Brains, Buddhas, and Believing

We saw in the last chapter that there are important considerations that might recommend the project of “naturalizing” the intentionality of the mental—considerations centering on the problem of mental causation and on commitments, taken as axiomatic, regarding the understanding of causation that must figure in addressing that problem. In this chapter, I want to make a case for thinking that reason itself is centrally implicated in the nature of intentionality—and that it’s chiefly insofar as reasoning will not admit of efficient-causal explanation that intentionality essentially resists explanations such as proposed by Fodor and Dennett and their philosophical fellow travelers.

The avowedly eliminative physicalist Paul Churchland has recognized that in recent debate it has particularly been “the realm of the intentional, the realm of the propositional attitude, that is most commonly held up as being both irreducible to and ineliminable in favor of anything from within a materialist framework.” On Churchland’s account, this reflects a retreat on the part of critics of the properly scientific sort of program he commends; the “locus of opposition” to physicalism has thus shifted, he says, insofar as arguments appealing to constitutively subjective phenomena—phenomena like emotions, qualia, and “raw feels,” formerly taken to represent “the principal stumbling blocks for the materialist program”—have been shown to fail (1981, 67–68).

Churchland takes arguments from intentionality, then, to represent a not very promising last line of defense for critics of physicalism. While there are philosophers on both sides of the debate who may not share Churchland’s confidence that proponents of physicalism have finally dispatched the other line of argument,¹ my interest is in showing that the argument from intentionality has rather more going for it than Churchland grants. We can begin to develop a case for this by asking why it makes sense, in the first place, that “the realm of the intentional” can be glossed, with Churchland, as “the realm of the propositional attitude.” Why is it reasonable to think that the “aboutness” that characterizes the mental is paradigmatically reflected in the ways—believing, doubting, affirming—in which one might relate to propositions?

I first mentioned the category of intentionality in my introduction, with reference to Franz Brentano, whose recovery of this term of art from medieval scholastics made it an idea of central concern in twentieth-century philosophy (continental and Anglo-American alike).² Typically invoked in discussions of intentionality is a canonical passage from Brentano: “Every mental phenomenon is characterized by what the Scholastics of the Middle Ages called the intentional (or mental) inexistence of an object, and what we might call, though not wholly unambiguously, reference to a content, direction toward an object (which is not to be understood here as meaning a thing), or immanent objectivity” (1973, 88). Brentano explains that “inexistence” here means essentially the same thing as “immanent objectivity”—that, in other words, it is characteristic of the mental somehow to contain its content, which therefore “exists” in thought: “We can… define mental phenomena by saying that they are those phenomena which contain an object intentionally within themselves” (1973, 89).

This formulation reflects something of Brentano’s basically representationalist epistemology—a point we can appreciate by noticing that in the foregoing, he distinguishes “mental phenomena” not from physical objects per se but from “physical phenomena.” In considering, that is, the distinguishing characteristics of the “mental” and the “physical,” Brentano is in both cases considering only (what he thinks is autonomously intelligible) what appears to us in thought; “physical phenomena” are on this account mental representations of “physical” things, and these representations—as contra the mental acts that are aware of these as one’s representations—are distinguished from “mental” ones only in that they do not themselves “contain” further objects. (A tree-representation appearing in thought is not itself about anything further.) While intentionality, on this account, is thus the criterion of mental phenomena, it’s important to appreciate that for Brentano, it’s really phenomena all the way down. Insofar as his account of intentionality thus turns our attention inward, Brentano’s account raises all manner of difficult questions about the ontological status of the objects of awareness (note Brentano’s insistence that “object” here is “not to be understood here as meaning a thing”), and about how we are to understand the intentionality of thoughts whose contents (whose “in-existent objects”) are nonexistent objects. (What, for example, is “intentionally contained” in thought when one thinks of unicorns?)³

What I here want to emphasize is just that Brentano’s introduction of the idea that mental events are constitutively contentful does not obviously entail the idea that, as Wilfrid Sellars argues, “the categories of intentionality are, at bottom, semantical categories pertaining to overt verbal performances” (1956, 94). Nothing about Brentano’s canonical formulation obviously suggests that understanding intentionality might involve understanding such things as language and reasoning. How, then, might it come to be thought that Brentano’s “reference to a content” involves—perhaps essentially—the peculiar kind of “aboutness” that is exemplified by language? What might it be about thought’s representing its contents that essentially involves reasoning?

Roderick Chisholm—himself centrally involved in introducing Brentano’s thought (and other themes from European phenomenology) to the mainstream of Anglo-American analytic philosophy ⁴—differed from Sellars precisely over the question of whether we are to understand “the intentional character of believing and of other psychological attitudes by reference to certain features of language,” or conversely (Chisholm and Sellars 1957, 215).⁵ Despite, however, Chisholm’s resistance, and despite its not being obvious from Brentano’s introduction of the category that intentionality essentially involves semantics, something like Sellars’s view is reflected in a use of the word “intentional,” common in recent Anglo-American philosophy, to characterize not just contentful mental acts but the level of description at which we speak of anyone’s meaning something. While there are many ways to tell the story of intentionality and its role in twentieth-century philosophy,⁶ I am here aiming for an account that makes sense not only of Brentano’s idea of thought as contentful but also of this contemporary use of the qualifier “intentional” to characterize (say) any instance of someone’s saying something (particularly insofar as that essentially differs from their simply producing acoustic disturbances)—an account, indeed, that shows why it is as we should expect that these ideas should go closely together.

Sellars himself (Robert Brandom reports) said in this regard that “the aim of his work as a whole was to begin moving analytic philosophy from its humean phase into a kantian one” (Brandom 2000, 32).⁷ Concisely suggesting what’s involved in thus ushering in a Kantian phase, Sellars said that “Kant insists on the irreducibility of judgmental to non-judgmental content. This, indeed, was the very heart of his insight” (1967, 61). This characterization of Kant’s project reflects the idea that Kant took the basic units of experience to be not (as for empiricists) discrete, causally describable sensations, but judgments such as are not themselves explicable in terms of the empiricists’ sensations. (Sensations by themselves, Sellars said, “are no more epistemic in character than are trees or tables” [1991, 336].) Our task, then, is to see what sense it makes to think about intentionality from the point of view of the characteristically Kantian emphasis on judgment; insofar as this turns out to be a helpful way to think about intentionality, we will see that it’s not inappropriate to say that Kant’s transcendental unity of apperception represents perhaps the most significant forerunner to the contemporary idea that people typically take Brentano to have introduced.

The “transcendental unity of apperception”—also referred to by Kant as “pure apperception,” “original apperception,” and the “synthetic original unity of apperception”—is central to Kant’s project.⁸ This represents the basis of Kant’s “Transcendental Deduction of the Pure Concepts of the Understanding”; it is, indeed, the transcendental unity of apperception from which these concepts are deduced. These represent, in turn, the basis of Kant’s claim to have developed an account of the objective validity of knowledge. While Kant’s “Copernican Revolution” famously turned attention away from objects in the world and toward the knowing subject—emphasizing that we cannot be cognitively acquainted with things-in-themselves, only with things-as-they-appear-to-us—it was because he thought he could show that how things appear to us is necessarily structured in certain ways that Kant could nevertheless claim to offer an account of the objectivity of knowledge. His claim is that the conceptual order he deduces represents a condition of the possibility of experience’s being about a world at all—that our “having the world in view” (as John McDowell says)⁹ is possible only insofar as experience is conceptually structured in the ways Kant aimed to show.

