On What Does Rationality Hinge?

in International Journal for the Study of Skepticism
AutorIn: Yuval Avnur 1
Mehr anzeigen Weniger anzeigen
  • 1 Scripps College, Claremont yuval.avnur@gmail.com

The two main components of Coliva’s view are Moderatism and Extended Rationality. According to Moderatism, a belief about specific material objects is perceptually justified iff, absent defeaters, one has the appropriate course of experience and it is assumed that there is an external world. I grant Moderatism and instead focus on Extended Rationality, according to which it is epistemically rational to believe evidentially warranted propositions and to accept those unwarrantable assumptions that make the acquisition of perceptual warrants possible and are therefore constitutive of ordinary evidential warrants. I suggest that, even though Extended Rationality might be true, it cannot do the work that Coliva wants it to do. Although my objections do not show that it is false, they can serve to clarify what sorts of problem a theory of justification or rationality could possibly address. This provides an alternative to Coliva’s view of the skeptical problem and the question, on what does rationality hinge?


The two main components of Coliva’s view are Moderatism and Extended Rationality. According to Moderatism, a belief about specific material objects is perceptually justified iff, absent defeaters, one has the appropriate course of experience and it is assumed that there is an external world. I grant Moderatism and instead focus on Extended Rationality, according to which it is epistemically rational to believe evidentially warranted propositions and to accept those unwarrantable assumptions that make the acquisition of perceptual warrants possible and are therefore constitutive of ordinary evidential warrants. I suggest that, even though Extended Rationality might be true, it cannot do the work that Coliva wants it to do. Although my objections do not show that it is false, they can serve to clarify what sorts of problem a theory of justification or rationality could possibly address. This provides an alternative to Coliva’s view of the skeptical problem and the question, on what does rationality hinge?

1 Introduction: A Dispute within the Family

Annalisa Coliva’s Extended Rationality: A Hinge Epistemology is one of the clearest, most wide-ranging, and convincing books on skepticism I have read. It is also incredibly dense, with an astonishing quantity of philosophical ideas and arguments per page. As if that weren’t enough, Coliva manages to carve out some genuinely novel territory in the debate about rationality and the structure of justification. I think Coliva’s efforts were largely successful. Here, I can only critically discuss a small, albeit central, part of it.

The two main components of Coliva’s view are Moderatism and Extended Rationality. According to Moderatism: “a belief about specific material objects that P is perceptually justified iff, absent defeaters, one has the appropriate course of experience (typically an experience with content that P) and it is assumed [by the subject] that there is an external world (and possibly other general propositions, e.g., ‘My sense organs work mostly reliably’, ‘I am not the victim of massive cognitive deceptions,’ and so on)” (54). We can call, as Coliva sometimes does (e.g., p. 20), the propositions that must be assumed in order for one’s experience to justify one’s perceptual beliefs ‘hinge propositions’. So, a subject must assume some hinge proposition in order for her experience to justify belief in that experience’s content. However, the hinge assumption itself does not amount to a justified belief. That one must make this assumption in order to have perceptual justification contradicts Liberalism, and that this assumption need not (and apparently cannot) be epistemically justified, or “warranted,” contradicts Conservatism. In this way, Moderatism breaks the stalemate of the Liberal vs. Conservative debate, and, according to Coliva in the first three chapters, it avoids the serious problems those two views face.

The problems for Liberals and Conservatives that Coliva focuses on concern the structure of perceptual warrants, Moore’s much discussed ‘proof’ of an external world, bootstrapping, transmission failure, the contents of perception, and closure. These are all traced more or less directly to the problem of skepticism about the external world, or more precisely skepticism about whether our ordinary perceptual beliefs are justified. I will leave the many important and subtle moves made in those first three chapters aside. I grant that Moderatism fares better than Liberalism and Conservatism in addressing such problems.1 For those who are worried about other skeptical problems aside from the external world, the fifth and last chapter proposes a way to apply Moderatism to problems involving the epistemology of induction, testimony, and logic. In each case the strategy is to specify some hinge propositions that one must assume in order for one’s ordinary beliefs to be justified, where that assumption itself need not, and perhaps cannot, be justified. I will also set this material from the fifth chapter aside, though there is, again, much to admire and discuss. Instead, I will focus entirely on the arguments of the fourth chapter.

