Moderatism and Truth

Abstract

According to moderatism, perceptual justification requires that one independently takes for granted propositional hinges. This view faces the truth problem: to offer an account of truth for hinges that is not threatened by skepticism. Annalisa Coliva has tried to solve the truth problem by developing a new form of alethic pluralism. I argue that the resulting view cannot offer a coherent characterization of “skeptical switch scenarios” while providing an effective anti-skeptical strategy. In a more positive vein, I defend an approach that combines a correspondence conception of truth with epistemological disjunctivism.

Moderatism belongs to a family of “hinge epistemologies” that originate from Wittgenstein’s On Certainty (OC) (Wittgenstein, 1969). They hold that our epistemic practices take place within a framework of “hinges” that lie beyond any rational doubt. The idea came to light with a fruitful analogy: “If we want the door to turn”—writes Wittgenstein—, “the hinges must stay put” (OC 343). Similarly, for our epistemic practices to be possible, some hinges must stay put. According to moderatism, hinges are not and cannot be epistemically justified. ¹ Yet, moderatism analyzes hinges as propositions. ² Given that propositions have truth conditions, moderatism faces the truth problem: to offer an account of truth for hinges that is not threatened by skepticism.

Annalisa Coliva (2015, 2018, 2021, 2022) has offered an intriguing attempt at solving the truth problem. Her strategy is to combine the claim that ordinary external world propositions have a substantive truth property like correspondence with the claim that hinges only have a deflationary truth property. The resulting view constitutes a novel form of alethic pluralism that is meant to provide the materials for a successful anti-skeptical strategy. ³

There has been some discussion about the tenability of Coliva’s alethic pluralism (Coliva, 2021; Moruzzi, 2021). In this paper, I tackle a different issue. Assuming that alethic pluralism is available, does its conjunction with moderatism offer a successful anti-skeptical strategy? I will offer a negative answer to that question. More specifically, I will argue that the conjunction moderatism and alethic pluralism cannot offer a coherent characterization of what I will dub “skeptical switch scenarios” while providing an effective anti-skeptical strategy. In a more positive vein, I will defend an alternative picture. First, I will suggest that the propositions that Coliva identifies as hinges have a substantive truth property that can be derived from the correspondence truth property of ordinary external world propositions. Second, I will submit that this view can be fruitfully combined with epistemological disjunctivism.

The article has five sections. Section 1 spells out moderatism. Section 2 presents the truth problem and introduces Coliva’s solution. Section 3 presents my argument against Coliva’s view. Section 4 discusses two possible replies. Section 5 presents an alternative view and defends it from objections.

1. Moderatism

Coliva characterizes moderatism as follows:

Moderatism. Absent defeaters, a subject, S, is justified in believing p if and only if S has an appropriate course of experience (typically, an experience as of p) and S independently takes several hinges for granted (Coliva, 2015, 8–11, 33–34, 117, 2022, 59).

Moderatism is consistent with several hinge epistemologies. However, Coliva incurs additional commitments. Hereafter, I use “moderatism” to denote the account that results from adding these commitments:

Propoasitionality. Hinges are propositions.

Constitutivism. Hinges are constitutive of human epistemic rationality.

Generality. Only very general propositions qualify as hinges. They include <There is an external world>, <My sense organs work mostly reliably>, and <I am not a brain in a vat (BIV)>. ⁴

No justification. Hinges are not and cannot be epistemically justified.

Propositionality opposes moderatism to nonpropositional views of hinges (Moyal-Sharrock, 2004). A prominent argument for nonpropositional views invokes bipolarity. For the early Wittgenstein (1921, 3.144, 3.221, 6.111–6.126), it is part of the essence of a proposition to be bipolar, that is, to be capable of being true and being false. Hinges are not bipolar because they play a normative role, and—so the argument goes—norms are not capable of being true and being false. Therefore, hinges are not propositions. ⁵

Coliva grants that hinges are not bipolar but insists that propositionality is consistent with failures of bipolarity. She offers exegetical and independent arguments for her claim. In her view, by the time Wittgenstein wrote On Certainty, he used the word “proposition” as expressing a family resemblance concept that included both (bipolar) empirical propositions and hinges (OC 51; Coliva, 2022, xxix, 23–24). Moreover, Coliva reads On Certainty as attributing both descriptive and normative functions to hinges. Hinges are like certain pictures that guide us in assembling a piece of furniture: they have descriptive content and tell us how we ought to assemble it (Coliva, 2022, 9). Finally, Coliva claims that failures of bipolarity are consistent with the possession of a truth property (Coliva, 2022, 110). Hinges can work as paradigms of truth (Coliva, 2022, 22). So, they can figure as the contents of assumptions and play various roles that require the possession of truth-conditions, like being meaningfully negated and figuring in the antecedents of suppositional arguments (Coliva, 2022, xxviii).

Coliva’s arguments are not decisive, though. Many proponents of nonpropositional readings have argued that we should not take all linguistic appearances at face value. It can seem that a hinge is playing propositional roles when something else is playing those roles. Perhaps the actual hinge has a doppelgänger with propositional form. Or, less controversially, a hinge can lose its hinge role in some contexts. When that happens, the hinge has features of propositions (Moyal-Sharrock, 2004).

Coliva has several replies to these moves. First, the doppelgänger view cannot accommodate Wittgenstein’s contention that, depending on the circumstances, the same proposition can be treated as a rule of testing and as something to be tested (OC 96–99; Coliva, 2022, 34). Second, any broadly contextualist view of hinges faces the problem of “semantic ignorance” that afflicts traditional forms of epistemic contextualism (Coliva, 2022, 34). The problem is that contextualist views introduce forms of context-dependence that competent speakers are unaware of. Third, Coliva suggests that we should develop an anti-skeptical strategy that is not hostage to a controversial theory of meaning (Coliva, 2002, 17, 97; see also Pritchard, 2016; Wright, 2004). Hence, Coliva defends a form of invariantism about hinges according to which they lack doppelgängers and do not vary from context to context.

