The world shapes our concepts and, in turn, is shaped by them. This truism is reflected in some of our best thinking about religion, both classical and contemporary. Hume takes the former route, tracing the content of religious ideas and concepts to external, material forces. He takes “religion” to mean “belief in invisible, intelligent powers”—a belief caused by fear of unknown causes in combination with our tendency toward anthropomorphism. Durkheim also writes in this spirit. He finds that the alternation of density and dispersion in social life gives content to the general term “religion,” and to the sub-concepts “sacred” and “profane.” For Schleiermacher, religion is “a sense and taste for the infinite”; content shaped, apparently, not so much by the infinite as by his desire to inoculate religion—he insists not only Christianity—against what he saw as encroachment from science and ethics.1 Other critics, reversing this polarity, track the effects of religious concepts on the world. Thus, we have Max Weber’s claim that the Protestant concepts of a calling and of predestination helped capitalism take root in parts of Europe and North America.
A third option also tells a causal story, but is suspicious of the contrast between religious concepts and the world. It finds that, on examination, “religion” and its kin—“world religions,” “Hinduism,” “Islam,” and others—are vague to the point of vacuity, that they as much as have no content. For these critics, the search for causes is in the service of explaining why so many of us have taken such concepts, or apparent concepts, to represent smooth, well-formed, objective features of the world. One line of thought has them having been “manufactured” to support the rise of such institutions as university departments, professional journals, and academic conferences.2 Another has them constructed to serve “the interests of American power,”3 and yet others find them serving “the logic of European hegemony.”4 Here the search for forces is less about understanding the meaning of religious concepts and more about dissolving the illusion of meaning. Such doubts are all the more striking for having arisen in the midst of what is widely seen as a resurgence of religion world-wide.
Surveying this scene we may not know whether we are confronted by richness or confusion. In part, this is to be expected. Sorting out the interplay between concepts and the world—what Robert Brandom calls the dynamics of conceptual behavior5—is a tricky business. After all, even the most innocent-seeming empirical term may be unexpectedly complex; apparently there is no single test for the hardness of both metal and plastic.6 If “is hard” resists regimentation it is hardly surprising that “is religious” does, too. And, to some extent, no doubt our authors are talking past each other: Are Hume and Durkheim referring to the same phenomena? What about Durkheim and Schleiermacher? How can we even go about deciding? Are the skeptics looking at what the rest of us see?
I suspect there are no fully satisfactory answers to these questions. Certainly I do not have them. But the wager here is that we will come to some clarity about the behavior of “religion” by asking how concepts work in general—what Brandom labels “statics.” Thus, we will begin with what astronomers call exercises in averted vision. In dimly lit conditions we take best advantage of the physiology of the eye by looking just to the side of the object of inquiry. In the present case, we must look away from dynamics—away from history, from social institutions and practices, indeed away altogether from the material basis of the study of religion. Instead, we will focus on the nature of concepts, on the internal structures and processes whose contributions to meaning may be well hidden but are no less real. As compared to dynamics no doubt these structures and processes can seem pale and antiseptic. But any account of conceptual content must reflect the interrelationship between the material world and the nature of concepts themselves. Kant offers a powerful diagnosis of the conflicting pressures that shape such general terms as religion. According to Kant, the conflict goes as deep as human discursivity itself; that is, to the way in which we generate and use concepts.
Let us approach Kant’s diagnosis by way of a familiar passage from Fielding.
When I mention religion, I mean the Christian religion; and not only the Christian religion, but the Protestant religion; and not only the Protestant religion, but the Church of England.7
Two elements of Parson Thwackum’s remark put us on the right track. First, concepts—whatever else they are—are composed of further sub-concepts. The sphere or extension of a concept is not the class of individuals to which it applies; you don’t give the content of Christianity or Protestantism or the Church of England by pointing to all the people or artifacts belonging to or falling under them. Rather, you give further sub-concepts. Second, there is in principle no absolutely lowest concept. Thwackum’s purposes were served by stopping at the Church of England, but a moment’s thought shows that we can go on as far as we like, meaning not just the Church of England, but just that portion of it believing in gay ordination, and not only that portion of it believing in gay ordination, but that portion of it believing in gay ordination and in the ordination of women, etc.
By facing it in the opposite direction, toward higher genera rather than toward lower species, this picture of concepts yields some insight into how the general term “religion” is sometimes used today. Beginning with Fielding’s highest genus, Christianity, we need only add species of equal generality alongside it—Buddhism, Islam, Judaism, and so on, so that “religion” will be composed of smaller parts in the way that species are parts of a genus, in just the way the extension of the general term “automobile” is composed of smaller parts—“sedan,” “convertible,” “compact,” and so on. Since concepts are not composed of individuals, this account puts no weight on numbers; convertibles and sedans are on par, as are Santería and Islam and Judaism and Hinduism. It appears to be on this basis that many academic departments of religion are structured today.
