37

Intuitions and the Understanding Introduction In this essay, I take for granted that intuitions play a significant role in providing a priori justification within various domains, including philosophy itself.1 I focus on the question of what intuitions are, and how they might be able to supply the justification at issue. I will develop my account partly by contrasting it with Ernest Sosa’s well-known views on these matters. For concreteness, let us begin by looking at the role of intuitions in thought experiments and, in particular, in the famous case of Twin Earth, as described by Hilary Putnam.2 In putting forward his Twin Earth thought experiment, Putnam was concerned to refute a view that he (appropriately) called the “Received View of Meaning.” He took it to consist in the conjunction of the following two claims: (1) The meaning that a speaker associates with a word is determined by individualistic facts about that speaker, and, (2) The meaning that a speaker associates with a word determines that word’s extension. In order to refute the Received View, Putnam described a possible world in which, in another part of our galaxy, there is a Twin Earth, a planet as much like Earth as possible, except that the liquid that flows through its lakes, rivers, and faucets, although superficially quite similar to water, is not H2O but XYZ. On Earth, we suppose there to be a user of the word ‘water,’ Oscar, who, while not knowing anything about water’s chemical composition, is nevertheless perfectly competent at using it. While on Twin Earth, there is a molecule for molecule duplicate of Oscar’s, Toscar, who also counts as a competent user of ‘water’ by the relevant standards, (p.138) although he, too, does not know anything about the chemical composition of the local liquid.3 Putnam asks whether the extension of Oscar’s word ‘water’ is the same as Toscar’s. He believes that, when we reflect on this question, we will see clearly that, in the scenario he described, Oscar’s word ‘water’ will have H2O in its extension but not XYZ; while Toscar’s word ‘water’ will have XYZ in its extension but not H2O.4 If Putnam were right about this judgment, the Received View, which predicts that the extensions would be the same, would stand refuted. And, since the Received View is enmeshed with a large number of other important and well-entrenched claims—claims involving privileged access, psychological explanation, and the relation between meaning and reference—the implications of that refutation would be dramatic. Accepting Putnam’s judgment, then, threatens to upend many central and seemingly well-supported philosophical theories. Given this fact, his judgment had better be pretty strongly justified if it is to be accepted: we don’t want to rewrite large swathes of the philosophy of mind, the philosophy of language, and epistemology, if Putnam’s judgment is just something that he is inclined to believe, without adequate justification. Clearly, Putnam believes his judgment to be strongly justified; and so does each of us when we arrive at the same intuitive judgment about this particular case. The question is: What could be the justification for this judgment about this highly contrived and specialized case, far removed from any actual scenario that we have previously encountered or thought about? How do we come to know what would be true in such remote and unusual scenarios?5 In the earlier work alluded to above, I argued that we could not hope to explain the source of the justification for this and other thought experimental judgments simply by appeal either to perception (broadly understood to include sensory experience, memory, introspection, and testimony) or to the understanding—our conceptual competence with the relevant concepts. Rather, I argued, we must make appeal to the traditional notion of intuition, understood as a species of intellectual seeming or insight. According to this traditional idea, in reflecting on Putnam’s scenario, and on his question, we have an intuition to the effect that Oscar’s and Toscar’s tokens of ‘water’ have non-overlapping extensions, and it is that intuition that leads us to, and that justifies, the corresponding Twin Earth judgment, and thereby to a rejection of the Received View. (p.139) A crucial feature of this justification, as I have been emphasizing, is that it must be strong enough to trump the justification we previously had for (what is often) a large number of central beliefs that conflict with it. It must be able to justify our tossing out those well-entrenched conflicting beliefs, or, at the very least, to justify our reconsidering them. Any intuition-friendly view of a priori justification confronts three urgent questions: (I) What are intuitions? (II) Precisely what kind of justification do they supply? (III) How exactly do they supply the justification that is claimed for them? Sosa on Intuitions In a series of influential and important essays, Sosa has developed non-skeptical answers to these central questions, answers that may be summarized in the following theses:6 (S-I) An intuition that p is an intellectual seeming that p, which is itself to be understood as consisting in an attraction to assent to p that is based merely on an understanding of p. (S-II) Intuitions supply a foundational and a priori justification for propositions. (S-III) An