Constitutivism is the second tenet of moderatism. Commenting on the hinge analogy, Wittgenstein writes:

[I]t belongs to the logic of our scientific investigations that certain things are in deed not doubted.

But it isn’t that the situation is like this: We just can’t investigate everything, and for that reason we are forced to rest content with assumption. If I want the door to turn, the hinges must stay put (OC 342–343).

Hinge epistemologists generalize these remarks from scientific investigations to any human epistemic practice (Wright, 1985, 2004). They also transform Wittgenstein’s talk of “logic” into the claim that hinges are constitutive of any human epistemic practice. A weak reading holds that any epistemic practice requires that some hinges stay put. This view is consistent with the claim that different contexts require different hinges. A strong reading holds that any epistemic practice requires that a select group of hinges stays put across contexts. There are also attempts at combining both types of hinges into a unified framework. ⁶

Coliva’s constitutivism is a development of the strong reading. Coliva grants that our reliance on a select group of hinges is a contingent matter. However, that select group of hinges is necessary to engage in human epistemic practices. More radically, Coliva sees a select group of hinges as part of an analysis of the concept of human epistemic rationality because they are necessary for humans to rationally form, revise, retain, and evaluate beliefs (Coliva, 2022, 17–19, 22, 36–37, 2023, 10). Their status is therefore analogous to the constitutive rules of a game. If one does not follow the constitutive rules of chess, one is not playing chess. Similarly, if a select group of hinges does not stay put, one is not engaging in a human rational-epistemic practice (Coliva, 2022, 12, 56; Coliva & Palmira, 2021, 407).

Coliva’s commitment to invariantism and constitutivism compels her to develop a hinge epistemology centered on a restricted number of very general hinges (Coliva, 2022, xi, xxiii, 96). After all, it is implausible to hold that particular hinges stay put across all contexts and are necessary for engaging in human rational-epistemic practices. Thus, Coliva departs from Wittgenstein and other hinge epistemologists who treat <I have hands> (OC 250) or <My name is NN> (OC 425) as hinges. Her argument is straightforward. If <There is an external world> or <I am not a BIV> did not stay put, “one could not even start investigating a mind-independent reality” (Coliva, 2018, 269). By contrast, one can start investigating a mind-independent reality without taking for granted propositions like <I have hands> or <My name is NN>. After a severe car accident, someone might wonder whether she has hands. An amnesic subject could start investigating who they are. More generally, if one can conceive of a human rational-epistemic practice as taking place while a given proposition does not stay put for the participants in that practice, then one has a reason not to treat that proposition as a hinge (Coliva, 2022, 5). ⁷

If one holds a nonpropositional view, it is a category mistake to think of hinges as justified or unjustified (Moyal-Sharrock, 2004; Wright, 1985). If one holds a propositional view, one must offer an argument for no justification, the fourth tenet of moderatism. Coliva’s defense of no justification relies on an argument by elimination (Coliva, 2015, 129, 238, 2022, ix–xv). Consider the hinge <There is an external world>. Given moderatism, an independent commitment to <There is an external world> is necessary to obtain empirical justification. So, if one were to justify <There is an external world> empirically, one’s justification would be circular (Coliva, 2022, 8, 37–38). It is also unclear how <There is an external world> could be a priori justified. Consider the claim that a priori justification is grounded in intuition. Without an account of this putative faculty of intuition, the proposed view remains mysterious (Coliva, 2022, xv). Or consider the claim that a priori justification is grounded in the inferential role of the relevant concepts. For Coliva, the resulting view “would immediately be hostage to the theory of meaning we are prepared to subscribe to” (Coliva, 2022, xv). Finally, one could hardly provide an epistemic justification for <There is an external world> by highlighting its pragmatic benefits (Wright, 2004). Coliva thinks—as many other philosophers do—that epistemic justification must speak to the truth of the target proposition (Coliva, 2022, xv, 37). These arguments generalize to other hinges. ^{8 , 9}

2. The Truth Problem and Alethic Pluralism

Although moderatism offers a promising version of hinge epistemology, it also faces the truth problem: to offer an account of truth for hinges that is not threatened by skepticism. Alas, it is hard to see how hinges could satisfy both desiderata.

Consider correspondence. On this view, a proposition p is true just in case p corresponds to a mind-independent reality (e.g., a fact that obtains prior to and independently of its being the subject matter of a representation) (Coliva, 2022, 98). ¹⁰ Nevertheless, correspondence leaves open the possibility that hinges do not correspond to a mind-independent reality. When it comes to ordinary propositions like <That is an apple>, there is an easy way of blocking this possibility. It suffices to avail oneself of evidence that speaks to the truth of <That is an apple>. However, no justification tells us that hinges are not and cannot be epistemically justified (Coliva, 2022, 101). If we want to use moderatism to close the door on skepticism, correspondence is not a promising view—at least in so far as hinges are concerned. As Coliva puts it:

[T]he most powerful antidote to any argument based on radically skeptical scenarios, which raise the possibility that, in ways that are totally unknowable to us, there is no external world (or we are BIVs, etc.), is to avoid thinking of truth as mind-independent. That is, as a correspondence between our representations and mind-independent facts, whose obtaining is in principle unknowable to us (Coliva, 2022, 65).

This line of reasoning suggests that the way out is anti-realism. For our current purposes, we can define anti-realism as the view according to which a proposition p is true just in case p is maximally justified (Wright, 1992, 42). Nevertheless, no justification makes this view unavailable (Coliva, 2022, xx). Therefore, anti-realism is not an option.