The transcendental deduction of the conceptual order begins, then, from what Kant takes as an essentially basic fact about experience: that it consists in the imposition of some perspectival unity on the irreducibly plural data of perception or (Kant’s term) “intuition.” Experience essentially consists in the ordering or “synthesis” of a sensory “manifold” such as might be characterized (following John McDowell) in terms of causally describable “impingements by the world on a possessor of sensory capacities” (1996, xv). Among the ideas here is that to be the subject of any experience just is to have a unifying perspective on the world as seen from some perspective. Despite the centrality of this idea for his project, though, Kant had a hard time expressing clearly just what the transcendental unity of apperception is, and the idea is developed rather differently in the first and second editions of the Critique of Pure Reason; indeed, this is among the points on which the two editions significantly differ, and we can usefully get a handle on some of the issues in play by scouting the different ways Kant develops it.

In the first edition, Kant develops the idea in response to David Hume, particularly insofar as Hume had argued that there is nothing more to subjectivity than a “bundle or collection of different perceptions, which succeed each other with an inconceivable rapidity, and are in a perpetual flux and movement” (Hume 1739, 252). What is real, Hume had thus urged, is only episodic cognitions, and there is no enduring subject that “has” these fleeting states; insofar, rather, as episodic perceptions exemplify “our notion of diversity, it can only be by mistake we ascribe to [the series] an identity” (1739, 255). Consisting, then, in an always-vanishing subjective “now,” awareness represents the paradigm case of something essentially momentary; to that extent, we are misled by the phenomenological continuity that characterizes awareness. The phenomenologically continuous character of experience, Hume argued, is explicable in terms of memory and recognition: “as the relation of parts, which leads us into this mistake, is really nothing but a quality, which produces an association of ideas, and an easy transition of the imagination from one to another, it can only be from the resemblance, which this act of the mind bears to that, by which we contemplate one continu’d object, that the error arises” (1739, 255; emphasis added). It’s only, then, insofar as successive mental states “resemble” one another that they are experienced as states of an enduring subject; in fact, what is real (Hume says in terms that Dharmakīrti would approve) is only existents “such as consist of a succession of related objects” (1739, 255).

Kant first elaborates the idea of the transcendental unity of apperception by way of a compelling rejoinder to this; how, he asks, could we recognize two moments or representations as similar without presupposing just the continuity that is supposed to be explained by this recognition? How, in particular, are we to understand the unified perspective from which it could alone make sense to say that any two such moments are thus “taken” as similar? “Without consciousness that that which we think is the very same as what we thought a moment before,” Kant says, “all reproduction in the series of representations would be in vain.” Insofar as mental events are considered as discrete and momentary, he explains, none could ever represent a unified perspective on a series thereof; each such event “would be a new representation,” and its content could “never constitute a whole”—could never constitute a unified perspective on its predecessors (A103). If, that is, mental events are genuinely discrete, no one of them could so much as take its own predecessors as phenomenally similar; to the extent that there is no single “act” somehow comprising others, experience would consist in a series of utterly particular, always unfamiliar cognitive events.

Among the questions this raises, though, is what performs the synthesis that thus characterizes experience, what the agent of this act is. To press this question is effectively to ask for an empirical locus of the action of “synthesis”; but that such a thing could be identified seems to be just what the characterization of this as “transcendental” is meant to rule out. P. F. Strawson notes in this regard that many misgivings about Kant’s account thus involve the thought that insofar as we ascribe states to a subject, the latter must be “an intuitable object for which there exist empirically applicable criteria of identity” (1966, 107). If that’s right, though, Kant’s point would look very much like the Cartesian sort of argument he so unambiguously eschews. The representation “I think” is not, Kant emphasized in this regard, to be understood as anything like the content of a perceptual encounter with one’s “self.” Indeed, the distinctiveness of Kant’s constitutively transcendental approach is clear from his critique of Descartes, against whom he argued that “since the proposition ‘I think’ (taken problematically) contains the form of each and every judgment of the understanding and accompanies all categories as their vehicle, it is evident that the inferences from it admit only of a transcendental employment of the understanding” (A348; emphasis added). The “I think” of transcendental apperception, then, is nothing perceptible; it is, rather, something like the logical or conceptual limit of any conceivable act, even one of inquiring about the empirical self. This point remains obscure, though, as long as Strawson’s worry remains in play.

Among other things, Kant here more clearly makes the strictly formal point that any experience counts as such only insofar as it is at least possible for it to be expressed in terms of the content of some subject’s judgment. Note that this is very different from saying (what is surely false) that all experiences are expressed as judgments; the point is only that a subject could thus attend to her own experience. To emphasize the formal character of this point is, among other things, to say that it does not warrant any inferences about what (if any) kind of existent the referent of this “I” must be; it’s not, then, because it gets us anything like a Cartesian “thinking substance” that Kant’s argument will provide the basis for a cogent critique of physicalism. (The argument here is not, to that extent, necessarily incompatible with the Buddhist doctrine of selflessness, depending on how that is understood.) The point, rather, is just that “experience” is essentially such that it’s only with reference to a point of view that we can make sense of anything’s being an experience; the idea of an experience of, say, feeling warm which is such that it’s unclear to the subject thereof whose experience it is would seem to be unintelligible.

Among the intuitions here is that only the kind of content that can thus be individuated by “that”-clauses could count as constituting genuinely objective knowledge; mental content that’s not so describable—the kind of content we would have, for example, given only bare sensings produced by impingements on our senses—would consist simply in subjectively known mental representations (would consist in a subject’s simply having something appear to her). “The difference between truth and dream,” Kant says to similar effect in the Prolegomena, “is not decided through the quality of the representations that are referred to objects, for they are the same in both, but through their connection according to the rules that determine the connection of representations in the concept of an object” (1783, 42). Any account that grounds our knowledge simply in the images that are causally produced by sensory contact would thus take something eminently subjective as the basis of supposedly objective knowledge; against that, the argument here advances the thought that only conceptually structured content could be intersubjectively available.

Just insofar, then, as experiences are constitutively had from some point of view, they are characterized by the same kind of unity that essentially characterizes propositions—by the fact that the various inputs to experience at any moment are commonly brought together as the unified states of affairs that are picked out by “that”-clauses. The point is that what is individuated by any such thought is complex states of affairs under some description, something of the world as characterized somehow or another. Insofar as we thus experience things always under some description, it makes sense to suppose that the rules that describe our bringing various inputs (a sensible “manifold”) under a unifying perspective are something like the same rules that describe the structure of propositions—that experiential inputs are brought under a point of view (are “synthesized in a consciousness”) in something like the same way that the parts of well-formed propositions are related to one another.¹⁰ What any experience is of, this is all to say, must be the same kinds of things that also figure in the contentful sentences that it’s the task of logic and semantics to systematize; “our original idea-contents,” as Sellars puts this, “turn out to be the sort of thing which serve as the senses of linguistic expressions” (1967, 66).

We should not here be misled by the fact that Kant’s paradigmatic expression involves the verb “think”; his point applies just as well to all of the other propositional attitudes, too. The argument is not, that is, that all experience consists in thinking—it is that we can only experience the same kinds of things we can in principle mean. The claim is not that all experiences explicitly involve judgments, understood as the conclusions to inferences or arguments; it is, rather, that experience is only contentful, is only meaningfully about anything, insofar as it involves the same kinds of things (conceptually structured things) that also figure in whatever we can say about it.¹¹

The transcendental unity of apperception—the idea that experience essentially consists in some subject’s unifying perspective on the world, which therefore always appears to a subject under some description—represents, then, the basis for an account according to which intentionality is a constitutively semantic affair. The line of argument here sketched is succinctly expressed by John McDowell: “Experiences in which the world is disclosed,” he says, “are apperceptive. Perception discloses the world only to a subject capable of the ‘I think’ that expresses apperception.” That is, there is no perceptual experience that is not someone’s perceptual experience, no experience that is such that its subject could not attend to it, under any propositional attitude, as his experience. And “if an experience is world-disclosing, which implies that it is categorially unified”—which implies, that is, that it is synthesized in the way that goes with its being an experience had from a unitary point of view—then “all its content is present in a form in which… it is suitable to constitute contents of conceptual capacities” (2009, 318–319). The extent, that is, to which we can say anything about an experience—as, for example, when we adduce it as affording reasons for belief or action—is just the extent to which its content always already occupies Sellars’s “logical space of reasons.”