Recall that there are two elements to Coliva’s view, and so far I have only discussed Moderatism. The title of the book derives from a particular version of Moderatism that Coliva favors, which is the subject of the fourth chapter. What is the status of our hinge assumptions, which enable us to have perceptual justification? To answer this question, Coliva proposes a view of rationality that extends beyond beliefs that are perceptually justified (though of course it includes those justified beliefs as well). The view is called ‘the extended rationality view’—from here on, ‘er’: “it is epistemically rational to believe evidentially warranted propositions and to accept those unwarrantable assumptions that make the acquisition of perceptual warrants possible in the first place and are therefore constitutive of ordinary evidential warrants.” (129) So, our hinge assumptions are epistemically rational despite our lacking evidence, or justification, to believe them. For example, it is epistemically rational for you to assume that you are not a brain-in-a-vat, even though you have no evidence, and no justification, to believe that you are not a brain-in-a-vat. What makes this assumption rational is that such an assumption is required in order for your perceptual beliefs to be justified and therefore rational, and, furthermore, this assumption is constitutive of what epistemic rationality is. As Coliva puts it, given these facts, one has a “rational mandate” to assume hinges (or, at least, to start out assuming them, prior to or independently of any empirical evidence).2

The bulk of Chapter 4 aims to address problems for Moderatism by appeal to er. In the rest of this paper, I will suggest that er is either unnecessary for, or cannot, address these problems. I then suggest that, in the latter cases in which er cannot address the problems, this is because no theory of rationality, or justification, could possibly address such problems. So, none of this shows that er is false. Rather, it shows that er cannot do the work that Coliva wants it to do. And, ultimately, it clarifies what sorts of problem a theory of justification or rationality could possibly address. This, I suggest in the end, provides an alternative to Coliva’s view of the skeptical problem and the question, on what does rationality hinge? To be clear, then, my suggestion is compatible with Moderatism and er, but diverges from Coliva on the significance of the latter and the relation of the former to some skeptical problems. My disagreement is a dispute within the family of Moderatists who are willing to accept er, rather than an outright rejection of Coliva’s view.

2 The “Humean Skeptical” Problem

It will be helpful to formulate an alternative to er that the Moderatist could accept instead. In discussing some problems for Moderatism, I will consider whether this alternative version fares just as well as er. I will call it Limited Rationality, or ‘lr’: One is epistemically rational in holding perceptually justified beliefs, but epistemic rationality is limited, in both its positive and negative pronouncements, to such beliefs, so that any required, epistemically unjustified hinge assumptions are neither epistemically rational or epistemically irrational. In other words, according to lr, the propositions for which one has a rational mandate according to er are, instead of being rational or constitutive of rationality, simply a-rational. Like breathing, wiggling your toes, or being alive, the hinge assumptions posited by Moderatism are simply outside the scope of epistemic rationality or rational evaluation. With lr in mind as a contrast to er, I now consider the problems for Moderatism that Coliva addresses.

The original question to which er is an answer is what Coliva calls the ‘Humean Skeptical’ problem: our basic assumption on which perceptual justification hinges is itself unjustified, or ungrounded. This, the Humean skeptic says, makes the whole of epistemic rationality questionable. To be clear, this is distinct from the more standard, “Cartesian” skeptical worry that our ordinary beliefs are unjustified. The Moderatist has an answer to that sort of skepticism (as Coliva argues in the previous chapters): our ordinary perceptual beliefs are typically justified, partly in virtue of our basic, hinge assumptions, which themselves are not epistemically justified. It is the status of our hinge assumptions that the Humean skeptic is concerned with, and this is the focus of the problems she means to solve by appeal to er.