Suppose that truth can only be predicated of a proposition when there can be evidence in favor of that proposition (OC 199, 200, 205). Given that there cannot be evidence in favor of hinges, it follows that truth cannot be predicated of hinges. ¹¹

Coliva resists this pessimistic conclusion by suggesting that hinges can be true in a minimal sense. Consider the hinge <There is an external world>. For Coliva, sentences that express this hinge satisfy a collection of platitudes that sufficiently characterize their truth property. For example, [1] they satisfy the disquotational schema:

(T_DISQ) The proposition expressed by “There is an external world” is true if and only if there is an external world.

Moreover, they [2] can be meaningfully negated and [3] embedded in conditional statements (Coliva, 2022, xxi, 8–9, 17, 105). In previous work, Coliva added that predicating truth of hinges also consists in “[4] being prepared to act on their basis and even [5] to judge and assert them, thus being disposed to present them as true… Finally, and more importantly, to say that they are true is equivalent to [6] holding that what they state is how things are, given our overall Weltbild” (Coliva, 2015, 149; see also 37, 156–157, 201). ¹²

Coliva’s approach is inspired by Horwich’s (1982) characterization of alethic minimalism but, unlike Horwich, Coliva combines her own alethic minimalism with alethic pluralism (Wright, 1992). The intuition behind alethic pluralism is that not all sentences or propositions are true in exactly the same way (see also Lynch, 2009; Pedersen, 2012). For example, in certain areas of discourse, truth is exhausted by the minimalist characterization. A potential example is logical truths, which do not seem to be true by corresponding to a mind-independent reality. For Coliva, something similar happens with hinges. They are “plain truths” in Lynch’s terminology, for their truth is not grounded in anything else (Coliva, 2022, 106). However, Coliva is prepared to concede that, in other areas of discourse, there is a more substantive truth property that goes beyond the relevant platitudes. When it comes to ordinary external world propositions like <That is an apple>, Coliva thinks that correspondence offers a plausible characterization of their truth property.

For simplicity, let us use alethic pluralism to refer to the following combination of views: correspondence for ordinary propositions (ordinary correspondence for short) and minimalism for hinges (hinge minimalism for short). When propositions with different alethic properties participate in entailment relations, alethic closure holds:

Alethic closure. If <p>has an alethic property T₁ and <q> has an alethic property T₂, and <p>entails <q>, then either T₁ = T₂, or having a truth property is the manifestation of minimal truth. ¹³

By conjoining moderatism with alethic pluralism, Coliva can retain propositionality while preserving much of the spirit of nonpropositional views. Norms like “Stop at traffic lights when red” do not correspond to a mind-independent reality. If hinges do not correspond to a mind-independent reality, they can work like norms while being true in a minimal sense. This captures some of the spirit of Wittgenstein’s view, which holds that a hinge is true “only inasmuch as it is an unmoving foundation of [the] language-games” (OC 403) and not because it “tallies with the facts” (OC 199) (Coliva & Palmira, 2020, 14). Since Coliva had already granted that hinges fail the bipolarity test (Section 1), her view seems to offer the ingredients to close the door on skepticism: “If ‘There is an external world’ is relevantly similar to ‘Stop at traffic lights when red’ then the skeptical worry that, being unjustified, it might turn out to be false would be off target and due to a mistaken conception of the very nature of that ‘hinge’” (Coliva, 2022, xix).

In Section 3, I challenge Coliva’s solution to the truth problem. More specifically, I show that the conjunction moderatism and alethic pluralism faces a dilemma. If Coliva’s hinges remain true in what I will dub “skeptical switch scenarios,” Coliva cannot improve upon dogmatist and disjunctivist treatments of skepticism. If Coliva’s hinges turn false in those skeptical switch scenarios, Coliva cannot offer a coherent characterization of those scenarios.

3. The Skeptical Switch Argument

Many anti-skeptics have tried to use externalist theories of semantic content to deal with radical skepticism (Davidson, 1986; Putnam, 1981). Some of Wittgenstein’s remarks hint at a strong version of this strategy:

The argument “I may be dreaming” is senseless for this reason: if I am dreaming, this remark is being dreamed as well—and indeed it is also being dreamed that these words have any meaning (OC 383).

But even if in such cases I can’t be mistaken, isn’t it possible that I am drugged? If I am and if the drug has taken away my consciousness, then I am not now really talking and thinking. I cannot seriously suppose that I am at this moment dreaming. Someone who, dreaming, says “I am dreaming,” even if he speaks audibly in doing so, is no more right than if he said in his dream “it is raining,” while it was in fact raining. Even if his dream were actually connected with the noise of the rain (OC 676).

The interpretation of these passages is tricky. However, they could be understood as follows. If an utterance of the sentence “I am dreaming” is taken to entail the negation of a hinge, that utterance is meaningless. After all, Wittgenstein construes hinges as constitutive of language games and, presumably, the dreamer is not a participant in a language game. If one insists that such an utterance is not meaningless, one may be conflating its empty use to deny a hinge with other uses, like those in which one expresses “one’s surprise with respect to an unusual circumstance” (Coliva, 2021, 220–221). If language bears constitutive relations to thought, the argument generalizes from language to thought.

Coliva finds this view unpalatable, for “it depends on a conception of meaning as use that is not without problems and that may not be commonly agreed upon” (Coliva, 2022, 37). As I see it, the problem lies not in the connection of meaning with use but in Wittgenstein’s tendency to connect both of them too tightly. Let us grant that words acquire their meaning by being used in non-skeptical language games. It is hard to see why, once they have acquired their meaning by being used in non-skeptical language games, words cannot retain their original meaning when used in some skeptical scenarios.