Armed with this preliminary story about concepts, let us examine some familiar doubts about “religion.” Jonathan Z. Smith has famously remarked that “Religion is solely the creation of the scholar’s study. . . . Religion has no independent existence apart from the academy.”8 We can see that, in at least one good sense, he is right. “Religion” is a concept. Concepts are composed of further concepts, and not of existing individuals. As existing individuals are not, concepts are the products of human cognition. So, “religion” is a human invention and has no—or, perhaps, allowing for leakages of various sorts—limited existence outside the academy. In fact, this claim is generalizable across the conceptual board: “Buddhism,” “Judaism,” “Islam,” and all the rest are human inventions. We have made them all up. If anything, Smith is underestimating the vast cosmic power wielded by scholars of religion. Of course, the fact that we have made up the concept religion does not mean that there are no religious persons or artifacts, it just means that when we judge of some existing person or artifact that she or it is religious, we must place the person or artifact under some sub-concept (some subspecies), some part of the concept religion. That person must be Buddhist or Jewish or Sunni or Shia, etc. From the point of view of the theory of concepts, the existing individuals to whom the concepts apply are as points—they are extensionless, they take up no space. Thus, we have hit upon one good sense in which the term “religion” is, as Willi Braun says, “substantively empty.”9 It is empty in the sense that it is not composed of individual religious persons or artifacts or indeed of any-thing. But this emptiness implies no deficiency in the general term. All concepts are empty in the antiseptic sense that none are composed of existing individuals but rather only of further sub-concepts.
At this point we may be ready to proclaim our general concept vindicated. The hierarchical nesting of conceptual genus and species may seem to guarantee “religion” a perfectly respectable meaning, and we may then be tempted to take a very short line with the critics. We may be tempted to echo Hilary Putnam’s memorable assessment of Count Alfred Korzybski’s program of General Semantics:
Korzybski used to claim that to say of anything that it is anything—for example, to say of my car that it is an automobile—is to falsify, since . . . there are many automobiles and my car is not identical with any of them. . . . He recommended that one should use the word etcetera as often as possible. In his view, it would be highly therapeutic to say, “That is an automobile, etc,” and not, “That is an automobile,” in order to keep in mind that the “that” referred to (my car) has infinitely many properties besides those mentioned in my statement.10
Like “automobile,” “religion” is a general term, and so may be thought to invite Korzybskian analysis. Korzybski would presumably recast Weber’s claim that “Catholics like to sleep well, Protestants to eat well,” to a series of observations about the eating and sleeping habits of many, many individual persons. We can imagine a suitably hedged reconstrual of Marx’s slogan about religion as opiate. Durkheim’s claim that social density produces religious representations at most describes a group of perhaps several thousand Australian aborigines—and at that only partially and with qualifications, etcetera.
Putnam diagnoses Korzybski’s error as follows:
That everything we say is false because everything we say falls short of being everything that could be said is an adolescent sort of error. “Enough is enough, enough isn’t everything,” John Austin wrote, and that applies to interpretation as much as to justification.
True, universal judgments about religion fall short of saying everything that could be said—most of us know Protestants who like to sleep well, it is hard to think that religion was an opiate for the American civil rights movement, and collective effervescence does not always produce religious representations. But this again implies no criticism of the general term. Enough is enough, enough is not everything.
Putnam and Austin are making an important point, but it does not engage the present critics at a deep enough level. The more serious doubt is not about the legitimacy of general concepts (in general) but about the legitimacy of this one in particular. One way to bring out the deeper doubt is to press the analogy between “religion” and “automobile.” When, under normal circumstances, a competent English speaker asserts, “That’s an automobile,” she has fairly specific expectations about what she would encounter in the course of experience—four wheels, an engine, headlights, a roof, etc. “That’s a convertible” will yield all of these except that the roof will be found to be retractable. Furthermore, she can count on her English-speaking listeners to share those expectations (and for them to expect that she in turn shares them, and expects them to be shared). To be sure, ordinary vagueness is very much in the picture—a three-wheeled vehicle can’t be an automobile? What if there is no roof? Still, we smile at Korzybski’s suggestion because we know we can be precise enough about the meaning of “automobile” and “convertible” for the purposes of conversation and further inquiry. Can we say the same for the term “religion” as it is used in my first paragraph?