intuition that p is able to justify the belief that p if and only if it is explained by a competence to tell, on the basis of one’s understanding of p alone, whether p is true. Since I agree with Sosa on (S-II), I will simply assume it in what follows. My focus will be on assessing his picture of what intuitions are (S-I), and of how they justify (S-III). In particular, I will be interested in whether the understanding deserves the role that it is accorded by Sosa’s account.7 The Nature of Intuitions and the Understanding We may break up Sosa’s account of the nature of intuitions, (S-I), into two theses: (S-Ia) Intuitions are intellectual seemings. (S-Ib) Intellectual seemings that p are attractions to assent to p that are based on one’s understanding of p alone. I agree that intuitions are intellectual seemings. But why should we go along with Sosa’s other claim— that intellectual seemings are attractions to assent that are based on one’s understanding of the relevant proposition alone? Why attractions to assent? And: Why based on the understanding alone? (p.140) The view that intellectual seemings are to be understood as a species of attraction to assent is not without motivation. In its absence, intellectual seemings can seem mysterious. For what are they to be? We can agree, of course, that an intellectual seeming is a kind of seeming. And a seeming, we can say, is a conscious mental state that presents a proposition as true. It’s a conscious mental state that presents the world as being so, or perhaps even as being necessarily so. But what makes such a seeming an intellectual seeming?8 The intended contrast is with sensory seemings, seemings that some sense organ or other delivers, and that are individuated by their characteristic phenomenologies. But how are intellectual seemings individuated? Do they, too, have a characteristic phenomenology? This is an area of some controversy. Many philosophers balk at talk of states with a special intellectual phenomenology. Many philosophers report that they can detect no such states in themselves when they contemplate thought experiments and respond with intuitive judgments. But what then is an intellectual seeming to be, if it is not to be picked out by its distinctive phenomenology? In response to this problem, Sosa’s answer can seem quite compelling: an intellectual seeming that p is an attraction to assent to p. This proposal identifies an intellectual seeming with something we appear to have an independent purchase on—namely, assent. And it approximates the pre-doxastic character of a seeming by equating it not with actual assent, and not even with a disposition to assent, but merely with an attraction to assent. I can be attracted to assent to the proposition that the Müller-Lyer lines are unequal in length, even while not being disposed to assent to that proposition (as a result of knowing that they are in fact equal in length). But once we remove the mystery of what intellectual seemings are by equating them with attractions to assent, we face another problem. Intellectual seemings can’t just be any old attractions to assent, as the example used above shows, since such an attraction to assent to p might be based on one’s sensory experience of p; and such an attraction would obviously not be an intellectual seeming. This observation motivates Sosa’s (S-Ib): an intellectual seeming is not any old attraction to assent to p, but one that is based on one’s understanding of p alone. That restriction seems to give us a sufficiently distinctive account of what an intellectual seeming is, one that distinguishes it from a mere sensory seeming, since sensory seemings are attractions to assent that are not based on the understanding alone but are partly based on sensory experience. So, a compelling consideration in favor of an understanding-based view of the nature of intuitions is that it appears to give us a prima facie plausible account of what they are, one that distinguishes them in what appears to be the right way from mere sensory seemings. (p.141) Understanding and A Priori Justification The understanding also figures in Sosa’s account of the justificatory power of intuitions, (S-III). According to Sosa, what explains why an intuition is able to a priori justify a belief that’s based upon it is that the attraction to assent to the relevant proposition derives from a competence to tell, on the basis of the understanding alone, the true from the false, in the domain to which the proposition belongs. There are two questions here to which Sosa may be seen as giving answers: (A) How could intuitions be the source of any sort of justification? (B) How could they be the source specifically of a priori justification? In some sense, there ought not to be too much of a puzzle about why intuitions could be a source of some justification for belief. For intuitions are a kind of seeming, and, surely, seemings must be the source, at the very least, of prima facie justification. Where could rational belief begin if not with how things seem to be? Such justification may be defeasible by other beliefs that are more strongly grounded, but that concession cannot interfere with a seeming’s ability to provide prima facie justification. What I have just said about prima facie justification will be found plausible by those with a generally internalist