This point is illustrated by what I will call “skeptical switch scenarios” (also invoked in Coliva, 2022, 96–97). In a skeptical switch scenario, a subject moves from being in a good case to being in a bad case. For our current purposes, a good case is a situation in which a subject has and can have perceptual knowledge of external world propositions. A bad case is a situation in which the subject is, unbeknownst to her, cut off from the external world while she still has experiences that are subjectively indiscriminable from the experiences she would have in a good case. By way of illustration, imagine that Clara lived a normal life until she was 20 years old. During that period, Clara acquired lots of perceptual knowledge of the external world. Unfortunately, during Clara’s 20th birthday, a group of scientists captured her while she was sleeping, drugged her, and removed her brain, which they kept alive in a vat of nutrients. After that, they used cutting-edge technology to connect Clara’s brain to a supercomputer that now feeds it with stimulations that exactly mimic the stimulations that Clara’s brain would receive if she were having veridical experiences of the world. ¹⁴

It seems natural to use ordinary correspondence to explain what makes this a skeptical switch scenario. Before the envatment, the contents of Clara’s perceptual experiences and beliefs were true because they corresponded to a mind-independent reality. After the envatment, the contents of Clara’s perceptual experiences and beliefs were false because they did not correspond to a mind-independent reality. Without ordinary correspondence, we could hardly make sense of this as a skeptical switch scenario.

What about hinges? By Coliva’s own admission, hinges were true before the envatment. Otherwise, we could hardly understand why Clara managed to acquire perceptual knowledge of external world propositions during her first 20 years of life. However, several problems arise when one asks what happened with the truth-values of Clara’s hinges after her envatment. In what follows, I argue that the conjunction moderatism and alethic pluralism faces a dilemma. If hinges remain true after the envatment, moderatism cannot improve upon dogmatist and disjunctivist treatments of skepticism. If hinges turn false after the envatment, alethic pluralism cannot offer a coherent characterization of the skeptical switch scenario.

3.1. First horn

Recall Coliva’s take on the argument from bipolarity (Section 1): although hinges are not capable of being true or false, they are plain truths whose truth property is exhausted by their satisfaction of some platitudes (Coliva, 2022, xxi). This view commits her to claiming that hinges like <I am not a BIV> remain true after the envatment. So, after the envatment, <I am not a BIV> satisfies platitudes [1]–[6], and the satisfaction of those platitudes fully specifies its truth property. This view dovetails with Coliva’s favorite analogy between mathematical axioms and hinges: “the assumption of a hinge proposition is comparable to the positing, within the proof, of one or more of the theory’s axioms, which are typically considered to be true, at least within the theory, and, once fixed, aren’t (non-circularly) provable within the theory” (Coliva, 2022, 61). At least since Descartes (1996), mathematical propositions have been taken to remain true in partial skeptical scenarios like dreams.

Unfortunately, this view does not offer anti-skeptical relief. To see why, I would like to focus on two skeptical problems that feature in Coliva’s discussion: the underdetermination problem and the assurance problem.

The underdetermination problem originates from the following principle:

Underdetermination. For any subject, S, and pairs of incompatible propositions p and q: If S knows that p and q are incompatible propositions, and S’s reasons do not favor p over q, then S lacks epistemic justification to believe p. ¹⁵

Suppose that Clara knows that <I have hands> is incompatible with <I am a BIV>. Suppose further that Clara’s sole reason for the former proposition is an experience as of having hands. If Clara’s experience does not favor the hands proposition over the BIV proposition, then Clara lacks epistemic justification to believe <I have hands>.

Coliva thinks that dogmatism, an important rival of moderatism, lacks a good solution to the underdetermination problem, for it works with a conception of perceptual reasons that does not favor external world propositions over skeptical hypotheses (Coliva, 2015, 20–28, 58–64, 2022, 100–101). Additionally, the dogmatist holds that, for a suitable class of perceptually basic propositions, hinges are not necessary for perceptual justification (Pryor, 2000). If we use the propositional variable “p” to range over perceptually basic propositions, we can formulate dogmatism as follows:

Dogmatism. Absent defeaters, if a subject, S, has an appropriate course of experience (typically, an experience as of p), S is justified in believing p.

Since dogmatism posits perceptual reasons that do not favor external world propositions over skeptical hypotheses, dogmatism does not provide sufficient conditions for perceptual justification:

[A]ccording to the theory under scrutiny, which is based on an internalist notion of [justification], [a subject] could have that very same experience even if he were indeed dreaming of a hand in front of him. Intuitively, in such a predicament, it would seem odd that that experience, just by itself, could give him a [justification] in support of “Here is my hand.” Another way of putting the same point could be this: if one’s hand-like experience increases the likelihood of p and p*, how could it by itself justify p over p*? (Coliva, 2015, 59). ¹⁶

Unfortunately, the situation is not brighter for the conjunction moderatism and hinge minimalism. If hinges remain true after the envatment, one could have those very same hinges even though one was in a bad case. But, if one could have those very same hinges while being in a bad case, it is unclear how hinges could give one what dogmatism lacks. If ordinary, external world propositions are meant to correspond to a mind-independent reality, how could propositions that are only minimally true—and are therefore indifferent to the correspondence or lack of correspondence of ordinary, external world propositions with a mind-independent reality—give one the additional justification that dogmatism lacks? If justification supervenes on factors that speak to the truth of a proposition, and the truth of ordinary, external world propositions is defined by correspondence with a mind-independent reality, adding purely mind-dependent hinges will make no difference.

