Spontaneity Unchained: An Essay in Darwinian Epistemology
[from Idealismus
als Theorie der Representation?, Ralph Schumacher, ed., (Mentis Verlag;
According to Kant, all our cognitive achievements arise from the conjoint exercise of two complementary faculties, a “capacity for receiving representations (receptivity for impressions)” issuing in “intuitions”, and the “power of cognizing an object through these representations (spontaneity of concepts)”. (A50=B74)[1]
If the receptivity of our mind, its power of receiving representations in so far as it is in any wise affected, is to be entitled sensibility, then the mind’s power of producing representations from itself, the spontaneity of cognitions, should be called the understanding. … To neither of these powers may a preference be given over the other. Without sensibility no object would be given to us, without understanding no object would be thought. Thoughts without content are empty, intuitions without concepts are blind. It is, therefore, just as necessary to make our concepts sensible, that is, to add the object to them in intuition, as to make our intuitions intelligible, that is, to bring them under concepts. (A51=B75)
Our spontaneous understanding is thus chained to our receptive sensibility. On this account, concepts without intuitions are empty. But this is not an essay about empty concepts.
How do concepts and intuitions interact to produce cognitions? More precisely, how does a unitary cognitive act issue from the collaboration of two such prima facie different elements? On the face of it, the prospects for giving a useful answer to this question do not seem promising. In Kant’s story, intuitions are based on the receptivity of impressions; concepts, in contrast, on the spontaneity of thought. “[The] only use which the understanding can make of these concepts is to judge by means of them.” (A68=B93) To put it briefly, a Kantian concept is a predicate of possible judgments. Such a concept is consequently only mediately related to the objects which are themselves represented by the corresponding subject terms of such judgments and, for us, ultimately immediately represented only through intuitions. Thereby arises, to put it in Kantian terms, a homogeneity problem. “In all subsumption of an object under a concept the representation of the object must be homogeneous with the concept; in other words, the concept must contain something which is represented in the object that is to be subsumed under it.” (A137=B176)
John McDowell’s recent book Mind and World[2] moves within the framework of these Kantian ideas. McDowell sees in Kant’s story of the amphibious character of cognition, essentially involving both receptivity and spontaneity, the promise of an escape from an “interminable oscillation” between the idea that in order to be genuinely empirical our thoughts must somehow be grounded by the world, i.e., from outside the conceptual realm, and the recognition that what lies completely outside the realm of understanding, although it might indeed cause our empirical thoughts, cannot function as a reason for holding them.
On the one hand, the disturbing image of a coherent system of beliefs that floats free of genuine empirical constraints tempts us to embrace some version of the traditional foundationalist “Myth of the Given”.
[If] our freedom in empirical thinking is total, in particular if it is not constrained from outside the conceptual sphere, that can seem to threaten the very possibility that judgements of experience might be grounded in a way that relates them to a reality external to thought. … The putatively reassuring idea [of a Given] is that empirical justifications have an ultimate foundation in impingements on the conceptual realm from outside. So the space of reasons is made out to be more extensive than the space of concepts. (5-6)
On the other hand, when we come to recognize that the classical idea of Givenness is a myth, i.e., that “extra-conceptual impingements from the world” or a “brute impact from the exterior” (8) can play no justificatory role, we are correlatively drawn back to a coherentism according to which, in Davidson’s words, “nothing can count as a reason for holding a belief except another belief”.[3] But then
[it] can seem that we are retaining a role for spontaneity but refusing the acknowledge any role for receptivity, and that is intolerable. If our activity in empirical thought and judgment is to be recognizable as bearing on reality at all, there must be an external constraint. (9)
“We can dismount from the seesaw,” McDowell proposes, “if we can achieve a firm grip on this thought: receptivity does not make an even notionally separable contribution to the co-operation [between receptivity and spontaneity].” (9) Less darkly framed, we can free ourselves from perpetually alternating between these two unstable temptations by recognizing that