To the extent that we hesitate in answering we raise the prospect of a deeper form of emptiness than the one contemplated several paragraphs ago—the sort of emptiness we reserve for concepts directed at no object, what Kant in the Critique of Pure Reason calls “usurpatory concepts, such as fortune or fate, which, though allowed to circulate by almost universal indulgence, are occasionally called upon to establish their claim by the question: quid juris?, and then there is not a little embarrassment about their deduction because one can adduce no clear legal ground for an entitlement to their use either from experience or from reason” (A84/B117).11
II. The Spatial Theory of Concepts
Kant’s challenge is aimed at a set of a priori concepts—a priori in the sense that, so the claim, we must call them into play if experience is to be possible at all. We cannot then, without circularity, “establish their claim” by calling upon experience. Kant tells us that such concepts as these require a special form of justification, a “transcendental deduction,” in which we argue that either we have the right to employ them or we have no experience. By contrast, such empirical concepts as “religion” require only empirical support. Indeed, the hierarchical account of empirical concepts presented in the preceding section is essentially the one Kant advances in several places in the Critique, first in the “second observation” on the Table of Judgments (A72–73/B97–98; added in the second edition), and later in more detail in a particularly dark section titled Appendix to the Transcendental Dialectic (A642/B670–A668/B696; preserved from the first).12 Here is the key passage:
From the sphere of the concept signifying a genus it can no more be seen how far its division will go than it can be seen from space how far division will go in the matter that fills it. Consequently, every genus requires different species, and these subspecies, and since none of the latter once again is ever without a sphere, (a domain as a conceptus communis), reason demands in its entire extension that no species be regarded as in itself the lowest. For since each species is always a concept that contains only what is common to different things, this concept cannot be completely determined. It cannot, therefore, be related directly to an individual, consequently, it must at every time contain other concepts, i.e., subspecies, under itself (A655–56/B683–84).
We see here reflected several of Kant’s most characteristic doctrines. Since concepts are composed of what may be common to more than one thing they are all general. Thus, they cannot be directly related to individuals since, in the nature of the case, more than one object may fall under any given concept—say, “63 lime-green Mustang convertible with 74,359 miles and a two-inch scratch on the left rear fender.” No amount of conceptual detail can guarantee univocal reference. Rather, Kant argues that, for finite creatures like us, direct relation to objects involves being affected through sensation (Empfindung)—and then in applying general concepts to what strikes us. This is a central element in Kant’s humanizing of his Leibnizian heritage; it’s what he means when he says we have finite, discursive understandings.13 By contrast, a self-sufficient, infinite cognizer would presumably not require a contribution from sensation; an infinite cognizer would presumably bring its object into existence in the act of cognition.14 To give content to our concepts such finite creatures as ourselves must be able to exemplify them, as Kant likes to say, in concreto—in the world. This is generally not difficult when it comes to such directly perceivable middle-sized objects as automobiles; terms whose referents cannot be immediately perceived must be connected to objects that can be—thus, Kant writes of giving “magnetic field” its meaning through the immediate perception of the arrangement of iron filings (A225/B273).
In the first half of the quoted passage, Kant is laying out what has been variously termed the “conjunction”15 or the “spatial”16 model of the extension of a concept. The basic claim is that a concept’s extension (Umfang) should be conceived as a sphere, the parts of which are further concepts or further spheres. Just as bounded spaces admit of division without end, so there is no absolutely lowest element, no infima species, in the hierarchy of concepts. Every concept (except that of the highest genus) is composed of a hierarchically nested and indefinitely extended series of subspecies (sub-concepts), even as it itself forms part of a larger genus. The conjunction or spatial model of concepts in turn plays a crucial role in Kant’s larger picture of discursivity. The larger picture says that creatures like us must apply general concepts to what strikes us in sensation, and—appealing now to Kant’s concept of concepts—it specifies the structure of those general concepts as composed of hierarchically nested and indefinitely extended series of sub-concepts. That is, in forming ordinary empirical judgments I am calling into play the whole apparatus of nested genus and species, because, for creatures like us, that’s how concepts are structured. Thus, when I am struck through sensation and react “convertible” I am applying a concept that may apply equally to other objects, because by its nature it “contains only what is common to different things”—that is, such sub-concepts (subspecies) as “retractable roof” or “roller bar,” each of which is itself composed of further sub-concepts without limit. When Wendy Doniger writes of “Hinduism” she means a “complex system of interlocking, sometimes contradictory ideas and ideals—caste, karma, renunciation, and the worship of various gods—[that] has formed the religious scuttlebutt, the common wisdom, of all Hindus for many centuries.”17 Put in the terms that we are currently developing, Doniger is saying that the extension of the concept “Hinduism” is composed of four main sub-concepts or subspecies, “caste,” “karma,” “renunciation,” and the “worship of various gods.” And of course we know that each of these is composed of further sub-concepts; “caste” is composed of, for example, “Brahmins,” “Kshatriyas,” “Vaishyas,” and “Shudras,” the first of which is composed of the yet narrower concepts “scholars,” “teachers,” etc.
