The paper presents the key themes of my Extended Rationality: A Hinge Epistemology. It focuses, in particular, on the moderate account of perceptual justification, the constitutive response put forward against Humean skepticism, epistemic relativism, the closure principle, the transmission of warrant principle, as well as on the applications of the extended rationality view to the case of the principle of the uniformity of nature, testimony, and the justification of basic laws of inference.
It is natural to think that justifications stem from a system of assumptions. Take, for instance, the mathematical case. Theorems—that is, justified mathematical propositions—are obtained through proofs, given certain axioms. Indeed, different sets of axioms constitute different theories, which give rise to different sets of justified mathematical propositions. However, no matter how many sets of axioms are possible, and no matter whether the propositions eventually justified are different, the former are clearly necessary in order to generate proofs and therefore warrants for certain less basic mathematical propositions.
Take a more humdrum case. You see a ball roll between two poles.1 You form the belief that a goal has just been scored. Yet, clearly, that belief is justified only insofar as it is assumed that a football match is being played and this, in turn, justifies us in inferring that supporters of the team whose player sent the ball between the poles will be cheering. If it were a different game, whose point is actually that of letting the ball roll between the poles in one’s mid-field, while players of the other team should prevent that from happening, you would not be justified either in believing that a goal has just been scored, or in inferring that supporters of the team whose player sent the ball between the poles will be rejoicing.
Many further interesting examples can be found in Wittgenstein’s On Certainty (oc). For instance, he noticed that geological beliefs about the specific age of the Earth could only be justified by taking for granted that the Earth had existed for a very long time. Only that way could fossils and other evidence be brought to bear on the issue of the specific age of the Earth (see, for instance, oc 183–192). To see why, consider the hypothesis that the Earth, with all its fossils, had just been created five minutes ago. We would still have those fossils yet they could not be taken to prove anything about the specific age of the Earth. If a specific belief about the age of the Earth can be justified at all, it is only thanks to those fossils and the collateral general assumption regarding the long existence of the Earth.
All testing, all confirmation and disconfirmation of a hypothesis takes place already within a system [of assumptions]. And this system is not a more or less arbitrary and doubtful point of departure for all our arguments; no, it belongs to the essence of what we call an argument.
That is to say, the questions that we raise and our doubts depend on the fact that some propositions are exempt from doubt, are as it were like hinges on which those turn.
That is to say, it 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 105, 341–343)
2 The Moderate Account of Perceptual Justification
Yet, the main bulk of Extended Rationality does not deal with the case of beliefs about the distant past and their characteristic presuppositions regarding the Earth’s long existence. Rather, it concerns the structure of perceptual justification. That is, the kind of justification we have, based on current sense experience, for propositions about mid-size objects in our environment, which are the content of our beliefs. Examples of this kind of proposition are: “Here is my hand,” based on seeing a hand in front of one’s nose; “Here is a pc,” based on watching the screen while writing this very Précis; and so on. The claim at the heart of Extended Rationality is that, when specific empirical beliefs are at stake, perceptual justification is only possible based on a system of assumptions. That is to say, it is not enough merely to have a certain type of experience—hand-like or pc-like—in order to justify the corresponding empirical proposition, “Here is a hand” or “Here is a pc.” Rather, we need one or more general assumptions that allow us—no matter how defeasibly that might be—to bring those experiences to bear on a world populated by physical objects. Why so? The reason is, roughly, that experience by itself underdetermines the propositions that could legitimately be taken to be justified by it.
Compare this with the case of “A goal has just been scored.” We take the experience of seeing a ball roll between two poles to justify that proposition only thanks to already taking for granted that a football match is being played. However, as we saw, that very experience could be just the same if a different game were being played. If so, however, a different proposition (or set thereof) would be justified. Going back to the case of a hand-like experience: just by itself it could equally justify “Here is a hand”, “I am hallucinating having a hand”; “I am a biv (a brain in a vat) who is having a hand-like experience”, and so on. Hence, taking that experience to partly justify “Here’s a hand”, rather than any proposition compatible with that very experience, depends on already taking for granted that we are interacting with a world populated by physical objects, that our sense organs mostly work correctly (and, possibly, some other propositions, for example “I am cognitively lucid and not a victim of massive perceptual and cognitive deception”).
