure, you know that smoking causes lung cancer, but you don’t really know that”. The analysis that I eventually defended was that the “Yes, but not really”, as the revisionary philosophers’ response to the everyday claims, should be understood as follows: “You may believe-1 the proposition that you expressed with S (e.g. “Chairs exist”), but you should not believe-2 that proposition”. Belief-1 and belief-2 have different aims—roughly, practical and theoretical—and therefore different linguistic standards are suitable for framing beliefs-1 and beliefs-2. For example, while joint-carving concepts might be suitable for framing beliefs-2 (because beliefs-2 should ideally conform exactly to the world, under a plausible, although not indisputable view), the desideratum of joint-carvingness might sometimes be trumped by the need for cognitive processing efficiency, in the case of beliefs-1. Further, while an especially demanding standard for knowledge ascriptions might be suitable for framing beliefs-2 (to encourage the rigorous epistemic habits proper to that context), such a standard could jeopardize the role of knowledge ascriptions in the context of framing beliefs-1.

The revisionary philosophers’ repudiation of everyday existence and knowledge claims, on this view, consists of noting that the linguistic standards embedded in the claims are unsuitable for beliefs-2. This is not a criticism of the speaker (who is not expressing and may not even have the relevant beliefs-2), but a subtle kind of criticism of the proposition expressed. Given the subtlety of this remaining criticism, the strategy of explication, thus fleshed out, might admittedly call for some rethinking of the revisionary ambitions of the revisionary philosophers.

Finally, if this is after all not the best way for revisionary philosophy to proceed, I hope to have pointed to an independently important issue concerning explication, or conceptual engineering, or prescribing linguistic standards. Namely, in so far as we think of a good language as a means to good beliefs, the criteria for a good language will turn on the criteria for good beliefs. Further, it seems that we form beliefs for different purposes: to achieve non-cognitive aims and for the sake of intrinsically valuable epistemic achievement. Investigation into beliefs—what they are and what they are supposed to do—is essential for getting a better handle on the desiderata for linguistic choices.