In a more recent text, Coliva writes:

[I]f a certain kind of evidence e, like a perceptual experience, is compatible with mutually incompatible kinds of propositions, namely propositions about mid-size physical objects (p) or about BIVs being stimulated so as to have those experiences, absent any causal interaction with the relevant physical objects (q), in order for e to accrue to a justification for propositions of kind p rather than q, some extra condition has to be met. It is only in this way that we will have a justification for propositions of kind p and will be within our rights in taking a given experience, which is a mind-dependent kind of evidence, to bear on propositions about mind-independent objects (Coliva, 2022, xiv; also: 100–101).

If the conjunction of an experience as of <I have hands> with a hinge is compatible with mutually incompatible kinds of propositions, namely <I have hands> (before the envatment) and <I have no hands> (after the envatment), then in order for this conjunction to accrue to a justification for the proposition <I have hands> rather than <I have no hands>, some extra condition has to be met. Given moderatism and hinge minimalism, the conjunction of an experience as of <I have hands> with a hinge is compatible with mutually incompatible kinds of propositions, namely <I have hands> (before the envatment) and <I have no hands> (after the envatment). Therefore, the conjunction moderatism and hinge minimalism does not provide sufficient conditions for epistemic justification.

Coliva introduces epistemological disjunctivism as another competitor of moderatism (McDowell, 1995; Pritchard, 2012, 2016). ¹⁷ For our current purposes, we can work with a generic formulation:

Epistemological disjunctivism. In the good cases, a subject, S, can have perceptual reasons that guarantee the truth of external world propositions. ¹⁸

Before the envatment, Clara can see that there is a mango in front of her. Since Clara’s seeing that there is a mango in front of her entails that there is a mango in front of her, Clara has at least one perceptual reason that guarantees the truth of an external world proposition. Given this analysis, epistemological disjunctivism can avoid the underdetermination problem. After all, Clara’s possession of that perceptual reason is incompatible with the BIV hypothesis. If Clara sees that there is a mango in front of her, it follows that Clara is not a BIV. Unfortunately, Coliva discards epistemological disjunctivism because it does not offer assurance against Cartesian skeptical hypotheses:

[D]isjunctivism is impotent with respect to global skeptical hypotheses of a Cartesian kind. For we cannot tell, just based on our experience, that we are not BIVs, since everything would, ex hypothesis, look identical to us. That is why disjunctivism needs [hinge epistemology]. For the latter dismantles Cartesian skepticism, by reminding us that “I am not a BIV” is not an ordinary belief, which is open to doubt, verification and control, but is rather a commitment of ours—it is, in fact, a possible expression of our über-hinge commitment that we cannot possibly be mistaken (Coliva, 2022, 29).

Sadly, the problem of assurance also arises for the conjunction moderatism and hinge minimalism. If we cannot tell that we are not BIVs and the hinge <I am not a BIV> remains true after the envatment, then, for all we know, our perceptual experiences and beliefs could be massively wrong.

We could extend this argument to <My sense organs work mostly reliably>. Consider now Coliva’s favorite hinge: <There is an external world>. Even if one does not endorse hinge minimalism, there is a clear sense in which <There is an external world> remains true even after the envatment. Indeed, the skeptical switch scenario would be unintelligible if there was no external world before and after the envatment. In our short story, there is a human being, Clara, who lived a normal life until she was 20 years old. There are other human beings besides Clara. Those human beings performed actions that can only take place if there is an external world: they captured Clara while she was sleeping, drugged her, and removed her brain, which they kept alive in a vat of nutrients. They also used cutting-edge technology to connect Clara’s brain to a supercomputer that now feeds it with stimulations that exactly mimic the stimulations that Clara’s brain would receive if she were having veridical experiences of the world. Although many of these actions had a deceptive intent, they did not bring the external world to an end. So, the truth of <There is an external world> is fully consistent with the possibility that the contents of all one’s current perceptual experiences and beliefs are false.

In sum, if hinges remain true after the envatment, moderatism cannot solve the underdetermination and assurance problems. So, moderatism cannot improve upon two alternative anti-skeptical strategies: dogmatism and epistemological disjunctivism. ¹⁹

3.2. Second horn

Perhaps Coliva should claim that hinges like <I am not a BIV> turn false after the envatment. This move is only available, however, if Coliva is willing to revise her concessive response to the argument from bipolarity (Section 1). On the new view, skeptical switch scenarios are cases in which some hinges can be false. Unfortunately, this concession would also prevent Coliva from offering a coherent characterization of skeptical switch scenarios on behalf of hinge minimalism. ²⁰

If the hinge <I am not a BIV> turns false after the envatment, what could explain its change in truth value? One option would be to hold that, after the envatment, the hinge did not correspond to a mind-independent reality. Sadly, this contradicts hinge minimalism; it introduces a truth-determining factor that goes beyond the platitudes propounded by hinge minimalism. Another option is to hold that, after the envatment, sentences expressing <I am not a BIV> do not satisfy (some of) the platitudes propounded by hinge minimalism. Unfortunately, this view is implausible. Platitudes [1]–[6] are defined independently of any relation to a mind-independent world. If <I am not a BIV> was minimally true only in the good case, it would follow that <I am not a BIV> is not merely minimally true. The truth of <I am not a BIV> would perfectly covary with being in a good case. But this perfect covariation goes beyond hinge minimalism; it introduces an external dependency relation between <I am not a BIV> and being in a good case.

Moreover, this perfect covariation would make the skeptical switch scenario unintelligible. If, after the envatment, sentences expressing <I am not a BIV> did not satisfy (some of) the platitudes propounded by hinge minimalism, then Clara’s bad case would be discriminable from a good case. After all, envatted Clara would only need to ask herself whether [1] sentences that express that hinge satisfy the disquotational schema, whether [2] they can be meaningfully negated, whether [3] they can be embedded in conditional statements, whether [4] she is prepared to act on that content, whether [5] she is willing to judge or assert that content, and whether [6] that content is true relative to her own Weltbild. A negative answer to any of her questions would suffice to discriminate her bad case from a good case.