Notice, moreover, that the general propositions I claim must be assumed in order for our experiences to bear legitimately onto other propositions about mid-size objects in our environment, so that the latter are justified, are not needed to give us an indefeasible justification for these more specific empirical propositions. Ceteris paribus—that is, given those very assumptions and experiences—we could still be facing papier-mâché hands, for instance. What we need those assumptions for is to be able to overcome what one might call ‘our cognitive locality.’ Namely, we need them in order justifiably to go beyond our experiences and to bring them to bear on a universe populated by physical objects, whose precise identity and properties can, of course, still escape us in certain circumstances, rather than take them to be caused by appropriate neurological stimulations. To be more precise: if 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.
Hence, to repeat, the claim at the heart of Extended Rationality is that perceptual justification can take place only thanks to a system of very general assumptions, such as “There is an external world,” “My sense organs work mostly reliably,” “I am not a victim of massive perceptual and cognitive deception”, and so on. A problem as old as the very history of epistemology—epitomized by “Agrippa’s trilemma”—concerns the epistemic status of these assumptions. In the quest for justification, each horn of this trilemma is thought to be problematical: either we end up providing circular justifications; or we embark on an infinite regress; or else, we stop with unjustifiable and therefore a-rational and arbitrary assumptions.
Suppose we hold that each assumption, in its turn, needs to be warranted, in order for it to generate perceptual justification, together with the appropriate kind of experience. For, one may think, it is only if these assumptions are justified that our ordinary empirical beliefs based on them will rest on a secure base and will therefore be justified. Consider the football case: it is only if I am independently justified in believing that a football match is being played that my experience of seeing a ball roll between two poles provides a justification for “A goal has just been scored.” I think that in this case there is no dispute. Why so? Because it is indeed very easy to see how that assumption can be independently justified, for instance: I know that I paid for a ticket to the football match between teams A and B in the stadium where I am now sitting, watching the game; or, I know that every Sunday a football match is played in the stadium where I am, roughly at this time, and that today is Sunday; or else, if I am watching the match on television, I know that it has been advertised as the football match between the two teams; or that commentators keep repeating that this is a crucial football match, or saying that the team that prevails will win the World Cup, and I know that the World Cup is a football tournament; and so on.
Yet, as soon as we move away from the football example, things become much more complicated. Consider the historical case and the very general proposition that the Earth has existed for a very long time before our birth. One might think that that proposition is justified by a lot of our specific historical beliefs based, in their turn, on testimonies, both personal and documentary, often recorded in academic texts. However, as we saw before, those testimonies and documents could be just the same and yet have appeared and been recorded in academic books only a few minutes back. Therefore, clearly, it is not to be expected that a justification for such a general proposition could be obtained by inferring to it starting with premises that are justified just as long as that very proposition is taken for granted. That kind of justification would ultimately be circular and it would be no justification at all.
Nor is it to be expected that justification for it could ensue from coherence between it and our further beliefs. Justifications are epistemic goods—to put it in most general terms—that should speak to the truth of what they are supposed to justify. Yet, starting with the same evidence—apparent testimonies, documents and academic records—we could just as well produce a different and yet entirely coherent system of propositions. In that system the general assumption is that the Earth has just been created replete with everything we find in it and the corresponding specific empirical propositions are like “It looks as if Napoleon was defeated at Waterloo about three centuries ago.” Nothing makes the first system of beliefs more likely to be true than the second one. If we think otherwise it is either because we are more used to it and therefore think that it is epistemically kosher; or else it is because we consider its specific beliefs justified and think that this, in turn, gives us a justification for its basic presuppositions. However, in the former case, we would conflate our willingness to endorse a given system of beliefs with proof of its truth and, in the latter, we would try to provide a circular justification for its basic assumptions, starting from beliefs that are justified only insofar as those very assumptions are taken for granted.