In a hitherto neglected part of the long passage above, Kant says that the representation “I think” is “an act of spontaneity, i.e., it cannot be regarded as belonging to sensibility.”¹² He thus contrasts conceptual capacities—those in virtue of which a subject can take it that anything is the case—with perceptual “sensibility,” which instead involves a faculty of “receptivity.” Insofar as our sensory capacities passively “receive” the environment’s impingements upon our bodies, we can be said to be “affected” by perceptual objects, where that represents a causal relation.¹³ The conceptual synthesis of the sensory manifold that thus impinges upon us—the bringing of environmental stimuli under the kind of description that can be the content of a “that”-clause—is essentially different from that.

Among the points advanced by Kant’s reference to “spontaneity” here is that the unified character of experiential content is not something that could itself be given to us by the irreducibly diverse objects in any sensory field; it’s not, for example, determined by any particular cluster of vertical objects whether we see them as trees, or whether instead we see them as elms or as a forest.¹⁴ The description under which these are experienced is not itself part of what is caused by these objects’ “affecting” our senses; this represents, rather, how a subject takes them to be. Just insofar as what is thus present to thought is necessarily a complex state of affairs under some description, it cannot be thought that the intentionality of the mental will admit of an exhaustively causal description—this is the point Kant means to identify with his characterization of our conceptual capacities as a faculty of spontaneity.

There are, to be sure, problems with this characterization. In the Prolegomena, for example, Kant says that “if an appearance is given to us, we are still completely free as to how we want to judge things from it” (1783, 42). But surely it’s not right that we are “completely free” in this; we rarely (if ever) decide what to make of experienced states of affairs, and a characteristically all-or-nothing notion of freedom—one such as figures, too, in Kant’s moral philosophy—here represents an unhelpful characterization of Kantian spontaneity. If, however, Kant is thus “prone to suppose genuine spontaneity would have to be wholly unconstrained” (McDowell 1996, 96), that should not obscure the real insight, which has to do with the very idea of our being held responsible for our judgments and beliefs. It is in virtue of the “spontaneity” of the logical space of reasons that it’s so much as intelligible to us that we could be right or wrong in ever thinking as we do and that the activity of justifying our beliefs could thus make sense.

Here, recall (from chapter 2) Boghossian’s reflections on the normativity that essentially characterizes the possibility of our meaning anything by expressions. The innumerable truths entailed by any utterance, he said, are “truths about how I ought to apply the expressions, if I am to apply it in accord with its meaning, not truths about how I will apply it” (1989, 509)—a fact that cannot be accounted for by any empirical survey of the dispositions of language-users, since “to be disposed to use a word in a certain way implies at most that one will, not that one should (one can have dispositions to use words incorrectly)” (1989, 513).¹⁵ It is, Boghossian thus argued, the essentially normative character of linguistic understanding that most compellingly resists causal explanation. There is a case to be made for thinking that Kantian “spontaneity” essentially involves the same recognition—the recognition, as McDowell puts this point, that “we need to see intuitions as standing in rational relations to what we should think, not just in causal relations to what we do think. Otherwise the very idea of what we think goes missing” (1996, 68; emphasis added).

Robert Pippin offers a helpful interpretation along these lines. On Pippin’s reading, what is reflected in Kant’s discussions of the transcendental unity of apperception is the view that “it is criterial that we must be implicitly aware of our having ‘taken’ the world to be such and such, and thereby of its possibly not being such and such” (1987, 462). The argument that the content of any experience must be such as could be expressed by saying “I think (hope, believe, doubt…) that x” advances, then, not only the idea that the content of experience consists in the conceptually shaped states of affairs that can be individuated by “that”-clauses; it also advances the thought that experience essentially involves the subject’s awareness of herself as so taking things—the awareness of herself, that is, as actively appropriating something that may, to that extent, be mistaken. Pippin thus takes Kant to argue that for any of my experiences to inform an epistemically contentful claim, “I must take up the contents of intuition and the mental states that can be said to be produced by such intuitions, and make such a claim. It cannot both be said simply to occur in a relation of existential dependence on other mental states, and to be ‘objective’ (possibly true or false) representing of mind, unless I so ‘take’ it” (1987, 468).

On this reading, it is any subject’s awareness of himself as thus taking things to be somehow or other that makes intelligible the very idea that we might be asked to justify (by giving reasons for) any of the judgments we so form. It makes sense, that is, that the rightness of any judgment can be challenged only to the extent that it’s already integral to the experience of judging that we might mis-take things. That we necessarily experience things always under some description means, then, that we are answerable for our descriptions. The “spontaneity” of our conceptual capacities reflects, therefore, the essentially normative character of meaning one thing (and not another) by how we take anything to be—the fact, among others, that just what it is in virtue of which anyone could be thought right or wrong in describing any state of affairs is not itself something that will finally admit of causal explanation, any more than what it is to know what a sentence means will.

This, finally, gets at the point Sellars influentially expressed by saying that “in characterizing an episode or a state as that of knowing, we are not giving an empirical description of that episode or state; we are placing it in the logical space of reasons, of justifying and being able to justify what one says” (1956, 76). To characterize someone as knowing anything is to characterize her not only as rightly taking things as thus-and-so but as responsible for her so taking things—and when anyone is called on to justify her so taking things, we understand that what is demanded is not an account of how she happens to have come to believe, but reasons for so believing. Giving reasons, we have now seen Kant argue, essentially involves “spontaneity,” insofar as the constitutively normative dimension of this activity cannot itself be explained in causal terms; this, finally, is why the foregoing arguments are significant for a cogent critique of physicalism.

We now have in place a conception of intentionality as essentially semantic, such that thought’s being about its contents is appropriately understood on the model of linguistic items as about what they mean. This is why it makes sense to say, with many contemporary philosophers, that an “intentional” level of description is exemplified whenever we talk of understanding what somebody means by what they say. On Brentano’s canonical formulation of intentionality as the hallmark of the mental, we were encouraged to imagine the contentful character of thought on the model of a mental space that contains phenomenal representations, which are thus to be understood as having only “immanent objectivity”—and of which it will therefore always be possible to ask whether they can be known to represent anything “outside” of mental space. On my account, in contrast, thoughts have content in something like the same way that sentences expressing judgments do—and whatever we say about a sentence’s being about what it means, it’s clear at least that this will not admit of efficient-causal explanation.

The question now before us is straightforward: how could this account of intentionality possibly be right? To the extent that, with philosophers like Jerry Fodor, we take the problem of mental causation to be of paramount concern—and to the extent that we take it as axiomatic that only an account involving efficient, local causation could count as addressing that problem—the very fact that the view here proposed will not admit of causal description makes it seem a reductio ad absurdum of the view on offer. Just insofar as our understanding of intentionality is such that it will not admit of a certain kind of scientific explanation, it can reasonably be thought that we must be wrong to suppose either that the conceptual character of experience really is integral to it, or that it really will not admit of causal explanation.