the conceptual contents that sit closest to the impact of external reality on one’s sensibility are not already, qua conceptual, some distance away from that impact. They are not the results of a first step within the space of reasons, a step that would be retraced by the last step in laying out justifications, as that activity is conceived within the dualism of scheme and Given. This supposed first step would be a move from an impression, conceived as the bare reception of a bit of Given, to a judgment justified by the impression. But it is not like that: the conceptual contents that are most basic in this sense[4] are already possessed by impressions themselves, impingements by the world on our sensibility. (9-10)
McDowell formulates his central claim in a wide variety of ways. In the book’s Lecture IV, “Reason and Nature”, for example, it emerges as the thesis that “conceptual capacities [are] already operative in actualizations of sensibility” (67). Alongside “actualizations of sensibility”, we find apparently equivalent references to “actualizations of receptivity” (66) and to the “impressions that independent reality makes on one’s senses”. (67) All three expressions are used to designate loci for the exercise of “conceptual capacities”, loci in turn characterized as bearers of “conceptual content”. And McDowell asserts more or less indifferently that “empirical thinking is rationally constrained by experience” (67), that “conceptual content is already borne by impressions that independent reality makes on one’s senses” (67), and that “spontaneity permeates our perceptual dealings with the world, all the way out to the impressions of sensibility themselves”. (69)
What McDowell’s story explicitly has no place for is what we might call “raw” impressions, i.e., a stratum of pure receptivity, representations that are in no sense conceptual. In his first “Afterword”, he criticizes a contrary picture which, as he sees it, Sellars and Davidson share, namely, that
Receptivity figures in the explanatory background of circumstances that belong together with evolving world-views in the order of justification. But receptivity itself cannot rationally interact with spontaneity …. (141)
Whereas Sellars and Davidson distinguish (raw) “sensory impressions” from the (conceptually-structured) “appearances” which constitute the flow of experience, McDowell seeks to identify them. By his lights, we must be able to say
that the impressions of the world on our senses, the deliverances of our receptivity, are—as such—the appearings (or at least some of them) that, as Davidson and Sellars agree, can innocently be taken to belong together with our world-views in the space of reasons, since they are already in the space of concepts. (141)
It is important to appreciate that, although he agrees with McDowell regarding what he also calls ‘experience’,[5] Kant aligns himself with — indeed, is the source and inspiration of — the Sellarsian-Davidsonian picture we have just been discussing. His own solution to his homogeneity problem involves a long and fairly complicated tale regarding what he calls “schemata”, a tale that includes an account of the central role of imagination in the “threefold synthesis” of perceptual experience,[6] but the nub of the matter is that, if “intuitions” are to be capable of contributing something to (perceptual) experience, then they must already be, so to speak, pre-processed, and, indeed, pre-processed by an exercise of the faculty of concepts:
The same function which gives unity to the various representations in a judgment also gives unity to the mere synthesis of various representations in an intuition; …. (A79=B104-5)
Among Kant’s various uses of the word ‘intuition’, then, there is one — one to which McDowell also appeals — according to which an intuition presents more than just the bare being-affected by something entirely outside the realm of spontaneity, but rather already contains something conceptual. But intuitions in this sense can hardly be equated with the Kantian counterparts of McDowell’s “actualizations of [pure] receptivity” (66) or the “impressions that independent reality makes on one’s senses”. (67)
Rather there is also another strand in Kant’s story that adverts to “impressions” which are, in our sense, raw. He writes, for example, that
We can … with regard to [a priori] concepts, as with regard to all cognition, seek to discover in experience … the occasioning causes of their production. The impressions of the senses supplying the first stimulus, the whole faculty of cognition opens out to them, and experience is brought into existence. That experience contains two very dissimilar elements, namely, the matter of cognition [obtained] from the senses, and a certain form for the ordering of this matter, [obtained] from the inner source of the pure intuition and thought which, on occasion of the sense-impressions, are first brought into action and yield concepts. (A86=B118)[7]
And, in an important footnote, he argues for the indispensability of imagination as “a necessary ingredient of perception itself” on the grounds that, although the senses supply us with (raw) impressions, “something more than the mere receptivity of impressions is undoubtedly required” to generate “images of objects”, “namely, a function for the synthesis of them”. (A120, fn.)