Another possibility is to think that we have a priori justification for “The Earth has existed for a very long time.” Where would it come from, though? Intuition is an appealing answer, but only momentarily, because one then faces the problem of explaining its nature and workings. This remains one of the philosophically most arduous tasks.2 Perhaps we have some kind of a priori yet inferential justification, coming from reflection on the very meaning of the terms involved. Notice, however, that this would immediately be hostage to the particular theory of meaning we are prepared to subscribe to. For it is only on the basis of some inferential-role semantics, which take either a holistic or a molecularist form, that we can sensibly claim that, for instance, it is constitutive of the meaning of “Earth” that it has existed for a very long time. Yet, a direct referentialist could simply say that “Earth” refers to the planet we are all living on now, whether it has existed for a very long time or only for five minutes, and that this is the meaning of “Earth.”
Faced with this kind of difficulty—to repeat, distrust in justifications for general assumptions, stemming from specific beliefs that would be justified only by already taking them for granted; as well as in coherence theories of justification, and mistrust in intuition and in inferential a priori justifications stemming from meaning-constitutive considerations—recent years have seen the emergence of yet another proposal, which belongs to the a priori camp broadly construed. This proposal provides for non-evidential warrants, called entitlements, for very general background presuppositions, such as, “The Earth has existed for a very long time.” Entitlements however, at least in the main way in which they are currently thought of,3 are not meant to speak to the truth of these propositions. Yet, if this is the case, it is very hard to see how entitlements could be genuine epistemic warrants for them, since they are neither evidential warrants nor guides to the truth of the relevant propositions, capable of providing a viable solution to the original problem they were meant to address. Namely, the problem of how these general assumptions could actually be epistemically justified.
Similar considerations to the ones just rehearsed for “The Earth has existed for a very long time” could be made for “There is an external world,” “My sense organs work mostly reliably” and “I am not a victim of massive perceptual and cognitive deception,” which, arguably, are the presuppositions thanks to which our sensory experiences can be taken (defeasibly) to justify our beliefs about mid-size objects in our environment. If this were the situation, since we can neither provide immediate justifications for these propositions, nor mediate ones, it would seem that the skeptical outcome would ensue. That is to say, it would seem that the only plausible alternative would be to hold that these are just a-rational assumptions and that, even if we think we are justified in believing ordinary empirical propositions, we are not.
I think that in broad outline this is the path that (save for considerations regarding coherence and entitlements) led Hume to his skepticism. However, it is again Hume who, to my mind, offered the first seeds to try to escape it, as paradoxical as that might seem. Seeds that were developed much later on, in a different direction, by Wittgenstein in On Certainty, as I think Peter Strawson was the first to recognize in his Skepticism and Naturalism: Some Varieties (1985).
According to Hume, we cannot help believing that there is an external world, so that our sensory experiences are constantly brought to bear on a world populated by mid-size objects that are taken to exist independently of our minds, even when they are not directly perceived by us. For Hume, it is part of our psychological constitution that we cannot but form beliefs and devise actions accordingly. That is the way we live. That is the human condition; but notice that, for him, the human condition is the Humean condition of being forced by nature to follow certain forms of psychological and practical conduct that fall outside rational sanction. Rationally, however, we have to recognize that our most basic beliefs are not justified and neither are our more specific empirical beliefs based on perceptual evidence.
Wittgenstein, in contrast, put forward the view that even though we cannot justify these very general assumptions (or indeed, in his view, even more specific ones which are equally necessary for certain sorts of empirical practices and inquiries), we cannot help but make them thanks to our upbringing within a community that shares language and certain epistemic practices or, more generally, a form of life. However, his idea was that the human condition is not the Humean one at bottom. That is to say, one in which there is an unbridgeable gulf between what reflection imposes on us—that is, the recognition that all justification for ordinary empirical propositions rests on unwarrantable assumptions—and what we cannot help doing, given our psychological and more culturally determined nature—like going on living as if, thanks to those assumptions, our ordinary beliefs were justified. Rather, the human condition, in his view, is one in which we simply have to recognize that whatever degree of justification we possess for our ordinary empirical beliefs, and that we do in fact possess, it takes place within a system of assumptions, which are neither justified nor justifiable.4 Therefore, according to Wittgenstein, the human condition is importantly different from the Humean one, primarily because justifications are indeed possible, at least for ordinary empirical propositions, but only thanks to a system of unwarrantable assumptions.