I want to develop this line of argument chiefly in conversation with John McDowell, who takes his bearings from Kant (typically by way of Wilfrid Sellars). Among my reasons for so proceeding is that I have for some time been struck by the purchase McDowell’s thought affords both with respect to the thought of Dharmakīrti and with respect to contemporary proponents of the project of “naturalizing” intentionality—yet I am frustrated by McDowell’s avowedly “therapeutic” stance, which obscures the extent to which his commitments are mutually exclusive of such alternatives. Typically representing his work not as advancing a “constructive” philosophical project but as “exorcizing” certain philosophical “anxieties”—as showing, he says, “how we need not seem obliged to set about answering the questions that express the anxieties” (1996, xx)—McDowell urges a “rejection of the traditional predicament, not an attempt to respond to it” (1996, 112). On my reading, though, McDowell consistently advances a straightforward transcendental argument, among the conclusions of which is that physicalist accounts in philosophy of mind (as I here understand those) cannot be right.¹⁶

Insofar as his account thus precludes particular alternatives, it must be acknowledged, notwithstanding his frequent protests to the contrary, that McDowell’s arguments entail essentially metaphysical commitments. In thus trying to make explicit what I take McDowell to be committed to, I am offering a reconstruction, from a large corpus of works that itself reflects a developing perspective (which is another way to say that McDowell can be very hard to pin down), that will not always be consistent with all of the works cited or with McDowell’s own stated aims. Not only, though, will this way of developing the argument prove to be helpful when we try, in chapter 6, to characterize Mādhyamika objections to an understanding of the Buddhist project such as Dharmakīrti’s, but it also seems to me worth the effort to gain some perspective on the thought of this recently influential philosopher.

Let’s begin, as McDowell often does, with reference to Wilfrid Sellars’s influential critique of the “myth of the given,” as elaborated in Sellars’s “Empiricism and the Philosophy of Mind.” Sellars here aimed to argue against the kind of “givenness” that consists most basically in the view that “epistemic facts can be analyzed without remainder… into non-epistemic facts” (1956, 19)—in the view, that is, that our experience is finally constrained by something (the world as “given” to us in sensory impingements) that is not itself within the logical space of reasons. McDowell characterizes the kind of view against which Sellars thus directs his critique as involving “highest common factor” arguments.¹⁷ Such arguments exploit the fact (if it is one) that an illusory cognition can be phenomenologically indistinguishable from a veridical one of the same thing. That it is possible to have mistaken cognitions that seem the same as veridical ones is taken, by proponents of such arguments, to recommend the conclusion that there must therefore be some type-identical kind of thing—a mental representation—that is commonly present to both cognitions, a “highest common factor” in virtue of which the illusory and the veridical cognition can commonly seem to be of the same thing.¹⁸

It’s clear from this characterization that the kind of view targeted by Sellars dovetails with what we have seen Fodor characterize as a “Representational Theory of Mind”; recalling, then, Fodor’s “methodological solipsism,” we can understand highest common factor arguments as recommending the view that cognition is autonomously intelligible—that it is not, that is, constitutive of cognition for it to have what McDowell typically refers to as “objective purport.” The direction of explanation ought therefore to be, on this kind of view, from seeming to being. Sellars expresses this as the claim that “looks-F talk, with which it is possible to form a class of statements about which subjects are incorrigible”—incorrigible since one can doubt whether things are as represented in awareness but cannot doubt that that is how it seems—“is a foundation of knowledge, and so must be prior in this sense to is-F talk, with which it is possible to express only corrigible, inferred beliefs” (1956, 35). That cognition is of a world is not, on such a view, among the things we can be certain of; only that it seems so is indubitably known. This is, Hilary Putnam says, effectively a line of argument for the view (“almost universally” held) that experience is essentially “something ‘inside’ us, something in a private mental theater (identified by materialists with the brain, of course)” (2002, 177).

Among Sellars’s arguments to this effect is one that invokes the essentially normative character of ordinary linguistic usage regarding such things—a normative character, he argues, that is only intelligible given that “the concept of looking green, the ability to recognize that something looks green, presupposes the concept of being green” (1956, 43). This is evident from the fact that “looks” locutions (“that looks green to me”) do not concern essentially different content than “is” locutions. Thus, to say that something seems thus-and-so is to relate tentatively to some of affairs; to say, instead, that one sees that thus-and-so is not to introduce a proposition concerning an altogether different state of affairs—it is to characterize the same one as “making an assertion or claim, and… to endorse that claim” (1956, 39).¹⁹ The relevant sense of thus endorsing a claim is that this involves taking the content thereof to be “in the logical space of reasons, of justifying and being able to justify what one says”—and this just is to take the content of the claim as rationally relatable to all manner of other claims. Sellars can thus conclude this line of thought in a way that echoes Kant’s transcendental deduction of the categories: “one can have the concept of green only by having a whole battery of concepts of which it is one element.” Indeed, he says, “one has no concept pertaining to the observable properties of physical objects in Space and Time unless one has them all” (1956, 44–45).

The argument here, McDowell says, is generally to the effect that “reality is prior, in the order of understanding, to appearance” (1998, 410)—that, inter alia, experience constitutively has objective purport; experience is intelligible only as being of a world. Advancing this argument, McDowell takes his bearings from a basic disjunction: perceptual appearances, he argues, must be understood either as “objective states of affairs making themselves manifest to subjects, or [as] situations in which it is as if an objective state of affairs is making itself manifest to a subject, although that is not how things are” (2009, 231).²⁰ The relevant disjunction, then, is between views on which cognition is essentially of a world, and views according to which it is so much as intelligible that it might not really be so. Chief among the reasons for thinking these disjunctive, McDowell takes Sellars to have shown, is that there is no way to get from the second of these to the first; if, that is, it’s so much as intelligible that cognition merely seems to have objective purport (“although that is not how things are”), the entailed skeptical problems undermine any understanding of the thought that it does.

This disjunction represents the basis for the kind of transcendental argument that can be encountered throughout McDowell’s works—one to the effect that insofar as there is no sure way from things’ seeming some way to a subject to their really being so, it must be a condition of the possibility of our knowing the latter that cognition is always already of a world. Of course, Buddhists such as Dignāga and Dharmakīrti mean (we will see in chapter 5) to challenge precisely the idea that we are right about what McDowell characterizes as the “epistemic position we are manifestly in” (1998a, 344); their Buddhist project is predicated on the thought that awareness is not really of what it seems to be of. McDowell might be said, in this regard, to argue only that if one would take cognition really to concern a world, it must do so intrinsically—that cognition’s intrinsically being of a world can only be denied at the cost of denying that it could be of a world at all. His point, though, is stronger than this conditional one; he takes Sellars to have shown that we can make sense even of the phenomenological facts (even of cognition’s so much as seeming as it does) only given the rightness of his picture. Thus, the problem with the “highest common factor” conception is that on such a view, “appearances can never yield more, in the way of warrant for belief, than do those appearances in which it merely seems that one, say, sees that things are thus and so”; on McDowell’s reading of the “Sellarsian transcendental argument,” however, “that thought undermines its own entitlement to the very idea of appearances” (2009, 231). We can, in other words, find it intelligible that experience has any content at all only to the extent that it intrinsically has objective purport.

McDowell concisely represents the whole line of argument in terms of “Kant’s advance over Hume,” insofar as Hume takes from his empiricist predecessors “a conception according to which no experience is in its very nature—intrinsically—an encounter with objects” (1998a, 344). Hume was heir, that is, to the Lockean idea that insofar as we can always doubt whether experience is really of what it seems to be of, we ought therefore to derive our account of knowledge simply from what seems to be the case. Hume drew skeptical conclusions from this, arguing that cognition’s seeming to be of something could never finally provide assurance that it is. On McDowell’s reading, Kant credits Hume’s argument but builds on it; Kant’s advance consists in recognizing that “since there is no rationally satisfactory route from experiences, conceived as, in general, less than encounters with objects—glimpses of objective reality—to the epistemic position we are manifestly in, experiences must be intrinsically encounters with objects” (1998a, 344). Crediting the argument that we could never know, of experiences that are intelligible just in terms of how things appear to a subject, that they really afford a perspective on an objective world, Kant effectively reversed the argument; taking the Humean argument as a reductio ad absurdum with regard to the view of knowledge Hume presupposes, Kant concludes that it cannot be right that experience is intelligible simply in terms of what appears to its subject.²¹ It is, rather, a condition of the possibility of our experience’s so much as seeming to be of an external world that it already be intrinsically so.