The account of concepts in play here, when more fully developed, will figure prominently in what follows. It constrains the nature of definition (chapter two), it motivates the distinction between regulative and constitutive, and so supports the objectivity of unitary, systematic generalizations (chapter three), it sets the terms for a legitimate sense of non-conceptual content (chapter four), it poses a challenge to any adequate concept of the self, including Kant’s own (chapter five), and it introduces a holistic element into the business of interpretation (chapter six). In the remainder of this section I want to make a first pass at its main features.
The general view is that concepts—say, Hinduism—can be represented as conjunctions of two or more partial concepts, as, for example, caste, karma, renunciation, and the worship of various gods. Put in this very broad way, Kant is standing in a tradition stretching back at least to Boethius, for whom “homo”stands under “animal rationale” by way of “mortale” but not “immortale.”18 But Kant introduces several innovations into the conjunction model of concepts as it was traditionally understood. One is the basic move of conceiving of a concept’s extension as a bounded space. Motivating this conception is apparently Kant’s view that existence is never established through concepts alone, that existence claims require some intuitive content, some Empfindung. We cannot, for example, bring Hinduism into the world simply by assembling it out of smaller concepts. Thus, the reason Kant turns to space to model concepts is that he needs a way to talk about an individual’s relation to its species from within the theory of concepts—that is, without appealing to intuition. The solution is to say that an individual is to its species as a point is to the space that includes it. Since points have no extension they are included in but cannot serve as parts of the including space.19 This allows Kant to reject theories of concepts which make individuals their parts. For Kant, that would be to allow conceptual content in the absence of intuition—as though a higher-level concept could be given content simply by combining two narrower ones, and so on all the way down. In view of the importance of the spatial analogy to the account of concepts, from this point forward I will speak of Kant’s spatial theory or model of concepts rather than of the conjunction theory or model.
Second, Kant moves the discussion of concepts away from the Cartesian rubric of representation—so ably criticized by Richard Rorty in Philosophy and the Mirror of Nature—to that of action. Rorty indicts both Descartes and Kant for routing our cognition of objects through the fictitious intermediaries of ideas or representations, making it seem natural to ask whether they function as a mirror or a veil.20 But in fact Kant thinks of concepts as “predicates of possible judgments” (A69/B94). He assimilates using a concept to judging in accordance with a rule: “the concept of a dog signifies a rule in accordance with which my imagination can specify the shape of a four-footed animal in general, without being restricted to any single particular shape that experience offers me or any possible image that I can exhibit in concreto” (A141/B180). Competent English speakers are typically following a rule according to which one reacts “car” when one encounters an object having four wheels, an engine, headlights, a roof, etc. As I am reading Doniger, one may legitimately react “Hinduism” when one encounters something characterized by caste, karma, renunciation and the worship of various gods (and of course much else). For Bruce Lincoln, “four domains—discourse, practice, community, and institution—are necessary parts of anything that can properly be called a ‘religion’.” So, in order to act (to judge) in accordance with this rule, we must be presented with something that reflects each of these four “domains.” Again, Thomas Tweed takes religions to be “confluences of organic-cultural flows that intensify joy and confront suffering by drawing on human and suprahuman forces to make homes and cross boundaries.” Understandably, Tweed spends a chapter spelling out what he means by this definition. According to Kant, what Tweed is really doing is specifying the conditions that legitimize a certain kind of action; namely, using this concept to form a judgment.21
Third, Kant’s denial of a lowest or infima species makes an important contribution to the spatial theory of concepts. In the long quotation, above, Kant remarks that no concept, no matter how finely drawn, is ever without parts—even if they have not yet been identified. In the Logic, Kant develops the pragmatic element in this thought:
Even if we have a concept that we apply immediately to individuals, there can still be specific differences in regard to it, which we either do or do not note, or which we disregard. Only comparatively for use are there lowest concepts, which have attained this significance, as it were, through convention, insofar as one has agreed not to go deeper here.22
That is, having achieved whatever practical purpose engages us, we do call a halt—as did Thwackum—but further investigation could always turn up a yet-narrower species.
Now, in one form or another, these three features of Kant’s theory of concepts—generality, rule-governedness, and the denial of an infima species—or something very much like them, are not uncommon in contemporary portrayals of concepts.23 A fourth element of Kant’s discussion concerns the several ways that the parts of a concept, its sub-concepts, can be related to one another—what Robert Hanna calls a concept’s “conceptual microstructure.”24 At this point we come to his signature deployment of the distinctions between analytic and synthetic and a priori and a posteriori. These distinctions must loom large in any discussion of Kant’s epistemology. I will be developing and trading on them over the next several chapters. The distinction between analytic and synthetic will figure prominently in chapter two, as it is central to Kant’s treatment of definition, and in particular to his own favored definition of religion, “The recognition of all our duties as divine commands.” Indeed, only with this distinction in hand can we fully appreciate the load Kant wants his definition to bear. The second distinction, that between a priori and a posteriori, shows up in the discussion of reason in chapter three and of non-conceptual experience in chapter four. For the moment, I only note that, apart from the spatial model of concepts, neither distinction can be given even an initial formulation: An affirmative judgment is analytic if and only if the subject concept can be analyzed as a conjunction of concepts one of which is the predicate concept. If the predicate concept can be connected to the subject concept independent of experience then the judgment is a priori; if only by experience, then it is a posteriori.