This is the kind of picture about the structure of perceptual justification that I present and defend in some detail in Extended Rationality. It can been seen, among other things, as the attempt to make good one of the horns of Agrippa’s alleged trilemma. According to that trilemma, no justification is ever possible because there are no immediately justified propositions, which can serve as the basis for all others,5 and so the quest for justification ultimately leads to an infinite regress; nor can justification be produced in a circular way6 or by resting on unjustified assumptions. The view I present and defend in the book agrees that, when it comes to very general propositions, such as “There is an external world,” we cannot immediately justify them (whatever that might mean as we have briefly explored above), nor can we justify them in a circular way by advocating beliefs which are justified only as long as these assumptions are already taken for granted. However, it aims to vindicate the idea that even if these assumptions are neither warranted nor warrantable, they can serve to produce a justification for ordinary empirical propositions, once we enjoy the appropriate kinds of experience.
I call this view the “moderate” conception of perceptual warrant, as it can be seen as lying in between the so-called “liberal” view, proposed in recent years by Jim Pryor (2004), and the “conservative” view defended mostly by Crispin Wright (2004). In outline, the first one corresponds to the intuition that perceptual justification is not theory-laden. As long as there are no defeaters, our perceptual experiences give us an immediate justification for ordinary empirical propositions such as “Here is a hand.” In contrast, the conservative view has it that a warrant for ordinary empirical propositions can be had only if certain general assumptions are independently justified.
The idea I defend is that, contrary to the liberal position, we need assumptions to overcome our cognitive locality—that is, if we want to form defeasibly justified beliefs about physical objects in our environment based on our experiences. Yet, contrary to the conservative view, these assumptions need not be warranted, for, in fact, they cannot. The reasons why they cannot are explored at some length in the book, following the kind of reasoning we have quickly rehearsed here for the case of “The Earth has existed for a very long time,” and by analyzing the details of several contemporary attempts to provide independent justifications for “There is an external world,” and the like. For present purposes, let me stress that the moderate architecture of the structure of perceptual warrant just says that a specific empirical proposition P, for instance “Here is a hand,” is perceptually justified iff one has the relevant kind of experience, such as a hand-like one, and the background assumption that there is an external world is in place (possibly together with other ones such as, “My sense organs are mostly working reliably,” “I am not the victim of massive perceptual and cognitive deception,” and so on), while there are no defeaters. Since this definition is compatible with various ways of thinking of the status of such an assumption, which range from an externalist positing that the world is just like that, to making it the content of a doxastic attitude of a specific subject, moderatism is introduced as a family of possible views and not as just one single position. Yet, they would all be different species of the same genus—the genus I call, following the Wittgensteinian metaphor, “hinge epistemology”—because they all hold that perceptual justifications take place “within a system” (oc 105) of assumptions, that is of propositions that lie beside the route of inquiry and that make justifications within the inquiry possible in the first place.
Furthermore, these species of the same genus are compatible with different accounts of how we should think of the content of perceptual experience in order for the latter partially to constitute a justification for ordinary empirical beliefs. Indeed, it is my conviction that the moderate architecture of the structure of perceptual warrants has been endorsed, in one version or another, by many different philosophers, like naturalists of a Humean persuasion (provided they were prepared to forsake Hume’s skeptical attitude at the reflective level), Wittgenstein in On Certainty, and naturalists inspired by him, like Strawson. In addition, pragmatists would turn out to be moderates, in my view, as would those externalists about the nature of perceptual justification who are prepared to recognize a role for general assumptions, like Ernest Sosa in recent writings.