All of this, for McDowell, recommends the conclusion that our perceptual awareness can be epistemically contentful only to the extent that our conceptual capacities are always already in play therein. An experience, that is, can only be “world-disclosing”—which (recall) “implies that it is categorially unified” such that it is an experience whose content is that such and such is the case—to the extent that “all its content is present in a form in which… it is suitable to constitute contents of conceptual capacities” (2009, 319). Just as we should conclude from the impossibility of inferences from seeming to being that we cannot take the former as foundational, so, too, the fact that there is no way to get from essentially nonepistemic facts to epistemically contentful experience—from, for example, causally describable impingements upon our sensory apparatus to the kinds of conceptual items that figure in the logical space of reasons—means that it cannot be right to think the latter must be explicable in terms of the former. It must be the case, rather, that we always already experience things (even perceptually) only as some way or another, only under some description such as could in principle figure (whether as premise or conclusion) in the justification of further belief or action.

McDowell has said he takes “transcendental” to characterize “any philosophical thinking whose aim is that there not be a mystery in the very idea of objective purport” (1998c, 365). He thus suggests that the foregoing arguments show that we are mistaken to think there is any problem concerning the relation of “mind and world”—or rather, that any account on which this relation is problematic is, ipso facto, to be eschewed. Notwithstanding McDowell’s confidence, the idea that conceptual capacities are necessarily in play even in perception can reasonably be taken as quite mysterious. Considerations centering on the problem of mental causation can be taken to recommend just the view—viz., that “epistemic facts can be analyzed without remainder into non-epistemic facts”—that Sellars targeted as a “radical mistake” (1956, 19). It seems reasonable, that is, to think that epistemic facts must admit of analysis in terms of such essentially nonepistemic facts as brain events if we are to address the problem of mental causation. As long as this concern remains in play, it seems to me that, whatever their cogency, the foregoing arguments do not as decisively dispel the mysteriousness of the mental as McDowell typically seems to suppose.

McDowell can be taken to invoke the Aristotelian idea of second nature as addressing precisely this worry. Recurrent particularly throughout his Mind and World, “second nature” generally represents the idea that “human life, our natural way of being, is already shaped by meaning” (1996, 95)—the idea that it’s just our nature for humans to “acquire a second nature in part by being initiated into conceptual capacities, whose interrelations belong in the logical space of reasons” (1996, xx), and for our cognitive relation to the world therefore to involve conceptual capacities. If we but remembered this, McDowell suggests, we would see “that operations of nature can include circumstances whose descriptions place them in the logical space of reasons, sui generis though that logical space is” (1996, xx). Faced, then, with the kinds of philosophical demands ventured by those who are preoccupied with the problem of mental causation, “We need not connect this natural history to nature as the realm of law any more tightly than by simply affirming our right to the notion of second nature” (1996, 95; emphasis added). It’s just as “natural” for our mental lives to be constitutively semantic as it is for the central nervous system to work as it does.

It seems clear, though, that it does not address the issues typically raised by Fodor et al. simply to help ourselves to this idea of “naturalness”; this seems, indeed, to represent a question-begging assertion that there just isn’t a problem here. To be sure, McDowell does not propose the idea of “second nature” as solving the kinds of problems posed by cognitive-scientific philosophers of mind; it is, in fact, particularly at this juncture that McDowell is most apt to characterize his as the kind of Wittgensteinian “therapeutic” exercise that aims to show the demand for a “solution” to be misguided.²² That McDowell himself has nevertheless elaborated a position with essentially metaphysical entailments is most clear, I think, at the points where his project seems to make contact with characteristically cognitive-scientific proposals—points, that is, where he attempts most explicitly to distinguish his concerns from those of cognitive-scientifically inclined philosophers and to indicate which of the positions of the latter he does or does not mean to exclude.

For example, McDowell ventures that he is “rejecting a picture of a mere animal’s perceptual sensitivity to its environment: a picture in which the senses yield content that is less than conceptual but already such as to represent the world. What I am rejecting is a picture of what perceptual states and occurrences are for an animal” (1996, 121). This effectively raises what has been called the problem of infralinguals—the problem that despite the neurobiological and other continuities between higher apes and humans (not to mention those between human infants and adults), the account of intentionality here on offer “seems to introduce a fundamental ontological and methodological discontinuity where there ought to be continuity” (Garfield 1988, 72). This problem, indeed, is often taken to represent one of the principal considerations in favor of physicalist accounts of the mental.²³

The difficulty of these considerations makes it all the more striking that immediately after claiming to reject “a picture of what perceptual states are for an animal,” McDowell nevertheless claims to have “said nothing about how things look when someone tackles scientific questions about how an animal’s perceptual machinery works.” He allows that it’s hard to imagine addressing scientific questions about this “without exploiting an idea of content that represents the world but cannot be conceptual in the demanding sense I have been using”—hard to imagine, that is, “since no animal’s perceptual machinery (not even ours) possesses the spontaneity of understanding.” Given, however, his emphasis on the ineliminably “spontaneous” character of human understanding, it sounds like a non sequitur when Mc-Dowell concludes these reflections by insisting that “I do not mean to be objecting to anything in cognitive science” (1996, 121)—and indeed, by allowing that it would “be dangerous to deny, from a philosophical armchair, that cognitive psychology is an intellectually respectable discipline, at least so long as it stays within its proper bounds” (1996, 55).

McDowell thus thinks that we can uphold the sui generis character of conceptual mental content—the kind “that belongs with the capacities exercised in active self-conscious thinking”—and still allow “the respectable theoretical role that non-conceptual content has in cognitive psychology” (1996, 55). Reflecting what I take to be a significant tension running through the foregoing remarks, McDowell nevertheless holds that we cannot understand our distinctively conceptual capacities as “a welling-up to the surface of some of the content that a good psychological theory would attribute to goings-on in our cognitive machinery” (1996, 55). Particularly, however, insofar as a great many projects in “cognitive psychology” would have us understand our conceptual capacities precisely in terms of a “welling-up to the surface” of subpersonal processes, it’s not at all obvious at these junctures in McDowell’s thought just what he would have us think are the limits of cognitive psychology’s disciplinarily proper “respectability.”