On this reading, Kant himself distinguishes between raw impressions, the immediate products of “actualizations of receptivity”, which fall entirely outside the conceptual realm, and pre-processed intuitions (i.e., appearances), products of the “synthesis of apprehension”, which stand in a necessary relationship to the understanding. This makes Kant’s picture of our relationship to the world, like that ascribed to Sellars and Davidson, one, in McDowell’s terms, “from sideways on”:
We find ourselves always already engaging with the world in conceptual activity within … a dynamic system. Any understanding of this condition … must be from within the system. It cannot be a matter of picturing the system’s adjustments to the world from sideways on: that is, with the system circumscribed within a boundary, and the world outside it. That is exactly the shape our picture must not take.” (34)
Our picture must not take that shape, McDowell argues, precisely because such a picture
cannot depict anything genuinely recognizable as an understanding of a set of concepts with empirical substance. These supposed concepts could be bound up with impacts from the world only causally, not rationally; and … that leaves their status as concepts with empirical substance, potential determinants of the content of judgements that bear on the empirical world, a mystery. (35)
Brusquely put, then, McDowell concludes that we need to acknowledge “a rational connection between intuitions and thoughts”. (68, my emphasis)
The world itself must exert a rational constraint on our thinking. If we suppose that rational answerability lapses at some outermost point of the space of reasons, short of the world itself, our picture ceases to depict anything recognizable as empirical judgement; we have obliterated empirical content altogether. (42-3)
McDowell’s goal is to unite the experiencibility of (empirical) reality with its independence of our thoughts, and his strategy is to deny the existence of any “boundary” between such reality (“the world”) and “the conceptual sphere”.
Although reality is independent of our thinking, it is not to be pictured as outside an outer boundary that encloses the conceptual sphere. That things are thus and so is the conceptual content of an experience, but if the subject of the experience is not misled, that very same thing, that things are thus and so, is also a perceptible fact, an aspect of the perceptible world. (26)
McDowell thereby aligns himself with the problematic early Wittgensteinian thesis that the world consists of facts (Tatsachen) or states of affairs (Sachverhalten), i.e., of constituents which themselves intrinsically possess propositional form, that is, the form of a judgment.[8] As he himself recognizes, however, this ontological thesis threatens to undermine the historically hard-won distinction between semantic intelligibility and the quite different sort of intelligibility correlative to natural lawfulness.
The conceptual sphere does not exclude the world we experience. To put it another way: what we experience is not external to the realm of the kind of intelligibility that is proper to meaning. But in so far as what we experience includes merely natural facts, this can look like a call to regress into a pre-scientific superstition, a crazily nostalgic attempt to re-enchant the natural world. (72)
How then should we conceive the relationship between our rational spontaneity and the world of nature? McDowell considers three “styles of response” to this question. Two of them, he ultimately rejects: on the one hand, the sort of “bald naturalism” that proposes to “reconstruct the structure of the space of reasons out of conceptual materials that already belong in a natural-scientific depiction of nature” (73), and, on the other, an anomalous monism in the Davidsonian style, on the face of it equally congenial to Sellars’ “scientific realism”, which settles for the ontological claim that “the very things that satisfy the sui generis concepts … whose applicability signals the presence of spontaneity, are already in principle available to an investigation whose concern is the realm of law”. (74-5) The remaining alternative, however, i.e., his own, is prima facie disturbing. McDowell proposes to abandon the equation of nature with the realm of law. In the place of a naturalism of (natural) lawfulness, he suggests, we need to recognize a “naturalism of second nature”.
We need to recapture the Aristotelian idea that a normal mature human being is a rational animal, but without losing the Kantian idea that rationality operates freely in its own sphere. The Kantian idea is reflected in the contrast between the organization of the space of reasons and the structure of the realm of natural law. … [We] need to see ourselves as animals whose natural being is permeated with rationality, even though rationality is appropriately conceived in Kantian terms. (85)
If we can successfully recover the requisite, properly robust, idea of second nature, McDowell claims, “we can keep nature as it were partially enchanted, but without lapsing into pre-scientific superstition or a rampant platonism”. (85)
But just how is this supposed to work? A nature that is “as it were partially enchanted” is suspiciously similar to a woman who is “as it were only a little bit pregnant”. Once we abandon the equation of nature with the realm of law, we lose our grip on any positive conception of nature. To put it differently, McDowell’s strategy seems to leave us with no suitable contrast for the notions of nature and natural intelligibility. In particular, a nature which includes Bildung — “having one’s eyes opened to reasons at large by acquiring a second nature” (84) — among the natural means for the actualization of innate human potentialities is a nature that a fortiori includes, and so cannot meaningfully contrast with, culture.[9]
This is not to deny that we have available various non-conceptual notions of “meaning”. There is, for instance, Grice’s “non-natural meaning” (meaningNN) — “Those spots meanNN measles” — and, drawing on Gadamer’s distinction between “a merely animal mode of life, in an environment, and a human mode of life, in a world”, (115) McDowell himself is at some pains to develop the picture of the sort of “meaningfulness” embodied in structures of what Gibson called “perceptual affordances”, viz. “a succession of problems and opportunities, constituted as such by [immediate] biological imperatives”. (115) But McDowell also makes it clear that “mere animals do not come within the scope of the Kantian thesis” (114) as he proposes to understand it:
[The] objective world is present only to a self-conscious subject, a subject who can ascribe experiences to herself; it is only in the context of a subject’s ability to ascribe experiences to herself that experiences can constitute awareness of the world. … It is the spontaneity of the understanding, the power of conceptual thinking, that brings both the world and the self into view. Creatures without conceptual capacities lack self-consciousness and—this is part of the same package—experience of objective reality. (114)
The availability of non-conceptual notions of meaning and meaningfulness, then, is simply beside the point. On McDowell’s view, it is conceptual meaning that is embodied in “the layout of reality itself”. (26) His “partially enchanted” nature is suffused with the forms of judgment per se.