3 Extended Rationality and Skepticism
One could then be tempted to think that moderatism inspired by some of Wittgenstein’s considerations in On Certainty would offer only a momentary relief from skeptical worries for—the train of thought would go—it would remain that if those assumptions are not justifiable, then they may well turn out to be false. Hence, nothing guarantees that our epistemic practices rest on a secure basis. Yet this, according to Wittgenstein, would be right only if it made sense to call those assumptions into question. That is to say, it would be right only if those assumptions were in the business of epistemic appraisal at all; if it made sense to apply to them the very categories of truth and falsity and, more importantly and less contentiously, the very categories of being justified/unjustified, or even known or unknown. But the main thrust of On Certainty, at least according to the kind of, so-called, ‘framework reading’ I myself have put forward,7 is that those very general assumptions are not like empirical propositions of a more general kind, contra what G.E. Moore held. Rather, they are similar to rules; that is to say, they play a normative role and, like rules, are not subject to truth or falsity, nor to assessment in terms of justification or lack thereof.8 Compare with “Stop at traffic lights when red.” It is intuitive to think that it does not correspond to a pre-ordinate fact, and so that it does not make sense to think of it as either true or false in any robust sense of that word. Nor, for the same reason, would it make sense to think of it as either justified—that is as supported by further facts or experiences—or as unjustified—as disconfirmed by further facts and experiences. 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.”
I myself embrace the Wittgensteinian view that justifications for ordinary empirical propositions are possible thanks to a system of assumptions—that is, owing to a system of more general propositions, which, as such, cannot be justified. However, I do not wish to endorse the view that these assumptions are rules, devoid of empirical content, if that is indeed Wittgenstein’s considered view on the issue.9 Yet, if this is a sensible avenue to explore as far as the status of “There is an external world” is concerned, it actually seems to be in danger of re-opening the door to the skeptical challenge. For now, how would one block the conclusion that this is merely an assumption we make which, however, is actually unjustified and therefore not rational, exactly as a skeptic would hold? This is the challenge the extended rationality view I present and defend in the book is meant to face. Accordingly, if either empirical, or coherentist, or a priori kinds of warrant for “There is an external world” are unattainable and entitlements are only putative epistemic warrants, we may defend the epistemic legitimacy of making that assumption by claiming that, even though unwarranted, it is in fact constitutive of epistemic rationality itself. Just as both rules and moves are part of any game so, I argue, both constitutive assumptions and perceptual justifications, which are possible thanks to them, are part of epistemic rationality. To ban constitutive assumptions from epistemic rationality simply because they are not warranted (as they cannot be), like skeptics do, is due to too narrow and unmotivated a conception of the extent of epistemic rationality. Namely, one that confines it to perceptually justified beliefs only. In contrast, epistemic rationality extends beyond the latter to those very assumptions that make it possible to produce ordinary perceptual justifications and to have the kind of practice (or method) of forming, assessing, and withdrawing from empirical beliefs on the basis of perceptual evidence, which is itself constitutive of our very notion of epistemic rationality. If so, it turns out that we are actually mandated by epistemic rationality itself to assume “There is an external world.” However, a rational mandate is not an epistemic warrant—namely, an epistemic good that speaks to the truth of what it is meant to warrant. Skeptics are right to think that we have no such warrant for “There is an external world.” However, they are wrong to think that, for that very reason, that proposition falls outside the scope of epistemic rationality and that, for that very reason, we cannot have perceptual warrants for our ordinary empirical beliefs.
One may then worry that even if “There is an external world” is epistemically rationally mandated, it may still be false and hence that the extended rationality view has done little to counter the skeptical challenge. It is here, however, that I think we should ponder more on the semantic assessment of that proposition and, in particular, on what it means to say that it is true. As is familiar, there are two broad notions of truth: a realist, mind-independent one, and an anti-realist, evidence-dependent one. According to the former, no matter what we think or judge, a proposition is true (or false) in its own right, because it corresponds (or fails to correspond) to some pre-ordinate, mind-independent fact. What is seldom noticed is that it is only on such a conception of truth that a skeptical concern with respect to “There is an external world” makes sense. For it is only on such a realist conception of truth that, despite the fact that nothing we take ourselves to know speaks against it, it might still be false. Still, in order to counter the skeptical challenge we cannot revert to a familiar anti-realist, evidence-dependent view of truth either. For it is a tenet of hinge epistemology that all specific empirical truths are known (or justifiably believed) only by taking that very general proposition for granted. Hence, we cannot use the former to prove the truth of the latter. Yet, as remarked, I do not wish to endorse what is usually taken to be the Wittgensteinian view, according to which hinges are not truth-evaluable. It is at this junction that I propose to endorse a minimalist view of truth with respect to them. Accordingly, they satisfy certain platitudes: they may enter the disquotational schema, and allow for meaningful negation and embedding in suppositional contexts. So much suffices for predicating their truth. However, the kind of truth-property they enjoy is neither of a robustly realist kind, nor of a familiar anti-realist, evidentialist kind. Rather, I propose to consider it of an anti-realist, yet non-evidentialist kind. Hinges like “There is an external world” are true because we hold on to them and because they make it possible for us to acquire justification and knowledge of ordinary empirical propositions. Therefore, by playing that role, nothing can actually speak against them. Thus, hinges are true, in my view. To suppose that despite all we take ourselves to know they might still be false would depend on still being in the grip of a realist conception of truth, which one would be entitled to endorse in this connection only if there were no other option.