How exactly would McDowell have us understand the relation between our “perceptual machinery” and the “spontaneity of understanding”? He seems in remarks such as the foregoing to allow that there is some such relation, but with his emphasis on the Kantian idea that the sui generis character of the space of reasons is aptly characterized in terms of “spontaneity” (in terms, that is, of its resisting causal explanation), it’s reasonable to wonder how or whether neurocognitive goings-on could be enabling conditions of—how, indeed, they could be in any way related to—the intentionality of the mental as McDowell would have us understand that. Notwithstanding, then, McDowell’s Wittgensteinian confidence that he has exorcized philosophical anxieties, it’s reasonable to think he still owes us an account of just the kind he thinks is unnecessary; “without something like a theory of second nature,” as Robert Pippin expresses this concern, “it is not hard to imagine all sorts of bald naturalists nodding in agreement, convinced that the ‘training up’ of ‘neural nets’ can handle second-nature considerations just fine” (2002, 65).²⁴

Particularly in light of such misgivings about his project, it’s worth highlighting some passages in his corpus where McDowell can be taken to allow that his position does, after all, entail metaphysically significant conclusions—that if what he says about our conceptual capacities is right, then certain claims (characteristically physicalist claims, e.g., about causal efficacy as the criterion of the real) cannot be true. Thus, “If we can rethink our conception of nature so as to make room for spontaneity, even though we deny that spontaneity is capturable by the resources of bald naturalism, we shall by the same token be rethinking our conception of what it takes for a position to deserve to be called ‘naturalism’ ” (1996, 77).²⁵ To the extent that McDowell’s appeal to second nature thus serves not only as an assertion that it’s “natural” for human understanding to exhibit spontaneity but also as a check on scientistic understandings of “naturalism,” the idea may not, after all, beg the questions at issue; McDowell’s claims regarding second nature may, rather, be valuable not just as a therapeutic reminder but as rejecting the view that “causal relations ‘must ultimately be supported by a characterization of the world as the domain of physical natural fact’ ” (McDowell 2009, 139, quoting Friedman 1996).²⁶

Elaborating what I take to be an essentially metaphysical conclusion, Mc-Dowell says of the view he thus rejects that “physicalism about causal relations reflects a scientistic hijacking of the concept of causality, according to which the concept is taken to have its primary role in articulating the partial world view that is characteristic of the physical sciences, so that all other causal thinking needs to be based on causal relations characterizable in physical terms.” This is, he continues, a defensible view of causation only insofar as there is any “reason to credit physical science with a proprietary capacity to penetrate to the real connectedness of things. But I follow Gadamer in holding that there is no such reason” (McDowell 2009, 139).²⁷ The strong conclusion McDowell here embraces—the conclusion that characteristically physicalist understandings of causation cannot be exhaustive—is entailed, I have been arguing, by his case for the ineliminably conceptual character of perceptual experience. The “scientistic hijacking” of the notion of causality—the claim, we might say, that only efficient causal relations are finally real and that only the kinds of things that enter into such relations can therefore be thought really to exist—is ontologically incomplete (representing only a “partial world view”); this is because we can make sense of the intentionality of the mental only by making reference to the “logical space of reasons.” What McDowell thus refuses with his characterization of them as “scientistic,” then, is just those scientific accounts that “claim to contain all the real objects there are” (1998b, 473n).²⁸

The ineliminable character of our conceptual capacities thus demands that we not take causal relations to consist only in “relations characterizable in physical terms.” With respect, then, to the concern—expressed by Boghossian and noted at the beginning of chapter 2—that affirming the ineliminably normative character of mental content seems to commit one, “implausibly, to the essential incompleteness of physics,”²⁹ we might take the lesson from McDowell to be that it’s only mistakenly that we could ever have assumed the completeness of physics in the first place.³⁰ On this reading, McDowell’s argument for the ineliminable character of our conceptual capacities (of our “faculty of spontaneity”) just is, among other things, a transcendental argument against physicalism—one that can be taken to show that it’s a condition of the possibility of experience’s objective purport that causation cannot be exhaustively describable in the terms of physics.

On the complete picture I have aimed to elaborate following McDowell, then, the argument is not only that our experience can have “objective purport” only insofar as experience always already involves our conceptual capacities (only insofar as perceptual content is essentially the kind of thing one might take as true); he is also, I’m now emphasizing, committed to arguing that insofar as we thus cannot make sense of our experience except as having a conceptual dimension, any understanding of causality that renders this problematic cannot be right. Thus, it is finally a condition of the possibility of our “having the world in view”—of our having the kind of mental lives and experience that we do—that causation cannot be exhaustively described in the terms typically favored by physicalists. If he says anything at all by the claim, then, McDowell says something quite significant in saying that “we need not see the idea of causal linkages as the exclusive property of natural-scientific thinking” (2009, 258).

If it’s right thus to take McDowell as making essentially metaphysical claims at these junctures, it’s not immediately clear how different the resulting picture finally is from the basically dualist kind of account one can take from Kant.³¹ While it is not the point of their projects to warrant the conclusion that consciousness or subjectivity represents an ontologically distinct kind of “stuff,” for both McDowell and Kant it seems we may still have the kind of “dual-aspect” view according to which we are asked to accept that there are two fundamentally distinct levels of description (call them causal and intentional) that make sense of human being—and for both thinkers, the question of how or whether these levels of description relate to each other is meant somehow to be ruled out. This foreclosure of the question of their relation need not, however, be understood in terms of the familiarly Kantian reference to constitutively unknowable “things-in-themselves.” The real insight may be not that there is a mysteriously but essentially unknowable “other side” of reality, but that there is nothing it could look like to answer the question of how these levels of description relate—that nothing could count as explaining how reasoning “really works” (how it causally intervenes in the world), since the very idea of there being some way reason “really works”—something else (something “natural”) that our responsiveness to reasons really consists in—could only be the idea of it as something other than reasoning.

It is, in other words, only under the right description that it could in any case be our conceptual capacities that we are talking about; it’s only as conceptual that intentionality can be individuated, only as semantically contentful that it could be someone’s reasons-for-acting that we are talking about. The demand that we show what these “really” are is incoherent as long as it’s thought that it can be met only by indicating something not itself conceptual or intentional; it makes no more sense to think we could thus reduce epistemic to nonepistemic facts than it does to say we might know something unknowable. Indeed, the very distinction between the “logical space of reasons” and that of “nature” is itself internal to the space of reasons; our knowing anything at all about this always already subsumes the very divide that seems to present a problem.

Philosophical intuitions tend to diverge sharply over whether this is a significant point; it’s not immediately clear whether it’s a truism (one to the effect that we can only think about things as thought-about-by-us) or whether instead it discloses a logically or metaphysically significant constraint. While he is, to that extent, here unlikely to move those who are strongly inclined to take the point as vacuous, I take it that it is this broadly Hegelian thought that McDowell elaborates when he revisits the themes of Mind and World in his 1998 Woodbridge Lectures.³²

Among the recurrent images in Mind and World is that of our conceptual capacities as somehow extending “all the way out”³³—an image evidently meant to suggest that we cannot make sense of the world itself as what is on the “other side” of our conceptual capacities, as constraining the latter from the outside. Despite, however, his own indebtedness to Sellars in developing this thought, McDowell comes, in his Woodbridge Lectures, to convict Sellars himself of smuggling in a form of just the sort of “givenness” he so influentially criticized; fairly or not, McDowell thus reads Sellars’s commitment to critical realism as itself involving a finally scientistic stance.³⁴ Mc-Dowell embraces Sellars’s guiding thought that a “conceptual episode’s being intentionally directed toward an element in the real order is analogous to, say, a linguistic episode’s containing an expression that functions as a name of an element in the real order”; but he resists the view (also found in Sellars, he thinks) that “this ‘aboutness’ must not be conceived as a relation between an element in the conceptual order and an element in the real order” (McDowell 1998b, 479).³⁵ According to this reading of Sellars, it must be something from the world under a true scientific description that finally constrains thought. According to McDowell, then, Sellars finally thinks Kant ought to have “cast the objects of the scientific image, which are distinct from the objects of the manifest image, in the role of things in themselves” (1998b, 469n23).³⁶ On this reading, Sellars’s point is that certain scientifically described facts about the world finally represent the real conditions of the possibility of our having the kind of experience we do.