Two reasons speak against the unintelligibility alternative. First, it is unclear why the envatment episode could possibly have any influence on a sentence’s satisfaction of (some of) the platitudes propounded by hinge minimalism. Second, Coliva is committed to thinking of the skeptical switch scenario as intelligible, as shown at the introduction to this section.

In sum, if hinges turn false after the envatment, Coliva cannot provide a coherent characterization of skeptical switch scenarios on behalf of hinge minimalism.

4. Two Possible Replies

In this section, I explore two possible replies to the skeptical switch argument and show that they compromise some key tenets of moderatism.

Objection 1. Coliva might want to block the skeptical switch argument by extending minimalism to ordinary, external world propositions (Coliva, 2022, 107). On this generalized minimalism, the truth property of all external world propositions would be exhausted by their satisfaction of some platitudes. Since correspondence goes beyond those platitudes and skeptical switch scenarios introduce failures of correspondence, generalized minimalism predicts that skeptical switch scenarios are unintelligible. If skeptical switch scenarios are unintelligible, one cannot raise questions pertaining to the alethic profile of hinges after skeptical switches.

Response. This move makes it hard to distinguish moderatism from the radical semantic anti-skeptical strategy that Coliva finds in Wittgenstein’s remarks on dreams (Section 3). As we have seen, Wittgenstein imposes very strict limits on intelligibility by connecting meaning and use too tightly. If Coliva endorsed generalized minimalism, she would connect truth and use too tightly. If meaning is analyzed in terms of truth conditions, she would also connect meaning and use too tightly. Finally, Coliva would owe us an explanation of the intuitive appeal of skeptical switch scenarios.

Objection 2. The skeptical switch argument only works if one takes for granted that hinges stay put across contexts. That view is not mandatory, though. As Wittgenstein himself makes clear, hinges can be “demoted” from their rule-like status to play merely descriptive roles. For example, after a severe car accident, one can find out, with relief, that one has hands. So, <I have hands> may lose its hinge status and be treated as a bipolar, empirical proposition that can be rationally assessed (OC 23; Coliva, 2022, xxix, 23). By parity of reasoning, an episode of envatment could demote the hinge <I am not a BIV> from its hinge status, turning it into a bipolar, empirical proposition that can be rationally assessed (Schönbaumsfeld, 2016, 117).

Response. The phenomenon of hinge demotion is real. Nevertheless, it is not available to moderatism. Coliva’s commitment to constitutivism leads her to focus on propositions that are necessary for humans to rationally form, revise, retain, and evaluate empirical beliefs (Section 1). If the skeptical switch scenario is coherent, then, after the envatment, Clara should be able to rationally form, revise, retain, and evaluate her beliefs in the light of her experiences. Coliva takes the skeptical switch scenario to be coherent and characterizes <I am not a BIV> as necessary for humans to rationally form, revise, retain, and evaluate empirical beliefs (Coliva, 2022, 32). Therefore, Coliva is committed to holding that <I am not a BIV> stays put after the envatment.

Imagine now that <I am not a BIV> could be demoted from its hinge status. This concession would have serious costs for moderatism. First, it would be hard for Coliva to stick to her arguments for propositionality, which presuppose that general hinges are not subject to contextual shifts (Section 1). Second, Coliva would need to either abandon or revise her constitutivism. The costs of both moves would be high, though. If Coliva abandoned her constitutivism, she would have no reason to identify hinges with very general propositions. Moreover, she would be left with no account of the concept of human epistemic rationality. If Coliva revised her constitutivism, the weaknesses of her anti-skeptical strategy would become perfectly clear. The most obvious modification would be to insist that putative hinges that can turn false after a skeptical switch are not genuine hinges. So, <I am not a BIV> would have two readings. If one took it to mean <I am not a recent BIV>, it would not be a genuine hinge because it could turn false after a skeptical switch. If one took it to mean <I have not lived a total envatted existence>, it would be a genuine hinge because it could not turn false after a skeptical switch. Notice, however, that this move would make it perfectly clear that moderatism cannot have an adequate response to the underdetermination problem. Coliva could only retain constitutivism if she restricted her account to very general hinges that remain true across all contexts, including contexts where a skeptical switch has occurred. Alas, those very general hinges are unsuitable to deal with skeptical switch scenarios, for the truth of very general hinges is indifferent to skeptical switches. More specifically, those very general hinges do not favor external world propositions over hypotheses that describe skeptical switch scenarios. ²¹

5. An Alternative

In this section, I present an alternative picture that successfully blocks the skeptical switch argument. In Section 5.1, I suggest that the propositions that Coliva identifies as hinges are best seen as having a substantive truth property that is grounded in ordinary correspondence. In Section 5.2, I show how one can pair that correspondence-first view with epistemological disjunctivism to solve the underdetermination problem. I also offer independent arguments to dismiss the assurance problem. In Section 5.3, I respond to two objections.

5.1. Correspondence-first

Correspondence-first is a semantic view of the propositions that Coliva treats as hinges. That semantic view should not be seen as a development of Coliva’s hinge epistemology. It is only meant to provide a semantic characterization of those propositions with the aim of restoring the coherence of skeptical switch scenarios.