orientation in the theory of justification. But those philosophers, like Sosa, for whom justification involves the reliability of the mechanism by which belief is fixed, will not be satisfied. They will insist that it is a condition on intuition’s being able to supply justification that it be a reliable source of truth about the relevant domain. And they will want to know how our having an intuitive response to a remote possible world scenario could possibly be a reliable source of truth about the modal facts? How, to return to the Twin Earth example, they might ask, could we reliably know, just by reflecting on the scenario in question, what the extensions of words would have been on Twin Earth? The demand for an explanation of intuition’s reliability about modal facts may be felt even by those who accept a broadly internalist view of justification. Some philosophers may feel that, regardless of whether reliability is a condition on justification, there had better be a scientifically respectable explanation of how a given epistemic faculty could be a reliable source of truths about a particular domain. Either way, we seem to be faced here with an embarrassing question: what’s the informational route whereby our intellectual seemings can be a source of reliable views about abstract or modal truths? The answer can’t be that there’s a causal channel open between abstract reality and our cognitive faculties that explains why we’re reliably attracted to assent to truths about abstracta, on analogy with the way in which sensory perception serves as such a causal channel to concrete reality. But what’s the alternative? Here, too, it looks as though play with the understanding provides a potentially satisfying resolution to these worries. If we have a competence to reliably tell, on the basis (p.142) of our understanding alone, which modal propositions are true and which false, we would have a demystifying answer to the potentially embarrassing question about reliability. The ‘causal channel’ in question would be open only between our cognitive faculties and the information that’s encoded in our concepts; and, presumably, everyone should be comfortable with the existence of that sort of mechanism. Finally, (S-III) promises to give a satisfying answer to question (B) as well. If the competence for reliable modal belief were based upon the understanding alone, that would explain why the justification involved is a priori, since, by relying on the understanding alone, one wouldn’t be relying on sense experience. Rejecting the Link between the Nature of Intuitions and the Understanding I believe that these considerations build a powerful case for deploying the notion of the understanding in an account of intuition in the way that Sosa does. Nevertheless, I believe that such accounts are ultimately mistaken. In what follows I will first try to explain why they are mistaken. I will then turn to defusing the considerations I have adduced in their favor and suggest an alternative picture. Let me start with a discussion of (S-I). According to Sosa’s account, a person’s intuition that p is an attraction to assent to p that is epistemically based on that person’s understanding of p alone. It is a distinctive feature of views that identify intuition with some species or other of assent that an intuition can have an epistemic basis, rather than merely serving as one. A presentational state, such as a visual percept, could serve as an epistemic basis for assent, but could not itself be epistemically evaluated. An attraction to assent, by contrast, in being a species of assent, may be epistemically evaluated and so could itself have an epistemic basis, as opposed simply to being able to serve as one. Sosa tells us very little about basing and even less about what it is to grasp or understand a proposition. This is not surprising: these notions have proven to be very difficult to explicate. But they are, of course, notions that are needed quite generally; and we are entitled to some confidence that each is in good standing, even in the absence of a satisfying account. However, what we are not entitled to be confident about is that when we finally get a satisfyingly general account of grasp, that it will be clear how mere grasp of p could serve as an epistemic basis for assenting (or being attracted to assent) to p. There is one model for this that we (think we sort of) understand reasonably well. And that is when grasp of p consists in grasp of some sort of explicit definition for p, and our basis for assenting to p consists in inferring p from its definition. But this is clearly a very special case—special both in that grasp rarely consists in grasping an explicit definition, and in that ‘basis’ rarely means inferential basis. But if the relation between grasp and assent is not inferential, what else could it be? Do we understand how the relation between grasp and assent could be like the relation (p.143) between a perceptual state and the belief that is based upon it? What notion of grasp would work like that? Rejecting the Link between Justificatory Power of Intuitions and the Understanding Let me turn to the role the understanding is said to play in explaining intuition’s justificatory powers. As we saw, Sosa’s account of this (S-III) claims that an intuition that p is a potential provider of foundational