One version of the kind of view McDowell is here concerned to reject—and we can set aside the question of whether it’s really Sellars’s view—is the idea that the posited scientifically describable constraints on experience are precisely such as can supersede (can explain away) even the fact of our knowing them. This is the idea that the scientifically describable facts represent, in other words, “what there really is” instead of the kind of first-personal experience we seem to ourselves to have (even where it is the experience of adducing those scientific facts that we are talking about). McDowell argues that this kind of view cannot coherently be thought to get, as it were, “behind” the kind of experience we are trying to understand, just insofar as such facts are only accessible to us as already the content of just such experience; against, then, Sellars’s supposedly scientific-realist construal of Kant, McDowell urges that “for Kant objects as they appear in the scientific image would be just another case of objects as they appear, with a transcendental background for that conception just as necessary here as anywhere. Sellars’s attempt to be responsive to Kantian transcendental concerns goes astray in his idea that an appeal to science could do the transcendental job” (1998b, 469n23).³⁷

McDowell thus argues that we cannot make any sense of the idea of explaining our conceptual capacities in terms of anything that is not itself already the content of those very capacities; just insofar as we think any such explanation, thought already sublates the very distinction between thought and reality that we would thus claim to have invoked. Wary, perhaps, of the variously egregious excesses of Hegel, McDowell suggests that “Tarskian semantics points to a sober interpretation”³⁸ of the clearly Hegelian point he is making here (1998b, 489). McDowell thus invokes Alfred Tarski’s “Convention T,” which specifies that in order for any theory of truth to be adequate, it must entail that for every sentence P of any language, “P” is true if and only if p—as, on a canonical example, “ ‘Snow is white’ is true if and only if snow is white.” Here, it’s important to appreciate that the seemingly tautological repetition on the right-hand side of such statements represents something essentially abstract—the proposition, for example, that is expressed by the natural-language phrase on the left-hand side. This distinction is obscured, however, by the fact that we can only express propositions in natural-language sentences. McDowell’s Hegelian turn can be characterized as emphasizing just this point; thus, he comments that “we have to use the words on the right-hand side of semantical statements” (1998b, 489; cf. 2009a, 63).

If I rightly understand him, the point is that we cannot suppose the proposition on the right-hand side amounts to an explanation, from essentially outside semantic space, of how the sentence on the left-hand side means what it does; Tarski’s point in deploying this apparatus, rather, is itself intelligible only given the essentially semantic character of both sides of the equation. With respect, then, to the essentially Sellarsian thought that “the categories of intentionality are, at bottom, semantical categories pertaining to overt verbal performances” (Sellars 1956, 94), McDowell is here arguing—against what he takes to be Sellars’s own finally scientistic compromising of a properly transcendental insight—that it matters that “we have to use the words that figure in specifications of what nonovert conceptual episodes are intentionally directed toward. In statements of meaning and aboutness, we relate the conceptual order to the real order, mentioning elements of the real order by making ordinary uses of the words on the right-hand sides of these statements” (1998b, 489; cf. 2009a, 63). The logical space of reasons (to return to the image from Mind and World) goes “all the way out,” then, in the sense that anything we could know or think about “the other side” of that space would, ipso facto, already be within it.³⁹

The logical space of reasons, on this account, is thus ineliminable—which is also to say that our conceptual capacities are ineliminable, that epistemic facts cannot be “analyzed without remainder into non-epistemic facts”—in the sense that anything we could know about what explains this would already be itself “contained” therein. There is, then, nothing it could look like to know that, say, intentionally describable states like knowing or believing might “really” be something else, or that our conceptual capacities are explicable in terms of things that are not themselves conceptual; just insofar as this could be known (could be an object of thought), it would be known “without moving outside the conceptual order—without doing more than employing our conceptual capacities” (McDowell 1998b, 489; cf. 2009a, 63).

Insofar as it thus involves reflection on what we are doing in knowing (or arguing for) the truth of any position, the foregoing nicely introduces an argument that seems to me a necessary complement to the account so far developed—an argument for thinking that the ineliminable character of our conceptual capacities imposes a metaphysically significant constraint on our thinking about intentionality, and that the problems with reconciling certain kinds of causal accounts with our first-personal experience of intentionality should finally be thought to count against the former rather than against the latter. The argument I here have in mind centers on what I follow Kant in calling practical rationality, and is to the effect that the physicalist’s own demand—the demand, in the face of the difficulties it raises regarding mental causation, that we justify our commitment to the ineliminably “spontaneous” character of conceptual thought—is itself intelligible only given the truth of the claim being challenged.

The complementary argument from practical reason is significant, I think, insofar as it can reasonably be thought possible to reverse the argument so far developed following McDowell—possible, that is, to concede McDowell’s point that conceptual and causal descriptions are irreconcilable, but to conclude that this is just so much the worse for descriptions of us as ultimately responsive to reasons. The position I have reconstructed following McDowell—according to which any understanding of causality on which our conceptual capacities becomes problematic cannot, ipso facto, be the right account—is finally only as strong, then, as our commitment to the reality of what McDowell calls the “epistemic position we are manifestly in.” The argument so far developed shows, in other words, that we could only be in this situation if, inter alia, our experience is not exhaustively describable in causal terms. But if one is prepared to deny that we really are in the epistemic situation McDowell takes it we are manifestly in, then the fact that some account of causal relations—one that assumes the completeness of physics, for example—is irreconcilable with a purported condition of the possibility of our being so can be thought to count not against our understanding of causation, but against our really being in the situation we think we’re in.⁴⁰

To conclude this, one would have to have good grounds for thinking it is particularly the causal description of us that is to be preferred in case of conflict between that description and how things seem to us. The present argument thus focuses on the question whether we really are entitled to think this kind of “theoretical” description more likely true than is what it seems like to reason about it. That we are not so entitled is what Kant argues in his Critique of Practical Reason, in which (Kant says) he is concerned “merely to show that there is pure practical reason” (1788, 3). Thus to show that reason is “practical,” on this account, is just to show that reason is “concerned with the determining grounds of the will” (1788, 12)—that explanations of what we do must include some reference to our reasons for doing so and, conversely, that our reasons for acting have some role in explaining how we act. More precisely, Kant’s question (as we saw in the first section of chapter 2) is “whether pure reason of itself alone suffices to determine the will or whether it can be a determining ground of the will only as empirically conditioned” (1788, 12). In aiming, then, to show “that there is pure practical reason,” Kant particularly has it in mind to show only that we are (as McDowell regularly puts the point) responsive to reasons as such—that it is as semantically contentful (and not as empirically conditioned) that reasons must figure in a complete account of what we are like.

That reason is “practical” in this sense is just what is denied by those who hold that it’s only under some other description—only, for example, as represented in brain states—that reasons can be thought to figure in psychology. On one widely entertained version of this denial that reason is really “practical,” the “common-sense” view of the mental—the view, in effect, that reason is “practical”—really amounts to a theory (often characterized as “folk psychology”). The point of saying so is that it’s in the nature of theories to admit of alternatives; to characterize our first-personal sense of ourselves as doing things for reasons as “theoretical,” then, is to suggest that this is essentially comparable to theoretical posits like phlogiston and ether—and that our sense of ourselves may, to that extent, be altogether superseded.⁴¹ While we may, that is, subjectively experience ourselves as responsive to reasons, it’s conceivable that a complete scientific description of our cognitive machinery will, though making no reference at all to items such as “reasons” and “beliefs,” nevertheless capture everything there is to understand about the mental. It’s against such views that Kant aims to show that our responsiveness to reasons as such is ineliminably among the “determining grounds of the will”—that human action cannot be exhaustively described without reference to the semantic content of reasons and beliefs.