That nature, in its own way, answers to the forms of judgment, i.e., to the Categories, is, of course, also a central Kantian contention. Like McDowell, that is, Kant ultimately rejects the idea of a boundary between the conceptual realm and nature. But where McDowell’s “relaxed naturalism” abandons the equation of nature with the realm of law in favor of a conception of a nature “partially re-enchanted”, Kant holds fast to the traditional conception. His strategy is rather “transcendental idealism”, i.e., to distinguish, so to speak, nature from reality in general, or, better, things as they are for us from things as they are in themselves. Of things as they are in themselves, i.e., apart from the conditions of possible experience, Kant insists, we can have, not only no empirical concepts, but in fact no contentful concepts at all. Kant thus erases the boundary between nature and spontaneity by locating the former’s lawfulness within the latter:
[The] order and regularity in the appearances, which we entitle nature, we ourselves introduce. We could never find them in appearances, had not we ourselves … originally set them there. (A125)
However exaggerated and absurd it may sound, to say that the understanding is itself the source of the laws of nature, and so of its formal unity, such an assertion is none the less correct, and is in keeping with the object to which it refers, namely, experience. (A127)
Kant immediately proceeds to reassure us that “certainly, empirical laws, as such, can never derive their origin from pure understanding”. But, he continues,
all empirical laws are only special determinations of the pure laws of understanding, under which, and according to the norm of which, they become possible. (A127-8)
There is no boundary between spontaneity and nature as the realm of law, in other words, because our conceptual understanding is the source of the lawfulness of natural laws.[10]
Needless to say, transcendental idealism of this stripe holds no attraction for McDowell. Indeed, he argues, if we hold to any naturalism that equates nature with the realm of law, we cannot avoid the conclusion that
since the operations of sensibility are, as such, natural goings-on, considered in themselves they can only be intuitions without concepts. (88)
But since such intuitions without concepts — what we earlier called “raw impressions” — cannot ground any empirical judgment, we are immediately returned to one pole of the potentially “interminable oscillation” which constituted our original predicament. The “deep-rooted, but … non-compulsory influence on our thinking that accounts for the predicament”, McDowell proposes, is precisely a naturalism of natural law, “the naturalism that leaves nature disenchanted”. (85)
McDowell’s attempt to thread a path between the mythic foundationalism of a conceptual given and a rational coherentism that floats free of genuine empirical constraints thus apparently brings us to a choice between two differently unpalatable pictures. On the one hand, we have Kant’s transcendental idealism, which, although arguably not equivalent to the monstrous “two-world” story of empirical and “supersensible” reality which haunts McDowell’s own reading of Kant,[11] does carry the potentially disturbing implication that what is genuinely mind-independent is incognizable as such.[12] On the other hand, we have McDowell’s own vision, surely no less disturbing, of a “partially enchanted” nature, within which “the dictates of reason are there anyway, whether or not one’s eyes are opened to them”. (91) But is there really no further alternative here?