Hence, the final and specific version of hinge epistemology I endorse has it that thanks to true and epistemically rationally mandated assumptions such as “There is an external world,” together with appropriate courses of experience, we can and do have perceptual justifications for ordinary empirical beliefs such as “Here is a hand.” However, to repeat, this is the species of the hinge epistemology genus I endorse. It is not the only possible one; even though I am convinced it is the one that has the best prospects of success, because it speaks to the skeptical challenge, albeit by developing an indirect response to it. Namely, not a response that contradicts the skeptic by providing what he is looking for and maintaining that it is impossible to attain, that is ordinary epistemic warrants for “There is an external world.” Rather, the extended rationality view is a response that shows that the skeptical quest is somehow illegitimate when it comes to very general propositions like “There is an external world.” For it asks for warrants that cannot be obtained; it is based on too narrow and unmotivated a conception of epistemic rationality; and, finally, it depends on a realist conception of truth that is by no means the only possible alternative.
4 Extended Rationality, Relativism and Oblomov
The extended rationality view also speaks to the peculiar relativist challenge that, unlike other forms of epistemic relativism originated by reflections in philosophy of mathematics and science, were to target the assumptions of the basic practice (or method) of forming, assessing, and withdrawing from empirical beliefs on the basis of perceptual evidence.10 The thought would be this: we are familiar with the view that different mathematical or logical theories are possible and hence that there can be different sets of axioms that would generate proofs for different theorems. Why not think, then, that even though our conception of epistemic rationality has its own background presuppositions, such as “There is an external world,” other ones would be possible too? In particular, those that would do without that assumption and that did not take our experience to bear on beliefs about mind-independent objects, yet could account for that very experience.11 In response to this challenge, I claim that this alternative conception would not even have the means to account for the very content of our perceptual experience, which is objective, that is to say as of objects and/or properties out there, and not as of mere subjective variations, like sensations of cold and heat. Nor would it be able to explain, so-called perceptual constancies. That is to say, the phenomenon whereby no matter how far or near an object we move, and so no matter how many changes in perceptual input there are, we keep having a representation as of the same object or property (cf. also Burge 2010). Arguably, all this happens without the intervention of concepts, through the workings of partially encapsulated perceptual faculties. Hence, it would be difficult to explain the very content of our perceptual experiences if it were not assumed that our perceptual faculties must operate, or at least have evolved, in an environment populated by physical objects. If, in contrast, one wished to say that those perceptual representations are indeed the result of the operation of concepts, rather than of the subpersonal and automatic workings of our perceptual faculties, then it would actually turn out that the best way of explaining them would involve an appeal to the very assumption this alternative model should have dispensed with, namely that there are mind-independent objects. For, if one insisted that perceptual constancies and the like were, in fact, the product of the intervention onto perceptions of our concepts, our overall cognitive system should be such as to embed a general rule that unifies subjective and scattered representations in such a way as to produce a representation of an object or a property out there.
Finally, I consider what I dub ‘the Oblomovian challenge’, after Goncharov’s novel Oblomov. I take that challenge to consist of the idea that we may opt out of playing the game of epistemic rationality altogether, as it were. Not in order to play any other game, with different characteristic assumptions, as a relativist would have it, but simply in order not to play that game, or alternative ones in the same ballpark, at all. I hold that only faced with the Oblomovian challenge can practical considerations be appealed to in order to motivate our acting the way we do. For instance, it would be very risky for us to live if we did not take our sensory experiences to bear onto physical objects and went on acting irrespective of what we seem warranted in believing based on those very experiences. Yet this challenge is not epistemological in nature, but rather practical or even psychological. Therefore, it is only to be expected that it will be responded to by appealing to that kind of consideration.