It would be hard to find two more controversial distinctions than these in the recent philosophical landscape. But for better or worse we have, in fact, been trading on what looks very much like the analytic/synthetic distinction for the last several pages. It is, for example, natural to read Doniger as saying that the concepts of “caste,” “karma,” “renunciation” and “the worship of various gods” stand in a special relationship to the concept “Hinduism.” If we go on to say—what seems unavoidable—that this special relationship consists in the fact that they are parts of the larger concept then we are at the threshold of Kant’s distinction: to say of any Hindu that he is, for example, a member of no caste is not just odd or in some way non-standard. It is self-contradictory. For Lincoln, to speak of a religion of unconnected individuals would amount to saying, “All religions must have a communal structure but this religion does not have a communal structure.” Similarly, for Tweed, it makes no sense to see an institution as a religion that has no effect on joy and suffering or that makes no reference to suprahuman forces. Note that Doniger, Lincoln, and Tweed are not ruling these things out “by definition.” At stake here is more than the merely verbal. Rather, they are ruled out on the grounds that they attempt to assert something contradictory about the objects to which the definitions are said to apply. When we turn from analytic to synthetic we find a similar naturalness. That is, all three authors seem to be tacitly relying on the fact that some of a concept’s parts (some of its sub-concepts) are entirely contingent and can be thought otherwise. Thus, for Doniger, location presumably falls under this heading: the concept “Hindu” need not include the Indian subcontinent; any continent will do. For Lincoln, we may presume that the general concept “religion” exhibits some degree of neutrality as to the kind of practice it includes; we could, presumably, talk about religions that lack initiation rituals. And for Tweed, one presumes the suprahuman forces may or may not be gendered.
Calling on Doniger, Lincoln, and Tweed, I have been echoing what many commentators—friendly and other—have said, namely, that Kant’s distinction between analytic and synthetic structure does have a certain naturalness that is reflected in our ordinary talk, including, apparently, our talk about religion. But of course we all know that talk is cheap. This or any such distinction may be both natural-seeming and bogus. Kant appears to want to base the distinction on the thought that some sub-concepts cannot be meaningfully denied of the subject concept. For the moment this must be a promissory note, to be redeemed in the next several chapters. Let us finish our overview of Kant’s theory of concepts with a final point about the internal relations between parts of concepts.
So far we have seen that, on Kant’s model, concepts are composed of conjunctions of two or more partial concepts, that they are rules for the formation of judgment, that they admit of no lowest species, and that their parts admit of analytic and synthetic and a priori and a posteriori connection. A further bit of conceptual microstructure requires Kant’s claim, new under the philosophical sun, that the connection holding between certain sub-concepts is at the same time synthetic and yet a priori. Thus, if I am going to judge that a person is a member of this or that caste, or refer to any sort of communal structure, or impute joy or suffering I must presuppose that all of the implicated substances will endure, interact causally with the surrounding substances, and reflect some extensive and intensive magnitudes. Here for the first time we meet with the fundamental purport of Kant’s transcendental idealism. That substances—say, a piece of clothing or a ritual implement—are permanent cannot be read off the relevant concept. That is, I do not contradict myself by thinking of Humean vestments that pop in and out of existence each time I look at them; in this respect it is fundamentally unlike trying to speak of a religion that has and, in the same sense, does not have a communal structure. In other words, “permanence” attaches to “clothing” synthetically and not analytically. At the same time, Kant argues that such things as articles of clothing and ritual implements can be known to be permanent (and causally interrelated, and more) independent of experience, or a priori. He thinks he has shown (in the arguments of the Principles of Pure Understanding, A158/B197–B294) that it must be so if experience is to be possible. Since experience apparently is possible, it must be so. Thus we arrive at propositions which are at once synthetic and a priori—facts about the world which can be known independent of experience.