Correspondence-first attributes a substantive truth property to a select group of ordinary external world propositions. Those are propositions that count as basic given our best psycho-semantic theories. Although I have views on this matter, we can treat that select group as a theoretical placeholder to be filled by one’s favorite psycho-semantic theory. Following Russell (1918) and Wittgenstein (1921), correspondence-first envisions explaining the truth of general external world propositions in terms of the correspondence (or lack of correspondence) between one’s select group of ordinary external world propositions and bits of reality together with logical relations between that select group of ordinary external world propositions and general external world propositions. As a result, the truth property of general external world propositions is substantive and grounded in ordinary correspondence. ²²

Consider the proposition <There is an external world>. Plausibly enough, this proposition remains true after the envatment. This verdict gets support from an intuition of correspondence. Even after the envatment, there is a match between propositions that entail this general proposition and bits of reality. In our story, the proposition <Clara exists> corresponds to the continued existence of Clara, who lived a normal life until she was 20 years old and was envatted afterwards. The same holds for other external world propositions, which describe situations in which other human beings captured Clara, and performed actions that can only take place if there is an external world. More generally, correspondence-first can specify correspondence relations between some ordinary external world propositions and bits of reality and assign truth conditions to <There is an external world> via logical relations to those ordinary external world propositions.

Consider now the proposition <I am not a BIV>. Intuitively, if not-p, it is false that p. Thus, this proposition is equivalent to: It is false that <I am a BIV>. What does it take for <I am a BIV> to be false? Correspondence-first offers a natural interpretation. It means that some ordinary external world propositions do not correspond to bits of reality. For example, <I am a brain> and <I am in a vat> are descriptions of the world that do not match reality, where I am (or have) much more than a brain; I am (or have) a body who is currently sitting in front of its laptop. Correspondence-first theorists rely on these and other ordinary external world propositions to derive, via logical relations, the truth property of <I am not a BIV>.

Of course, the history of philosophy has shown that the details will be complex. However, this difficulty is common to most conceptions of truth I am aware of. Thus, rather than dwelling on the shared problem of complexity, I would like to insist that correspondence-first enables us to provide a coherent characterization of the semantic value of Coliva’s putative hinges in skeptical switch scenarios.

5.2. Epistemological disjunctivism

Coliva thinks that dogmatism cannot solve the underdetermination problem because an experience as of p does not favor p over many skeptical hypotheses. However, the conjunction moderatism and hinge minimalism is not better placed to solve the underdetermination problem. In this section, I defend a conditional claim: If Coliva’s critique of dogmatism is sound, only epistemological disjunctivism can solve the underdetermination problem.

Schönbaumsfeld (2016, 2019) has deployed three complementary strategies to deal with skeptical switch scenarios. Following Pritchard (2012), she insists that 1) it is epistemically rational to disbelieve skeptical switch hypotheses if one lacks reasons to believe that those skeptical hypotheses obtain. She also argues that 2) we are justified in disbelieving skeptical switch hypotheses because we have background evidence that speaks against them. For example, we have evidence that the state of technology is insufficient to remove someone’s brain, keep it alive in a vat with nutrients, and connect it to a supercomputer that feeds it with stimulations that exactly mimic the stimulations one would have if one was having veridical experiences of the world. Finally, Schönbaumsfeld thinks that 3) epistemological disjunctivism can come to the rescue of hinge epistemology to solve the underdetermination problem (see also Pritchard, 2016).

In my view, those who accept Coliva’s critique of dogmatism are committed to rejecting 1–2 as solutions to the underdetermination problem. So, their only option is to endorse one or another form of epistemological disjunctivism.

If the skeptical switch scenario is coherent, envatted Clara cannot subjectively discriminate her bad case from a good case. So, even if Clara was envatted, she would still lack reasons to believe that the skeptical switch scenario is actual. Also, even if Clara was envatted, she would still have background evidence that allegedly speaks against the skeptical switch hypothesis. The point is even clearer with some variations on our original case. For example, before the envatment, the group of scientists managed to hide the advanced state of technology from Clara. Or, after the envatment, they implanted in Clara false memories of the primitive state of technology. Hence, both Clara’s lack of reasons to believe skeptical switch hypotheses and her possession of background evidence are fully consistent with the obtaining of skeptical switch hypotheses. So, none of these factors increases the likelihood of <I am not a BIV> over the likelihood of <I am a BIV>. Hence, if one grants Coliva’s critique of dogmatism, one cannot avail oneself of strategies 1–2.

The way out is to develop a view according to which, in the good cases, one can have reasons that one cannot have in the bad cases. Given that view, skeptical hypotheses are not fully consistent with one’s possession of those reasons. That is precisely what epistemological disjunctivism offers us: perceptual reasons that guarantee the truth of external world propositions. If one can have perceptual reasons that guarantee the truth of some external world propositions, correspondence-first is not an obstacle to solving the underdetermination problem. While correspondence-first restores the coherence of skeptical switch scenarios, epistemological disjunctivism puts the external world within our epistemic reach. ²³

To be sure, the conjunction correspondence-first and epistemological disjunctivism is not available to Coliva, who complains that epistemological disjunctivism does not provide assurance. To my mind, it is a mistake to saddle our anti-skeptical strategies with an assurance requirement. First, it is widely held that a theory of epistemic justification should identify the conditions under which it is epistemically appropriate (Pryor, 2005) or epistemically permitted (Goldman, 1986) to believe some propositions. As far as I can see, a theory of epistemic justification can guarantee that those conditions are met even though that theory does not offer assurance (pace Fumerton, 1995). Second, suppose that one insists on providing assurance, perhaps as part of a second-order epistemological inquiry. This requirement would speak against moderatism. There is nothing in the very idea of unjustified (and unjustifiable) hinges that are minimally true that can provide one with assurance that one is in a good case. Third, an assurance requirement is in tension with acceptance of the coherence of skeptical switch scenarios. If one grants that one could be in a skeptical switch scenario, one has conceded that one cannot subjectively discriminate one’s current case from a bad case. But, if the good case is subjectively indiscriminable from the bad case, one cannot be completely sure that one is not in a bad case. ²⁴

5.3. Objections and replies

Objection 1. The conjunction correspondence-first and epistemological disjunctivism leads to a questionable form of Mooreanism.