Let us take a more careful look at the “sideways on” picture that McDowell finds in both Davidson and Sellars. On this view, “raw” impressions
have an indirect epistemological significance, in that without them there could not be such directly significant circumstances as seeing that things are thus and so, or having it look to one as if things are thus and so. But it is only in that indirect way that impressions enter into the rational responsiveness of empirical thinking to the course of experience. We can have an innocent interpretation of the idea that empirical thinking is rationally responsive to the course of experience, but only by understanding “the course of experience” to mean the succession of appearings, not the succession of impressions. (141)
[What] we cannot say without confusion … is that sensory impressions, impacts of the world on our senses, impose rational demands on our empirical thinking. Or if we can say that, it is only by dint of packing some complexity into “impose”. Perhaps a sensory impression causes it to appear to a subject that things are thus and so, and the appearing has implications for what the subject ought to think. But sensory impressions themselves … cannot stand in rational relations to what a subject is to think. (139-40)
But just what is supposed to be wrong with this picture? Well, according to McDowell,
… if we follow Sellars and Davidson in distancing the appearings from the impressions, saying what they will let us say does not entitle us to find no philosophical mystery in thought’s bearing on the world. (142)
The curious double-negative here is evidently intended to suggest that the Sellarsian-Davidsonian picture ultimately fails to come to terms with a significant philosophical problem, a problem that the picture is in fact both taken to address and mistakenly judged to resolve. What is crucial to the picture is the idea that “appearings are just more of the same kind of things beliefs are: possessors of empirical content, bearing on the empirical world”.
And now we cannot make the question “How can beliefs (say) have empirical content?” look any less pressing by talking about a rational interplay between appearings and beliefs. The question is really “How can anything have empirical content?”, and it is no good just helping ourselves to the fact that appearings do. (142)
What is worrying McDowell, then, is “the mystery of empirical content”. But just what is it for a thought to have “empirical content”, and just what is “the mystery”? It is surprisingly difficult to extract useful answers to these question from McDowell’s text. We have, of course, just encountered the notion of “thought’s bearing on the world”, but if what makes a judgment or belief empirical is supposed to be its “bearing on the world”, it is not clear why a causal coupling of conceptually-structured appearings to non-conceptual “raw” impressions of the sort envisioned by Sellars and Davidson will not serve. For it is not clear why a subject’s conceptual responses to such “raw” causal impacts of the world cannot have evidential implications regarding what that subject ought to believe about the world, and that, surely, would be “bearing” enough.
We come closer to McDowell’s worry, I think, when we find him speaking of “something correctly or incorrectly adopted according to how things are in the empirical world” (139) or “the idea of a stand as to how things are, correctly or incorrectly adopted according to the layout of the world”. (142-3) What such formulations suggest is that what is at issue is not thought’s bearing on the world but rather the world’s bearing on thought. McDowell’s leading idea seems to be that a belief or judgment has “empirical content” just in case “the world” is the ground of its correctness or incorrectness. His leading thesis, correlatively, is that nothing can be such a ground of correctness or incorrectness unless it itself has a conceptual structure, the logical form of a system of facts. The thought that he takes from Kant is that the “impingements of the world on our sensibility” which constitute our receptivity rationally ground our empirical judgments and beliefs by being instances of “the world’s making itself manifest to us”.
When we are not misled by experience, we are directly confronted by a worldly state of affairs itself, not waited on by an intermediary that happens to tell the truth. (143)
On McDowell’s view, then, what is wrong with the “sideways-on” picture common to Kant, Sellars, and Davidson is that it cannot account for the normative bearing of “the world” on our empirical thoughts. “Nothing can count as a reason for holding a belief except something else that is also in the space of concepts.” (143) Hence
… the idea of an interaction between spontaneity and receptivity can so much as seem to make it intelligible that what results is a belief, or a system of beliefs, about the empirical world …only if spontaneity’s constructions are rationally vulnerable to the deliverances of receptivity, (138‑9)
and so only if the deliverances of receptivity per se already exercise rational constraints on belief. No “empirical content” (empirical significance) without rational grounding; no rational grounding without rational vulnerability; no rational vulnerability without rational constraint; hence no empirical content without rational constraint. So runs McDowell’s argument.
McDowell suggests that the Kantian-Sellarsian-Davidsonian “sideways on” picture casts impressions precisely in the role of potentially, but only contingently, truth-telling “intermediaries”. At this point, however, he arguably fails to keep properly in view the key distinction on which Sellars, Davidson, and Kant before them all insist, namely, the distinction between non-conceptual (“raw”) impressions (Kant’s “sensations”) and conceptually-structured appearings (Kant’s pre-processed “intuitions”, embodying the forms of judgment).