a priori justification for the belief that p if and only if it is an attraction to assent to p that is explained by a competence to tell, on the basis of one’s understanding of p alone, whether p is true. Any account of intuitions that gives a central place to the notion of understanding is likely to seem surprising against the background of the history of analytic philosophy. For, historically, explanations of a priori justification on the basis of the understanding alone—explanations in terms of what I have called “epistemic analyticity”—were thought to be rivals to accounts in terms of intuition.9 Much of the early interest in the notion of analyticity was fueled precisely by a desire to do the epistemology of a priori justification without appeal to intuition.10 One immediate puzzle about Sosa’s view concerns the foundational aspect. How could the justification provided by intuitions be foundational, on this view, when what is providing the justification is itself epistemically evaluable? That flies in the face of what foundational justification is. Since I believe it is correct that intuitional justification is foundational justification, I am committed to finding a way of thinking about what intuitions are that doesn’t equate them with a species of assent, but rather construes them on analogy with the sorts of pre-doxastic presentational states that visual perceptions are.11 A second problem for the account, from my point of view, derives from its reliance on a reliabilist conception of epistemic justification. I do not pretend to have anything new to say about this extensively discussed topic. As far as I’m concerned, the standard objections to such a conception are fatal. A reliabilist conception of justification, in my view, loses the topic.12 A final objection to (S-III) concerns the reach of such a theory of intuitional justification. Sosa makes it clear that he thinks intuitions supply all (or most) of our a priori knowledge. But if intuitions are constrained to justify only what can be reliably settled (p.144) on the basis of one’s understanding, then there will be lots of a priori knowledge that intuitions won’t be able to explain. As a result, the story in terms of intuitions will be, at best, radically incomplete. What are some of the propositions that look resistant to explanation by the understanding alone? I will mention two examples, although there are likely many others. The first is the famous color exclusion case. We can establish on the basis of a thought experiment that nothing can be red and green all over at the same time. But how could this be accounted for purely conceptually? Knowledge of the exclusion could not be built into one’s grasp of the concept red, because one could have red without having green. Mutatis mutandis, it could not reside in one’s grasp of the concept green. Hence, at best it could reside only in the joint possession of red and green. But what could the joint possession of red and green be if not the simultaneous possession of each? So it looks as though, in addition to grasping the ingredient concepts, some further act of insight is required if someone is to be able to get justification for the color exclusion claim from whatever information is encoded in our color concepts. A second, and arguably more important, case is provided by normative judgments, whether these are the judgments of morality or those of rationality. In a Trolley Problem, for example, one might conclude that it is not morally permissible to throw the fat man off the bridge in order to save the lives of the five innocent victims who are trapped below. But it is not plausible to say that this judgment derives solely from our understanding of the ingredient concepts. Hume (1978: book 2, part 3, section 3) famously said: ’Tis not contrary to reason to prefer the destruction of the whole world to the scratching of my finger. Perhaps one wouldn’t want to go as far as Hume. However, the kernel of truth in Hume’s memorable remark is that it is not contrary to our understanding of the ingredient concepts to prefer the destruction of the whole world to the scratching of my finger. Whatever deficit one would be manifesting in having such a perverse preference, it wouldn’t be a lack of understanding of what one’s preferences actually are. This argument is, in a sense, an application of Moore’s Open Question Argument, an argument that is widely accepted as showing that substantive moral conclusions are simply not built into our concepts of right and wrong. If someone were to insist that, according to their concept of ‘good’, the good always involves maximizing happiness, it seems that it is always coherent to ask: But is that the correct concept of good, the one that delivers genuinely normative results? Notice the contrast here with the concept square. If someone said: According to my concept square, a square always has four sides, it wouldn’t make any sense to ask: But is that the correct concept square? (p.145) By contrast, it seems always to make sense to ask about the correctness of a given normative concept.13 If this is right, and yet moral judgments are knowable a priori, this shows that the story in terms of understanding can’t be the whole story. There must be more to a priori justification of normative claims than is provided by the understanding account.