Hence, the overall picture I present and defend is that perceptual justification is possible only thanks to some general assumptions, which are mandated by epistemic rationality itself and that are universal as long as we are dealing with creatures like us that have to rely on their senses to form beliefs about physical objects in their environment and that have a certain kind of perceptual experience. Whether it is conceivable to imagine different creatures who can either rationally intuit truths about physical objects, rather than arrive at them through the operation of their senses, or that there might be creatures with altogether different perceptual experiences, it is not an issue that should really preoccupy us. For what we are trying to understand is the human condition and the kind of epistemic security for some of our actual beliefs that can be attained.
To repeat, at the heart of the moderate conception lies the claim that the assumptions that make the acquisition of perceptual warrants possible are not warranted. For we do not possess, nor can possess, any proof of their truth. Yet, given the extended rationality view, those assumptions are epistemically rationally mandated and therefore true (in a minimalist and anti-realist, non-evidentialist way, as we have seen). Thanks to them and to appropriate courses of experience, we do possess a perceptual justification for many ordinary empirical beliefs.
5 Closure and Transmission Failure
A number of important consequences follow from such a general picture. For example, it follows that the principle of closure for justification under known entailment is not unconditionally valid. For “Here is my hand” entails “There is an external world.” Yet, while we can justifiably believe the former (and the entailment), we cannot justifiably believe the latter. Still, in my view, this does not lead to any “abominable conjunction”12 of the kind “I justifiably believe there is my hand here, but I don’t justifiably believe there is an external world” sic et simpliciter. Rather, the kind of conjunction we get, once the extended rationality view is endorsed, is “I justifiably believe that here is my and and, although I don’t justifiably believe there is an external world, I am epistemically rationally mandated to assume there is.”
Furthermore, we have to recognize that beside the kind of warrant transmission-failure principle originally presented by Wright (1985, 2004), according to which an argument cannot generate (or enhance one’s previous) warrant for a conclusion if, and only if, the warrantedness of its premises depends on already possessing a warrant for its conclusion, there is another kind of warrant transmission-failure principle, which is indeed at issue in the kinds of cases that are of most interest to philosophers. Namely, the one according to which an argument cannot generate (or enhance one’s previous) warrant for a conclusion if, and only if, the warrantedness of its premises depends simply on the very assumption of its conclusion. It is for this reason that also on the moderate architecture of perceptual warrant, and not only on its conservative counterpart, Moore’s argument (“Here is a hand. If there is a hand here, there is an external world. Therefore, there is an external world”) is not cogent. Furthermore, it is because of this kind of transmission-failure that bootstrapping arguments designed to produce warrants for very general beliefs, such as “My sense organs are mostly working correctly,” out of specific perceptual beliefs justified by means of occurrent perceptions would not be cogent either.
6 The Extended Rationality View Extended
Finally, in the closing chapter of the book, I explore the possibility of bringing the extended rationality view to bear on all other aspects of epistemic rationality. Namely, in the case of those hinges, like the principle of the uniformity of nature, that allow us to form inductive justifications for generalized empirical propositions. The same goes for the case of “There is a past,” which is a hinge of the diachronic aspect of epistemic rationality, that is the one in which we form (or retain) justifications for propositions about the past based on memory. Moreover, the extended rationality view is extended to the presupposition of the social aspects of epistemic rationality, where we form justified beliefs on the basis of testimony—that is, “People are generally reliable informants.” The extended rationality view applied to the testimonial case therefore opens up an interesting way of resolving the long-lasting debate between a Humean, reductionist conception of testimonial justification and a Reidian, anti-reductionist one. For the assumption that people are generally reliable informants is needed in order to have testimonial justification (beside being told or reading that P, while no defeaters are present). Yet, it cannot be justified either empirically or a priori. For either kind of justification would face quite insurmountable problems. Yet, this does not mean that that assumption lies beyond epistemic rationality. Rather it falls within its remit, because it is itself constitutive of the social dimension of epistemic rationality. Lastly, the extended rationality view it is brought to bear on the issue of the epistemic status of basic rules of inference, such as modus ponens. Here again, it opens up a new avenue to resolve the long-lasting issue between those who think that we can only provide pragmatic warrants for them, or a priori ones based on contentious semantic assumptions.13 According to the extended rationality view, in contrast, they turn out to be constitutive of the deductive aspects of epistemic rationality. They therefore fall within the scope of epistemic rationality, even though they are not warrantable.