According to er, although hinge assumptions are unjustified, they are rationally mandated and are constitutive of rationality. So, that they are unjustified is only the beginning of the story of their status, and once we realize that part of what it is to be rational is to make these assumptions, the basic Humean worry is assuaged. Now, it seems to me that rational mandates and the claim that the hinge assumption is constitutive is not necessary here. One could just as effectively assuage the basic worry by pointing out that, though the belief in the hinge assumption is not justified, it is not therefore irrational. Rather, according to ir, it is a-rational, not evaluable for its rationality at all. It is epistemically groundless only in the same sense that wiggling your toes is epistemically groundless; it is just not the sort of thing that is up for rational evaluation. There is therefore nothing wrong with the assumption from an epistemic, or epistemically rational, perspective. Why? Hinge assumptions are requirements or prerequisites of rationality, and this is why they are not rationally evaluable. On this very general level, lr seems as well equipped as er to answer the Humean skeptic.3 But, to see whether this is really the case in detail, let us break up the Humean concern into separate versions, as Coliva does. Coliva interprets these as objections to er, and in particular er’s response to the basic Humean skeptic. If we avoid er in our reply to the Humean skeptic, as I just suggested we could, then these are problems for Moderatism in general, rather than for er in particular. That is how I will understand them, in order to see whether er is necessary for the Moderatist reply to the Humean skeptic.

3 er and Closure

The first problem concerns closure, and it is widely known as the “abominable conjunction.” As Coliva describes it, “Cartesian” skepticism appeals to a closure principle in order to conclude that our ordinary perceptual beliefs are unjustified. The familiar idea is that belief in some hinge proposition such as that we are not brains in a vat is unjustified, and given closure this implies that our ordinary beliefs are also unjustified. This is so because our ordinary perceptual beliefs (e.g., that we have hands) entail that we are not brains in a vat (in an otherwise empty world). As Coliva argues in previous chapters, and as we are assuming, Moderatism handles that: it is not a necessary condition on ordinary beliefs’ perceptual justification that we have justification to believe that we are not brains in vats. Thus, the unrestricted closure principle appealed to by the Cartesian skeptic is false. This is how we leave Cartesian skepticism behind and come to address the issue of Humean skepticism about our hinge assumptions (rather than the question about our ordinary perceptual beliefs). But, how big a price to pay is the denial of unrestricted closure for the Moderatist?

There are several potential problems for deniers of closure, and their formulation depends on the exact formulation of closure in question, or the version appealed to by the Cartesian skeptic. Some of these problems are solved by appeal to Moderatism in earlier chapters.4 What does appealing to er add to the defense of denying closure? Coliva cites two main additional advantages.

First, according to er, closure for rational mandate still holds (137). That is, though closure for justification (or warrant) fails in the way just described, it is still the case, according to er, that if one has a rational mandate to believe p, and if p then q, then one has a rational mandate to believe q. This may take some of the sting out of denying closure outright, since it gives us an unrestricted substitute: closure for rational mandate, if not for epistemic justification. However, lr can provide similar substitutes. First, the Moderate who accepts lr can accept closure for ordinary perceptual beliefs, since the hinge assumptions are not ordinary perceptual beliefs. Second, according to lr, closure can still hold for any beliefs that are within the scope of rationality. That is, for any beliefs that can be evaluated according to rationality, the closure principle holds for them. Third, the status of being justified or a prerequisite for justification can still be closed, even if the status of being justified is not. This seems to me to be as helpful (or unhelpful) in getting us to feel better about denying unrestricted closure as Coliva’s er-inspired proposal. In all three cases, we offer a version of closure that seems sufficiently complete, but that avoids commitment to Cartesian skepticism.

The second and main way in which Coliva appeals to er in defense of Moderatism’s denial of closure concerns the so-called “abominable conjunction.” According to Keith DeRose (1995), the conjunction “I know that I have hands and I don’t know that I’m not a handless biv” is abominable, in the sense that it couldn’t be true. But, deniers of closure are committed to its truth. The relevant version of the conjunction here would be: “I am justified in believing that I have hands and I am not justified in believing that I am a handless biv.” Again, Coliva offers some independent replies to this objection in earlier chapters, which seemed to me to be sufficient. But she also appeals to er for an additional defense against this objection. The idea is that once we add the relevant information, that although I am not justified in believing that I am not a biv, I am rationally mandated to believe it, the conjunction no longer seems absurd (138). And, without all of this added information, the conjunction seems absurd only because it is misleading in its omission of all of the relevant information. However, the Moderatist who accepts lr seems to have similar options for neutralizing the conjunction by adding the relevant information. For example, one could add that although I am not justified in believing that I am not a biv, this is because it is a requirement of justification that one assume this. Or, one could add that although I am not justified in believing that I am not a biv, it is not irrational for me to assume this because it is a necessary assumption for all perceptual justification, and is therefore outside the scope of what one can evaluate by appealing to rationality. Once “rational mandate” is spelled out, I doubt whether Coliva’s added information makes the conjunction much more palatable than these added claims would.