It’s important to emphasize that for reason thus to be among the “determining grounds of the will” is (perhaps notwithstanding Kant’s image of a “grounding” relation) precisely not for it to be understood as causally intervening in our actions; it is only under the right description that it could be our reasoning that we have in view, only as semantically contentful that it could be someone’s reasons-for-acting that we are talking about. This point is reflected in comments from Elizabeth Anscombe, who brings together the sense of “intentionality” in play throughout this book—the sense in which an intentional level of description is involved in experiencing someone who is speaking as not just making noises but as saying something—with the garden-variety sense in play when we talk of someone’s intending to do something. The latter idea, Anscombe says, “would not exist if our question ‘Why?’ did not. It is not that certain things, namely the movements of humans, are for some undiscovered reason subject to the question ‘Why?’ ”⁴² Her point is that “the description of something as a human action could not occur prior to the existence of the question ‘Why?’, simply as a kind of utterance by which we were then obscurely prompted to address the question” (1963, 83).

It is, in other words, intelligible that reasons as such might be among the “determining grounds” of any action just insofar as it’s the kind of action regarding which it makes sense to ask the agent why she did it. That the agent in such a case is responsive to reasons as such is not to be known, then, by (for example) inferring from the action that it must have been caused by the having of a reason; rather, there is something more like a logical or conceptual relation (an internal relation) between any action and the reasons that may be taken as among its “determining grounds.”⁴³ The very idea that someone might have acted intentionally (in the garden-variety sense of that word) emerges, then, only insofar as we already find intelligible the demand for justification; we discover our responsiveness to reasons as such simply in our already understanding what we are being asked when we are called on to justify anything we have done or claimed. Making just this point, Kant urges that “morality first discloses to us the concept of freedom”; it is, he says, “practical reason which first poses to speculative reason” the possibility of our freedom, practical reason that has “come in and forced this concept upon us” (1788, 27). The moral idea of freedom that Kant chiefly meant to advance in terms of conceptual “spontaneity” consists, then, at least in our finding “Why?” questions intelligible.

For Kant to show, in light of this, merely “that there is pure practical reason”—that our responsiveness to reasons as such must remain part of a complete account of what we are like—is therefore to show simply that we find demands for justification intelligible, and that the aspects of our being that can only be individuated relative to that fact drop out of view if it’s denied that reason really is thus “practical.” This point is not understood, moreover, as long as we persist in thinking it must mean that reasoning is causally efficacious—a point McDowell expresses as the thought that “the way certain bodily goings-on are our spontaneity in action, not just effects of it, is central to a proper understanding of the self as a bodily presence in the world” (1996, 91n). For a “faculty of spontaneity” always already to be in play just is for us to find it intelligible that we can individuate certain of our actions by entertaining the question whether they are as we ought to have done—a fact that cannot be individuated without reference to the logical space of reasons.

What Kant’s second Critique most significantly adds to the understanding of freedom that can be had following the first, then, is an argument to the effect that if any of the foregoing still seems problematic—if, that is, our interlocutor persists in thinking that it finally counts not against the completeness of our understanding of causation, but against the individuation of actions as “intentional” that the latter can only be effected relative to the semantic content of demands for justification—it is nevertheless a condition of the possibility of arguing so that the space of reasons already be in play. As Kant says to this effect, “if as pure reason it is really practical, it proves its reality and that of its concepts by what it does, and all subtle reasoning against the possibility of its being practical is futile” (1788, 3). Any argument for the finally eliminable character of reason’s being “practical,” then, is only intelligible given the idea of arguing for it; just to that extent, though, the fact that our reasoning has some role to play in our behavior is not something that can coherently be denied, since it is only by reasoning that one could do so. There is no making sense of the idea of anyone’s hoping to persuade us, in effect, that acts of persuading are not really real.

This, finally, is the kind of argument that I take Dharmakīrti’s Mīmāṃsaka and Mādhyamika critics most cogently to press against him. These Indian critics of Dharmakīrti, we will see in chapter 6, can be taken similarly to press the point that reason really is “practical,” variously advancing arguments to the effect that we cannot make sense of our knowing of seemingly intentional actions (actions like arguing for reductionism) that they “really” are explicable in exhaustively causal terms. To raise the question whether this is so is already to engage in reasoning on the subject, already to step back from the situation and ask whether things might be other than they seem to us to be; to step back thus from one’s immediate situation, though, just is to evince a responsiveness to reasons as such. If, then, we are to find the physicalist’s engagement in the argument intelligible, it must be that she finds intelligible the idea of persuading her interlocutor of the truth of her claim—and unless we can give a causal account of the intelligibility of that way of understanding the conversation, there cannot, ipso facto, be an exhaustively causal account of the intentionality that is on display in the physicalist’s reasoning for her case.

Reason’s being “practical” is thus shown finally to be a condition of the possibility of its also being “theoretical.” This point makes sense of G. F. Schueler’s claim (noted in chapter 2 with regard to Dennett’s “intentional stance” strategy) that there is necessarily “a ‘non-theoretical’ element at the heart of reasons explanations, namely the way I understand my own case when I act for a reason” (2003, 160). To argue, in other words, that any account of intentionality (even one that would explain it away) is itself intelligible only as the content of the constitutively intentional activity of reasoning about it is, among other things, to argue that our first-personal experience of rationality is logically basic and ineliminable; for me to understand anyone as responsive to reasons is finally to understand them, that is, in terms of what I am doing when I experience myself as acting for a reason.⁴⁴ Insofar, then, as we cannot finally go any “further down,” in understanding or attributing rationality, than to our own first-personal experience thereof, to argue (with Kant) that “there is pure practical reason” is also effectively to argue that an exhaustively impersonal account of the mental is finally unintelligible. To the extent, though, that we accept McDowell’s arguments that our first-personal awareness must nevertheless be constitutively of a world, to argue for the logical priority of the first-personal understanding of rationality is not to push toward solipsism; it is, indeed, the basis for what is arguably the best account of the objectivity of our knowledge. (Dignāga and Dharmakīrti, we will see in chapter 5, come to a different conclusion, based on considerations involving the first-personal character of experience.)

We now have a complete argument to the effect that in reflecting on the nature of intentionality, we cannot coherently eliminate the point of view from which we must do so. The common-sense experience of the mental is therefore not to be understood as analogous to a theory, insofar as it is not, in fact, open to us to accept or reject the very fact that it is only as reasoning that it makes sense to think about what “reasoning” might be. Any thought we could have on this or any other subject must, that is, be such as could in principle be expressed as the content of Kant’s “I think” (“I think that intentionality is causally explicable”)—and we have now followed a chain of argument to the effect that this just is to say that anything that could be the content of our theorizing must be conceptually articulable (must be “in” the logical space of reasons), and that this finally entails that putatively exhaustive understandings of causation cannot be right.

If, at the end of the day, it nevertheless seems intelligible that our responsiveness to reasons as such might admit of an alternative explanation (one according to which conceptual thought is finally constrained by something altogether outside of conceptual space), we have now seen that this cannot be right just insofar as we understand what it would mean to argue for that very view. To give up the idea that it is thus significant what we manifestly do in reasoning about this, as Lynne Rudder Baker says in concluding a transcendental argument of just the sort elaborated here, is “to give up the point of view from which thinking about anything, meaning anything, or doing anything intentionally is possible. To abandon the common-sense conception of the mental would be to relinquish the point of view from which the idea of making sense makes sense” (1987, 173).⁴⁵

Whether or not the line of argument I have thus developed following Kant, Sellars, and McDowell is finally judged cogent, the point for our present concerns is that this line of argument cuts not only against cognitive-scientifically inclined physicalists but also against Buddhists like Dharmakīrti. Despite the extent of Dharmakīrti’s own interest in refuting physicalism, the foregoing argument against physicalist attempts at “naturalizing” intentionality is not only one that it is not open to Dharmakīrti to make; it can also be enlisted to identify and characterize philosophical problems with Dharmakīrti’s own project. To that extent, Dharmakīrti’s own account of intentionality may finally have something essentially in common with the kinds of physicalist accounts he so unambiguously eschews.