I think it is a merit of the proposal put forward in Extended Rationality that it can be extended to various domains, beside the perceptual one. If, in a Carnapian spirit, the interest and plausibility of a philosophical view can be measured by its fruitfulness, among other factors, I think the moderate conception, developed along the lines of the extended rationality view, does well in this respect. For, as mentioned, it allows us to shed new light on issues such as the validity of the principle of closure, the nature of transmission-failure, the status of arguments such as Moore’s, and it can help illuminate the nature of inductive justification, of memory-based and testimonial warrants, and the epistemic status of basic logical laws.
Coliva A. (2013a). “Hinges and Certainty. A Précis of Moore and Wittgenstein: Scepticism, Certainty and Common Sense,” Philosophia. The Philosophical Quarterly of Israel 41: 1–12.
This example can be found in Wright (1985). Another usual example is the Zebra argument—“Here is a zebra” (based on seeing an animal in a zoo that looks like a zebra), “If this is a zebra, it is not a cleverly disguised mule”; “This is not a cleverly disguised mule”—famously put forward by Fred Dretske (1970). I discuss them at length in Extended Rationality (Chs. 1 and 3).
Cf. Wright (2004), examined in Extended Rationality (Ch. 2 and 4).
The attempt to build on that horn of the trilemma would lead to foundationalism. Both Pryor’s (2004) and Wright’s (2004) views can be seen as different ways of defending it. In Pryor we have immediate justification for ordinary empirical beliefs, thanks to perception and in the absence of defeaters, and from them derive a justification for very general propositions such as “There is an external world.” In Wright, in contrast, we have an entitlement—that is, a non-evidential justification—directly for those very general assumptions and, thanks to it and to an appropriate course of experience, a justification for ordinary empirical beliefs.
The attempt to build on this horn of the trilemma would lead to various forms of coherentism, whose fault is that they could give rise to maximally coherent, yet incompatible systems, among which we could make no epistemically sound choice. That is to say, we would have no means to determine which one is the correct one. Or else, we would have to produce locally circular justifications, that is justifications for general propositions like “There is an external world” based on specific propositions, such as “Here is a hand,” which, in their turn, are justified only insofar as we take for granted those very assumptions. In Extended Rationality (Ch. 3) I argue at length why such circular justifications would be no justifications at all.
The details of such a reading are developed differently by Moyal-Sharrock (2004) and Coliva (2010, 2013a, 2013b), but the main thrust is the same.
As always, with Wittgenstein, things are not entirely clear. My own reading, presented in Coliva (2010) and further developed in Coliva (2013a, 2013b), is that it might be possible to distinguish between the content and the role of a sentence. Hence, Wittgenstein’s, so-called, ‘hinge’ propositions would be propositions, being susceptible to truth and falsity at least in a minimal sense, which, however, have been removed from doubt and inquiry. Therefore, they would play a normative role, while retaining a descriptive content.
Following Boghossian (2006), I hold that an epistemic practice (or method) is basic iff it does not presuppose instances of itself and is presupposed by all other epistemic methods. Forming beliefs on the basis of perceptual evidence is one such practice (or method), as is reasoning in accordance with basic laws of inference, such as modus ponens, while both are presupposed in order to form and verify scientific theories. In contrast, to form beliefs about one’s future by casting oracles or by means of horoscopes is not a basic epistemic practice, for it requires both observation and inference. Notice, moreover, that to reason in accordance with modus ponens is basic when it does not concern embedded conditionals. These issues are explored in detail in Extended Rationality (Chs. 4–5).
In my view, it is important that the kind of relativistic challenge considered here be compatible with our experience for it to represent a serious alternative for us, as we are interested in explaining the fundamental traits of the human epistemic condition.
Famously, this is Keith DeRose’s (1995) phrase.