So, when it comes to making Moderatism’s denial of closure more palatable, er does not appear to be necessary. Rather, lr seems to afford the Moderatist about as much help as er does.

4 er and Relativism

So far, I have suggested that er gives us no special advantage in overcoming the general Humean skeptical worry about Moderatism or the specific worry about closure that Coliva considers. The next problem for Moderatism, which also seems to stem from Humean skepticism, is epistemic relativism. The problem is that, if nothing justifies or grounds belief in the hinge assumption, then any assumption is as good as any other. If so then the alternative hinge assumptions would in turn justify different, incompatible sets of beliefs. The different, conflicting systems of justification are all on a par, because there seems nothing to settle which assumption and which set of justified beliefs is the “right” one. Coliva appeals to er in addressing this challenge (128, 140): our hinge assumptions are constitutive of epistemic rationality, while other, incompatible assumptions are not. Therefore, our assumptions are not arbitrary, in that they are mandated by rationality.

There is something unsatisfying about this appeal to er. Relativism could be reformulated as the view that one could use a concept just like rationality but which mandates a different set of assumptions. Those different assumptions would then be constitutive of that alternative sort of rationality. Why use our rationality rather than that alternative version of rationality? They each mandate their own sets of assumptions, so er itself doesn’t seem to help us with this problem. That our assumptions are mandated from within our current system of justification does not make it less arbitrary that we use this, rather than a different system of justification which would mandate different assumptions. If anything, it merely shows that we are internally consistent in our appeal to rationality. Coliva sets out to address this worry.

Coliva’s reply (128) is that, first, given that the beliefs in question are those of humans, the relevant assumptions must concern our human perceptual experiences; and, second, they must concern basic epistemic practices, rather than those that depend on the exercise of other practices. Note, however, that this reply does not appeal to er; it is a comment on what version of relativism is a real threat. Understood as a comment about what kind of relativism is threatening, we can regard it as a challenge to Moderatism in general, rather than specifically to er. Whereas the relativism Coliva addresses considers alternative notions of epistemic rationality, with its alternative rational mandates, one could similarly consider alternative systems of justification, each of which require alternative, a-rational (relative to that system) hinge assumptions.

In any case, given what Coliva takes to be constraints on any viable relativist threat, Coliva conceives the problem of relativism in terms of two questions: (i) whether we can form warranted beliefs about mind-independent objects in our surroundings by methods other than perceptual experience, with different assumptions than the ones we actually make, and (ii) whether by using different assumptions we can come to have warranted beliefs not about mind-independent objects as we now typically think of them, but about mind-dependent objects (142).

er is used to address (i) and (ii) as follows: (i) In order for a system of justification to result in non-perceptually justified beliefs in mind-independent objects, the system must apply to creatures that are not at all like humans. For humans, the only cognitive access to mind-independent objects is through experience (144). (ii) The contents of our experience are such that they entail mind-independent objects, so any system of justification which makes our experiences the source of justification for our beliefs about the external world must be a system that assumes the existence of mind-independent objects (143).

But notice that these replies do not in any way depend on er. It is open to any Moderatist, including one who accepts lr, to accept these replies to (i) and (ii). We can appeal to the idea that a creature with non-perceptual sources of information about mind-independent objects is very different from humans, as Coliva does in response to (i), even if we endorse lr instead of er. And, for one who accepts lr, Coliva’s reply to (ii) amounts to a defense of the a-rational assumptions being the assumptions that match the contents of our experiences. The assumptions—whether they are constitutive or a-rational—that are required by Moderatism are shown to be non-arbitrary relative to the contents of our experiences.

So, Coliva’s appeals to er in response to relativism do no better than an appeal to lr could. They both depend on something external to rationality, justification, or the required assumptions: our human situation shows that our assumptions and system of justification is not arbitrary.