To review: We have now seen three ways in which the parts of concepts can be related to one another: through experience, through an as-yet unclarified notion of self-contradiction, and in virtue of “a third something,” namely, the possibility of experience (A155–56/B194–95). No doubt history is littered with claims for this or that piece of knowledge on which experience has been said, wrongly, to depend. A conspicuous example is Kant’s claim that physical space must conform to Euclidian axioms and postulates. The issue will take center stage in chapter three, when we look at the role of reason in coming to higher and higher-order generalizations about such objects as “world religions,” and again in chapter five, with the claim that Kant isolates the subject—the activity of reflection—from its social and institutional context.
So much for a preliminary sketch of Kant’s spatial theory of concepts. Of course a satisfactory account of general concepts will require more resources than Kant provides in the Critique and the Logic (or, to my knowledge, elsewhere25); what Kant says about the structure of concepts requires augmentation of various kinds. I have so far merely sketched Kant’s picture of discursivity. At its heart lies the spatial theory of concepts—persons reacting to what strikes them by following general, hierarchically structured rules. But, incomplete as it is, I want to suggest that the spatial model of concepts provides a useful platform for thinking about the vicissitudes of “religion” and the current varieties of skepticism about it. Indeed, our discussion may already be seen to have several preliminary implications.
1. Emptiness: The fact that we can construct a web of hierarchically nested, inferentially articulated general concepts does not by itself guarantee them legitimacy. In Making It Explicit, Brandom distinguishes between weak, strong, and hyperinferentialist accounts of conceptual content.
The weak inferentialist thesis is that inferential articulation is necessary for specifically conceptual contentfulness. The strong inferentialist thesis is that broadly inferential articulation is sufficient for specifically conceptual contentfulness—that is, that there is nothing more to conceptual content than its broadly inferential articulation.
Brandom explicates strong inferentialism along the lines suggested by Michael Dummett; that is, as involving the circumstances (including the perceptual circumstances) and consequences (including those consequences that bear on action) of conceptual activity. By contrast, the hyperinferentialist holds that we can create conceptual content from inferential articulation even absent such circumstances.26
On this schema, I am taking Kant to be an early, and probably the first, proponent of a strongly inferentialist view of conceptual content. His inferentialism is, to be sure, idiosyncratic. For example, Kant places his inferentialism in a non-conceptual context; he holds, roughly, that our thoughts bear on the world by way of states of mind lacking conceptual content—a view that has invited connections with several strands of modern religious thought. These issues will occupy us in chapter four. The present point is that we may understand the underlying critical thrust of the first Critique to be directed against what Brandom is calling hyperinferentialism: Absent legitimate application to the world, our concepts are meaningless. Hence the famous slogan, “thoughts without content are empty” (A51/B75). Since one of Kant’s targets was rational theology, the present question becomes whether the shift from God-talk to religion-talk is shift enough. It is easy enough to say that Christianity is composed of further subspecies of concepts—Roman Catholicism, Eastern Orthodoxy, Protestantism, etc., the last of which is composed of such further subspecies of concepts as Presbyterian, Methodist, Baptist, etc., the last of which is composed of such further subspecies as Full Gospel Church Fellowship, Cooperative Fellowship, Southern Convention, etc.—easy enough to flesh out inferential connections between them, and easy enough to note that there is in principle no absolutely lowest concept, just as there is in principle no absolutely lowest species. But none of this by itself is any more meaningful than is talk about the infinite, eternal, omniscient, omnipotent, wholly good deity; this is the point of having to exemplify our concepts in concreto. Thus, Kant is with contemporary critics of “religion” insofar as they insist that knowledge of the world cannot be had from concepts alone.
On the other hand, in Kant’s denial of a lowest or infima species, we have the suggestion that there is a limit to the precision and fixity that we may expect from the definitions of any of our general terms. This will be a central theme of chapter two. And this general lowering of expectation must affect how we take the widespread complaint—given forceful expression, for example, in Timothy Fitzgerald’s Discourse on Civility and Barbarity: A Critical History of Religion and Related Categories—that “religion” has no “fixed,” “essential,” or “uncontested” meaning.27 True enough—but then, on the Kantian view we are currently developing, neither does any other general concept on which depends the evident success of our everyday linguistic commerce. Enough is enough, enough is not everything.
2. Reification: The root of what is sometimes known as the “reification” problem does not lie in our concepts. Thus, Kevin Schilbrack, expressing what seems to be a widespread sentiment, identifies “the implication that the reality described is a bounded and even static object” as among “the most important conceptual problems with the concept ‘religion’.” As he sees it, the
process of reification occurs at the level of generalizing about “religion” and then it can occur again at the level of individual religions: the idea of Hinduism, for example, also suggests that there is something ahistorical and monolithic about being Hindu, and the same is true of the labels of “Buddhism,” “Christianity,” and so on. This opposition to religion thus draws on the widespread philosophical rejection of essentialism.28
No doubt there are accounts of concepts which do invite the conclusion Schilbrack describes, but recognizably Kantian ones are not among them. Kant’s opposition to reification has both metaphysical and epistemological aspects. Metaphysically, Kant holds that objects of experience can never be rendered completely determinate (“static,” “ahistorical,” “monolithic”)—for that would require establishing whether, for every property “x,” the object in question is either x or not-x. As Kant puts it, “To know a thing completely, we must know every possible [predicate], and must determine it thereby, either affirmatively or negatively. The complete determination is thus a concept, which, in its totality, can never be exhibited in concreto” (A573/B601). Epistemologically, Kant’s denial of an infima species is again in play: We can never know that we have come to anything more than a provisional stopping point in the conduct of inquiry.