Consider, for example, McDowell’s apparent concession that the “sideways on” view does not completely remove impressions from the domain of epistemology, since
[the] way impressions causally mediate between the world and beliefs is itself a potential topic for beliefs, and these beliefs can stand in grounding relations to other beliefs. Consider a belief that credits an observable property to an object. In the context of a rationally held theory about how impressions figure in causal interactions between subjects and the world, such a belief might be rationally grounded in a belief about an impression. One might be justified in believing that the object has the property by the fact that one has an impression of the type that is, according to one’s well-grounded theory, caused in suitable circumstances (for instance the prevailing illumination) by the object’s possessing that property. (144)
But notice the slide here from the prima facie unproblematic claim that a belief to the effect that an observable object has a perceptual property can being rationally grounded in a belief about an impression to the claim that such a belief can be justified by “the fact that one has an impression” of a suitable sort. On the Sellars-Davidson view, however, the latter claim makes sense only if “the fact that one has [a certain sort of] impression” is understood on the model of, e.g., “the fact that the object over there looks orange to one (here and now, in this red light)” — the justificatory reasoning at issue proceeding from a premise to this effect via generalizations regarding the ways that variously colored objects look in various conditions to a conclusion about the actual color (e.g., yellow) of the object in question — but that is a fact about a conceptually-structured appearing, not a fact about a “raw”, non-conceptual, impression. The fact that one is having a “raw” impression, in contrast, is a fact regarding the manner in which one is being causally affected by the world and, while reference to such a fact can enter into a correct (theoretical) causal-explanatory account of conceptually-structured perceptual appearings, it does not, as such, carry any immediate consequences regarding the veridicality of those appearings. Indeed, on Sellars’ story, a central point of “Jones’s” theoretical introduction of (non-cognitive, “raw”) impressions is to account for what veridical and non-veridical (cognitive) perceptual appearings have in common.[13]
In fact, the only way to avoid falling back into some version of the myth of the Given at this point is to recognize that, on the Sellars-Davidson “sideways on” view, any particular judgment to the effect that one is (then and there) having a certain sort of (raw) impression is available only as the conclusion of an explanatory (theoretical) inference from premises which will include beliefs about the ways in which things (then and there) appear to one. The epistemological bearing of theoretical beliefs regarding the way in which such (raw) impressions causally mediate between the world and perceptual judgments, in other words, is necessarily holistic and indirect.
In contrast, such beliefs would figure directly in theoretical explanations, for instance, of why the fact that one is spontaneously inclined to believe that, e.g., there is something orange before one’s eyes is any kind of reason at all to believe that there is something orange before one’s eyes. And once such a general theoretical framework is in place, it can be further deployed, in conjunction with premises regarding de facto “boundary conditions”, in the explanation of the soundness of justificatory reasonings of the sort McDowell has in mind, from premises regarding the way things appear in determinate objective circumstances to conclusions regarding the actual properties of objects.[14]
On McDowell’s view, however, we are still left with a “mystery”. Such a “merely” causal story, he insists, cannot account for the “empirical content” of our judgments and beliefs about the world, i.e., for precisely the sort of conceptual content by virtue of which they stand in those logical-evidential relationships constitutive of “the space of reasons”.
I want to compare this thought with another: A “merely” causal story cannot account for teleological functionality, i.e., for the sort of “purposiveness” exhibited in the biological world. The eventual fate of this thesis played a central role in the “disenchantment” of nature. The Divine providence and design once thought to be so clearly evidenced in the “great chain of being” simply evaporated in the light of a new post-Darwinian understanding of biological characteristics and relationships as the necessary outcomes of a continuous causal interaction between fortuitous random mutations and de facto environmental circumstances. The key insight of evolutionary theory, of course, is that the sort of manifest orderliness which appears to be a product of construction constrained by rational purpose, planning, or design, can be a byproduct of selection, arising from the blind but lawful operations of mere “brute” causal mechanisms.
I want to suggest that what we need to get past the two unpalatable pictures of McDowell’s partially enchanted nature on the one hand and Kant’s incognizable transcendental reality on the other is just this Darwinian turn. We need, that is, to understand empirical content as “constrained” by the world, not constructively but selectively.
This Darwinian model requires both a source of variation and an independent locus of selective pressure sufficient to induce differential rates of survival and reproduction. In the biological case, the locus of variation is the genotype; the locus of selective pressure, the environment. The genotype proposes, and the environment disposes. But, of course, the environment does not select this or that genotype as such. The organism-cum-genes interacts with the environment only causally. In particular, the genotype’s “proposals” take the form of inheritable traits. Such traits are, so to speak, a “middle term”, causally mediating between environmental impacts and the genetic “deep structure” which they express. This causal interaction is evolutionarily “successful” to the extent that the altered genotype is transmitted to the next generation, i.e., to the extent that the genes are able to express and carry out their own evolutionary biological function.