5 er and the “Oblomovian Challenge”

The challenge here is that “one might opt out and decide not to play the game at all.” (146) That is, even if we grant that the assumption that, say, there is an external world is mandated by epistemic rationality, why should we care about, and engage in, epistemic rationality in the first place? This challenge seems to arise from the Humean skeptical problem, since it is brought on by reflection on the idea that our hinge assumption is groundless. If the only thing that makes it grounded is that it is mandated by epistemic rationality, then the Humean skeptic turns to the idea that epistemic rationality itself is groundless, and therefore seemingly optional.

In this formulation of the challenge, it is posed as a challenge for er, rather than to Moderatism itself. However, I think a version of this challenge is easily discernable for Moderatism in general. Granted that some hinge assumption is required in order for our perceptual beliefs to be justified, one might question why we should care about justification at all, given that it hinges on a groundless assumption. So, I propose to construe the Oblomovian challenge as a challenge for all versions of Moderatism, not just to those that accept er.

It seems that Coliva is aware of the limitations of er when it comes to this problem. Instead, she suggests that there may well be pragmatic reasons (for humans like us) to adopt the mandated assumptions, as they seem to be necessary for survival, and perhaps to human life as we know it. These serve as non-epistemic reasons to engage in epistemic rationality. Epistemic rationality itself cannot provide reason for engaging in epistemic rationality: “What we are trying to do [in epistemology], rather, is to provide us with reasons to think that the rules of the game—indeed the very game of epistemic rationality—are kosher from an epistemic point of view.” (146) And er does do that, by showing that the rules, determined by the hinge assumptions, are constitutive of, and therefore mandated by, epistemic rationality.

The first, concessive point, seems to me correct: er cannot offer us an answer to the Oblomovian challenge. I’ll return to the idea that only practical reasons can do this below. But Coliva’s second point, which does appeal to er, seems on reflection not to require er. One could state an equally compelling, or conciliatory, response to the challenge by stating the following: if you’re looking for epistemic assurance about the hinge assumptions, the only one you can get is to notice that in order for there to be any epistemic rationality, you must make those assumptions. This is something that a Moderatist who opts for lr can say as well. And it seems to have the same effect (for better or worse) as an appeal to the idea that the necessity of these assumptions makes them constitutive of epistemic rationality. It is, it seems, in virtue of the fact that constitutive conditions are necessary conditions that the Oblomovian challenge is addressed in this way from an epistemic point of view. lr posits the same necessary conditions. So, again, er is not what is doing the work for the Moderatist.

Coliva addresses another problem for Moderatism in a subsequent section, namely that no connection between our hinge assumptions and the truth has been established. Unless such a connection is established, the game of epistemic rationality is undermined, given that its aim is to arrive at true beliefs. I will set this last problem aside, since Coliva explicitly appeals to a non-reality theory of truth, rather than to er, in addressing this last challenge. And my purpose is to critically assess Coliva’s appeal to er in order to solve Moderatism’s problems.

6 Conclusion: What er, and Epistemology, Can’t Do

The upshot of my critique of Chapter 4 is not that er is false. Rather, the upshot is that there are problems for Moderatism, stemming ultimately from the Humean skeptical challenge concerning the groundlessness of our hinge assumptions, whose solution does not require er. This is not an objection to er because these problems apply to any theory of rationality, or justification. They should not be grouped with the sorts of skeptical problems that Moderatism was meant to address. In other words, I don’t think the problems for Moderatism that have been discussed show that Moderatism is false, either. Rather, I think the problems discussed above are the sort of problem that cannot be solved by any theory of justification or rationality. They are not epistemological problems.