Response. “Mooreanism” can mean different things. It can mean that one knows or is in a position to know the denials of skeptical hypotheses. It can also mean that one has justified beliefs in the denials of skeptical hypotheses. Or it can mean that Moore-style (Moore, 1939) arguments provide cogent proofs of the denials of skeptical hypotheses. While the conjunction correspondence-first and epistemological disjunctivism is consistent with each of these claims, it does not entail any of them. And, while it would be good to deal with each of these issues, that is something I cannot do in this article.

Correspondence-first does entail that the propositions that Coliva treats as hinges are bipolar. Epistemological disjunctivism also entails that one can have reasons that favor external world propositions over the denials of skeptical hypotheses. However, these claims do not entail that one knows, is in a position to know, or has justified beliefs in the denials of skeptical hypotheses. On a traditional view, if a subject, S, knows p, then 1) p is true, 2) S believes p, and 3) S’s belief in p is doxastically justified. Correspondence-first is a semantic claim about condition 1, so it can be satisfied even if conditions 2 and 3 are not. If one thinks otherwise, one must hold that conditions 2 and 3 are not logically independent from condition 1. That is an implausible view. Epistemological disjunctivism introduces reasons that favor external world propositions over skeptical hypotheses. One can have those reasons even though one does not form any anti-skeptical belief on the basis of those reasons. So, the view is consistent with the lack of doxastic justification for believing the denials of skeptical switch hypotheses. Since neither correspondence-first nor epistemological disjunctivism are theories of the structure of epistemic justification or rationality, they make no predictions on the cogency of Moore-style arguments. ²⁵

To be sure, my arguments do have consequences for hinge epistemology. If one thinks that Coliva has correctly identified the hinges of human rational practices and that only non-bipolar propositions can work as hinges, then correspondence-first is inconsistent with hinge epistemology. However, there are hinge epistemologies that are not tied to semantic claims about meaning and truth (Pritchard, 2016; Wright, 2004). If one thinks that hinge epistemology must solve the underdetermination problem, then any reliance on epistemological disjunctivism should be seen as an alternative to hinge epistemology. Nevertheless, my approach is still consistent with attempts at using Wittgenstein’s remarks on hinges to deal with other issues, like the regress argument (Moyal-Sharrock, 2016), the coherence of total skeptical scenarios (Schönbaumsfeld, 2016, Ch. 2), and closure-based skepticism (Pritchard, 2016; Schönbaumsfeld, 2016, Ch. 4).

My own view is that hinge-like entities can play two roles in epistemology. If we gain factive perceptual reasons by employing perceptual-cognitive capacities, we will need to spell out some environmental regularities that enable those capacities to deliver factive perceptual reasons. I think that some of those regularities will come close to many of the hinges that feature in Wittgenstein’s writings. Moreover, theories of inquiry must spell out the conditions under which one can rationally answer some questions (OC 341). A question has a true answer only if the presuppositions of that question are true. Many of Wittgenstein’s hinges could be seen as presuppositions of questions that have true answers.

Objection 2. The conjunction correspondence-first and epistemological disjunctivism is more arrogant and dogmatic than the conjunction moderatism and hinge minimalism (Coliva, 2018, 274–275, 2022, xx).

Response. The charge of arrogance is unjustified. Assume that it is not ascertainable whether our experiences and beliefs correspond to a mind-independent world. Thus, anyone who denies that the truth of Coliva’s hinges consists in correspondence is as arrogant as anyone who accepts that the truth of Coliva’s hinges consists in correspondence. Since hinge minimalism denies that the truth of hinges consists in correspondence, hinge minimalism is as arrogant as correspondence-first.

Perhaps the conjunction correspondence-first and epistemological disjunctivism is unduly dogmatic. As Coliva (2022, 65) puts it, “it is simply question-begging to assume for a fact that there is an external world in the face of skeptical worries, which challenge the rational legitimacy of such an assumption.” Notice, however, that the conjunction moderatism and hinge minimalism is not less dogmatic. Coliva’s main reason to treat hinges as true is that false propositions cannot play a framework role. And she wants her favorite hinges to play that framework role. But what reasons do we have to think that those propositions rather than others play the framework role? As the skeptical switch scenario has shown, there is no easy answer to that question. And, as our reconstruction from Section 2 has made clear, Coliva has offered no independent reasons in favor of hinge minimalism. Indeed, Coliva embraces hinge minimalism as an alternative to other approaches that cannot be reconciled with her own view of hinges as true, albeit non-bipolar, propositions. So, Coliva’s view begs the question against those of us who are attracted to correspondence-first and epistemological disjunctivism.

6. Conclusion

Moderatism is a brand of hinge epistemology that construes hinges as unjustified and unjustifiable propositions that are constitutive of the concept of human epistemic rationality. This view faces the truth problem: to offer an account of truth for hinges that is not threatened by skepticism. Coliva has tried to solve this problem by combining a correspondence conception of truth for ordinary external world propositions with a minimalist conception of truth for hinges. I have argued that this conjunction of views cannot coherently characterize skeptical switch scenarios while offering an effective anti-skeptical strategy. After examining two possible replies, I have presented a two-component alternative. First, correspondence holds for a select group of ordinary external world propositions that, via logical relations, ground the truth property of the propositions that Coliva treats as hinges. Second, one can use epistemological disjunctivism to solve the underdetermination problem and cite independent reasons to dismiss Coliva’s demand of assurance. While this view is logically consistent with Mooreanism, it does not entail it. So, it can incorporate some of Wittgenstein’s ideas to shed light on several aspects of our epistemic practices. ²⁶