In the Critique, Kant also denies the absolute determinability of objects of experience on grounds having to do with embodiment. Thus, in the Metaphysical Exposition of the Transcendental Aesthetic (A22/B37ff), Kant argues that I must be able to locate objects of possible cognition at some distance from where I find myself. And that, he reasons, limits (a priori) my knowledge of objects to those that may be encountered in space and time. Of other possible predicates I must remain silent. This line of thought will be of signal importance in our discussion of non-conceptual content in chapter four. That objects of experience are not absolutely determinable marks a fundamental break with the Fregean tradition in the philosophy of language, a point that will be decisive in chapter six when we turn to linguistic meaning.
3. Holism: Brandom has argued in several places that we owe to Kant the thought that one must have many concepts in order to have any. Kant
takes it that the contentfulness of concepts essentially involves rational relations with other concepts, according to which the applicability of one provides reasons for or against the applicability of others. Applying one concept can oblige one to apply another, preclude one from applying a different one, and permit one to apply still others. Concept use, then, involves a normative dimension. Kant understands concepts as the rules that ultimately determine the correctness of such inferential moves.29
We will return to this passage in chapter six, in the context of a wider discussion of linguistic meaning. For now, I note only that Kant’s spatial model of concepts provides the background against which these rational relations should be set, requiring as it does that every concept be composed of a hierarchically nested and indefinitely extended series of sub-concepts, even as it itself forms part of a larger genus. The connections between concepts and sub-concepts are constitutive of them; that is, they are essential to each concept having just the content that it has. Further, we have seen that, for Kant, these relations can be analytic or synthetic. Thus, for Doniger, application of the concept “Hindu” obliges one to apply the concept “caste,” precludes one from applying “monotheistic,” and permits one to apply “lives in North America.” For Lincoln, “religion” obliges one to apply the concept “discourse,” permits (I take it) the sub-concept “text” (and the further sub-concepts “devotional literature” and “epic poem”) and precludes “solitary.” In chapter six (section III) I will be arguing that Kant’s holism about concepts can illuminate the kind of holism about interpretation we find in the work of Donald Davidson and others, including Brandom; work that has lately been so successfully applied to topics in the academic study of religion including critiques of the cognitive science paradigm, insider/outsider puzzles, the category of the primitive, conceptual relativism, and religion and democracy. The methodology of linguistic interpretation held little interest for Kant, but it is no surprise that his holism about concepts should be relevant to it; no surprise that the structure of concepts should constrain the methodology of interpretation.
For my purposes, Brandom’s assessment of Kant’s theory of concepts is particularly illuminating. While appreciative of the holistic, normative thrust of Kant’s theory, Brandom remarks that Kant “punted many hard questions about the nature and origins of this normativity, of the bindingness of concepts, out of the familiar phenomenal realm of experience into the noumenal realm.”30 I will return to the substance of this charge in the conclusion. At that point I will be arguing, contrary to Brandom, that Kant’s best view is that “phenomena” and “noumena” are parts of the more general concept of a thing. For now, I would like to see Brandom as placing Kant’s theory of knowledge within the realm of religious reflection. He thus illustrates how Kant’s epistemology can become so stubbornly intertwined with religious thought, for of course appealing to the noumenal realm is just what Nietzsche and Geuss find to be continuous with the basic purport at least of Christianity: the downgrading of the real in favor of the ideal. Brandom is right to take the bindingness of concepts as the leading thread of this discussion; the question is whether he misconstrues the nature of the constraint. In thinking about religion (at least in this and the next chapter) we are interested in empirical concepts (“religion” prime among them). I take it there is no danger of slipping off into the noumenal realm here; I take it as uncontroversial that Kant’s treatment of empirical concepts stays within Brandom’s “familiar phenomenal realm.” Religious illusion—in this case the lure of a bogus noumenal realm—beckons when we turn to certain a priori concepts, for now we are confronted with rules the following of which make possible any rule-following at all. As Kant puts it, such a rule “has the peculiar character that it makes possible its ground of proof, namely experience, and must always be presupposed in this” (A737/B765; see also A148/B188). The problem, then, is how to inquire into the bindingness of a concept on which, by hypothesis, that very inquiry must rely. Kant’s official position is that such an inquiry is dialectical, illusory. Indeed, to inquire into, as Brandom says, “the nature and origins” of this bindingness is to fall victim to the same kind of illusion as Kant finds in rational theology, namely, that which comes with the attempt to apply concepts beyond their proper sphere.31 This will be the central issue in chapter four between Kant and Schleiermacher.