The two problems for Moderatism, that I have in mind here are the basic relativist challenge and the Oblomovian challenge. As I suggested, I think the problems related to closure can be handled by er and lr equally well; the question of closure is one about the structure or principles of justification, and that is indeed an epistemological problem. In contrast the relativist and Oblomovian challenges seem external to the inner workings of justification or rationality. The relativist asks: Why not use some other version of ‘rational’ instead of the one we actually use? The Oblomovian asks: Why use our, or any, notion of ‘rational’ at all? These questions could be posed to any theory of justification or rationality, not just to Moderatism. Both questions demand, in effect, reasons to use or care about some notion rather than another, or at all. If Moderatism is right, then any such notion will require—and according to er be constituted by—some assumptions about the contingent causal relation between our experiences and reality. So one way to answer the relativist and Oblomovian challenge is to provide some reason to believe that the required assumptions are true. Coliva argues that perceptual justification, since it always depends on the very assumption that we are here inquiring about, cannot be appealed to. But another approach might be to search for a priori reasons to believe that the hinge assumption, or our hinge assumptions, are true. That would be spectacular, given that the assumptions are (deeply) contingent claims about how the world happens to be. But, if that could be done, it would provide an answer to the relativist and Oblomovian: our notion rests on (probably) true assumptions, so it is not an arbitrary one and it is one that we should care about insofar as we care about truth.

But suppose, as Coliva does, that no such a priori argument is available. So the only remaining sort of reason we could provide is pragmatic, as Coliva herself suggests. Perhaps what human life in fact requires, or the form that it in fact takes, provides practical reasons for accepting our hinge assumptions. If so, then that, too, would address the challenges. Our hinge assumptions are not arbitrary: they are the humanly practical ones. And the reason to care about rationality at all is that there is a practical point to doing so given the demands of human life. In other words, play the game of rationality because you are human, or because of your nature. This is of course far too glib. One potential problem is that, in appealing to the relevant aspects of human nature, one presumably must avoid empirical premises held on the basis of perceptual justification, which would, according to the Moderatist, appeal to the assumptions and system of justification that are here in question. So, in order for appeals to human nature to address this problem, they must be limited to those aspects of one’s own “inner” nature that one can observe within, or a priori. Needless to say, this is no easy task. (Notice, though, that at least one of Coliva’s appeals to human nature, namely the details of the contents of our experiences, is arguably discernable without appeal to empirical premises.)

My point in rehearsing these points is to illustrate that the demands made by the challenges to Moderatism are not the sort of thing a theory of rationality could possibly provide. These are not problems concerning the structure or inner nature of rationality. Rather, they are questions external to rationality, about its relation to our lives and the world that we contingently live in. These are either metaphysical or practical questions. With respect to relativist and Oblomovian challenges, rationality hinges, not on itself or its own internal mandate, constitution, or necessity, but on metaphysical, contingent claims about how the world happens to be and what we need from it.



I defend what amounts to a Moderatist view in Avnur (2012a) and (2012b).


The parenthetical qualifier is important, and perhaps underemphasized in Coliva’s discussion. For, it is possible to get empirical evidence that one is massively deceived. In such a case, one rationally comes to disbelieve the assumptions that Coliva says are constitutive of rationality. Thus, what is constitutive is that one has started out assuming the hinge propositions, not that one maintains the assumptions come what may.


Coliva considers a similar view on p. 124, but the version she criticizes appeals crucially to the idea that hinge assumptions are “norms” or rules rather than straightforward claims. She rejects this idea for compelling reasons. But one might think that the mere fact that assuming some claim, or even believing it, is a necessary prerequisite for having any justified beliefs at all is enough to suggest that such an assumption or belief is outside the scope of epistemic rationality. The only objection Coliva offers to this seems to be that this version of Moderatism provides no account of why the skeptic mistook the hinge beliefs to be within the scope of justification when in fact they are not (126). But it seems to me that an explanation is readily available: the skeptic (and most of the rest of us) are initially tempted to assume that norms or principles of justification and epistemic rationality apply to all beliefs. But, once one carefully distinguishes the role that some beliefs play in our system of justification, one sees that some beliefs, the hinges, are prerequisites for, and therefore not within the scope of, epistemic rationality. So, this more simple version of lr remains plausible, and remains an effective, Moderatist response to Cartesian skepticism.


And I argue that the problems for rejecting closure are not decisive in Avnur (2012a).

If the inline PDF is not rendering correctly, you can download the PDF file here.


Insgesamt Im letzten Jahr In den letzten 30 Tagen
Aufrufe von Kurzbeschreibungen 90 24 0
Gesamttextansichten 194 56 23
PDF-Downloads 41 24 4