4. Essentialism: We will not be in a position to appreciate the complexity of Kant’s views on the topic of essentialism until we have discussed his views on definition (chapter two) and on experience (chapter four), but one of the central points follows directly from the account of discursivity presented above. Suppose we claim, in an essentialist spirit, that “Islam is a religion of power,” or that “Water is H2O,” or that “God is love.” On the version of essentialism that interests me at present, we are making a claim about the nature of the object as it is in itself, quite apart from how it is affecting us or could ever affect us. We are claiming to know the real nature of the thing, and not merely how it appears. By contrast, as we have seen, Kant holds that our knowledge of objects depends on being affected by them and, in turn, on our ability to make judgments about them—judgments whose components include concepts nested into genus and species, and whose parts exhibit the kind of internal structure we have been discussing. There is no room here for “direct” (that is, non-conceptual) access to the nature of objects as they may be apart from all possible human cognition, no room for the Aristotelian faculty, nous, whose job it is to grasp essences apart from sense perception. Now contemporary essentialists have tended to separate out the metaphysical from the epistemological in Aristotle’s picture. Thus, Kripke and, at an early stage, Putnam, argue that statements about essences are, when true, metaphysically necessary but epistemically contingent. This formulation may seem to bring to Aristotle a recognition of our finite status as knowers while preserving reference to objects as they are in themselves and not as they appear to us. But this element of modesty does not go deep enough to satisfy a Kantian account of empirical cognition. Even so modified, the essentialist reserves the right to refer to objects in some more direct way than by means of our discursive, conceptual apparatus—the only way open to us. He is, after all, still purporting to say something about the essential properties of the water as encountered in that glass or of Islam as encountered in that country; namely, that they obtain as a matter of metaphysical necessity. But the notion of cognizing objects apart from the application of general terms is just what Kant thinks is incoherent.32
What of the third of my examples, “God is love”? If the claim is made in the same spirit as the remarks about water and Islam—as about an object of some sort—then the same moral holds: no reference except by general concepts. But suppose we are speaking of the concept “God”? Then essentialism comes to us in a different and more familiar form. Recall that, for Doniger, it is self-contradictory to speak of a Hindu of no caste; that Lincoln can make no sense of a religion of isolated individuals; and that, for Tweed, it is not merely odd to speak of a religion that has nothing to do with joy or suffering. The invitation here is to distinguish sharply between essentialism as applied to objects and as applied to concepts. In the next chapter we will see that Kant’s account of concepts undermines the former even as it embraces the latter.
5. Reason: When one looks at how scholars of religion conceive of their subject matter, one is struck by a certain tension—by movement both toward greater generality and toward greater specificity. It is, I would say, a conspicuous feature of the recent literature. Schilbrack documents one instance of this tension, citing Daniel Dubuisson’s recommendation that we abandon “religion” in favor of the broader “cosmographic formation”; it is, says Dubuisson, more “truly universal” and “less Western.” On the other hand, Timothy Fitzgerald would have us replace “religion” with three narrower terms: ritual, politics, and soteriology.33 Commenting on this back and forth, Schilbrack cautions, “There seems to be a conceptual need not only for the small, particular, local terms but also for generalized abstractions, for the imagined whole of which the small terms are parts.”34 Indeed. But what is the nature of this “need”?
The simultaneous movement toward greater generality and specificity is, according to Kant, the hallmark of Vernunft, reason. Coming to grips with its legitimate use and limitations calls for nothing less than a critique of its pure form. In chapter three we will explore the Kantian claim that the drive toward both generality and specificity derives from none other than the spatial theory of concepts; from, that is, the nature of the general terms we bring to experience, formed, as they are, from nested relationships of genus and species. If this is so, then the tension documented by Schilbrack would make itself felt even absent all considerations of power, politics, or indeed any other material force. In the Brandomian terms introduced at the outset of this chapter, the tension between generality and specificity has, on this picture, nothing to do with dynamics—the relationship between concepts and the world. It has only to do with statics—the internal structure of our discursive apparatus. Schilbrack remarks that, “Both the more inclusive and the more particular replacement strategies therefore seem to become entangled in the fact that no language is pure.”35 Kant’s view is that the entanglement is deeper than anything having to do with language, that it is traceable instead to the structure of the concepts we find ourselves constrained to employ. Only presupposing their use do questions about the purity or impurity of language even come into view.