The most straightforward way of mapping this model onto our present problematic is surely to associate variation with conceptual spontaneity and selection with sensory receptivity. Our creative understanding produces novel theoretical representations of the world in whose terms we then respond to its causal impacts on us. Spontaneity proposes, and receptivity disposes. But here, too, we must avoid the idea that receptivity selects this or that family of theoretical representations as such. In this instance, it is precisely appearings that form the correlative “middle term”, mediating between the “raw” causal impacts of the world in impressions and the systems of concepts which they express. The interaction between a novel theory and the world is epistemologically “successful” to the extent that the new system of representations is able to “save the appearances”, i.e., to express and carry out its causal-explanatory epistemological function.
To take this Darwinian step, however, requires that we explore more carefully the thesis, common to Kant and McDowell, that empirical concepts without intuitions must be empty. In particular, we will need to critically engage a certain picture of how empirical concepts are supposed to depend for their contentfulness on intuitions, for a spontaneity too tightly constrained by an already-conceptual receptivity could hardly serve as the independent source of variability required by the Darwinian model. To put the point in terms congenial to McDowell’s critique, if empirical concepts without corresponding intuitions must be empty, it is hard to see how any genuinely novel concepts arising from exercises of spontaneity could be full-bloodedly empirical. Absent rational control from the world, a system of theoretical concepts could, at best, possess the epistemic virtues of coherence. What they could not have, however, McDowell insists, is “empirical content”, i.e., a locus of correctness in the world. This is precisely McDowell’s criticism of Davidson:
Davidson embraces … the renunciation of rational control from independent reality. He thinks a merely causal, not rational, linkage between thinking and independent reality will do, as an interpretation of the idea that empirical content requires friction against something external to thinking. But it will not do. Thoughts without intuitions would be empty, as Kant almost says; and if we are to avert the threat of emptiness, 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. The items that were meant to be thoughts are still without intuitions in the relevant sense, and so empty. Davidson manages to be comfortable with his coherentism … only because he does not see that emptiness is the threat. He thinks that the only point of wanting a rational connection between intuitions and thoughts is reassurance that we are justified in endorsing the thoughts, as if we could take it for granted in any case that they are thoughts, that they possess content. But if we do not let intuitions stand in rational relations to them, it is exactly their possession of content that is put into question. (68)
But what is it for thoughts to be with or without intuitions “in the relevant sense”? How, that is, are empirical concepts supposed to depend for their contentfulness on intuitions? McDowell’s text is not particularly forthcoming on this point, but once we explicitly ask that question, Kant’s answer at least is clear:
An empirical concept arises from the senses through comparison of the objects of experience. From the understanding it receives only the form of generality. — The reality of such concepts rests on the real experiences out of which, in accordance with their contents, they are created. (JL,99)
In order to make a concept out of representations, one must … be able to compare, reflect and abstract, for these three logical operations of the understanding are the essential general conditions of the production of any concept whatsoever.[15] (JL,102)
Far from being an independent source of variability, concepts derived from “experiences” by such “logical operations” could not confront our receptivity with anything genuinely new. But how could concepts that are not chained to intuitions in this “abstractionist” way possibly be empirical?
Kant himself contrasts empirical with arbitrarily constructed concepts:
Arbitrarily constructed concepts are mathematical. …
Note: … — All empirical concepts must … be regarded as constructed concepts whose synthesis is not arbitrary but rather empirical.
Since the synthesis of empricial concepts is not arbitrary but rather empirical and, as such, can never be complete (because one can always discover additional marks of the concept in experience), empirical concepts can also not be defined.[16] (JL,153-4)
It is a crucial feature of Kant’s view, however, that even such arbitrarily constructed mathematical concepts are ultimately chained to intuitions, and indeed, to empirical intuitions.[17] Clearly, then, if we are to make room for any concepts genuinely without corresponding intuitions, we must in some way significantly part company with Kant.
To make progress at this point, it will prove helpful to draw upon Michael Friedman’s outstanding book, Kant and the Exact Sciences.[18] What especially serves our purposes is Friedman’s argument that Frege’s polyadic quantification theory finally liberated pure geometry from its Kantian dependency on empirical intuitions. According to Kant, geometry depends upon the temporal process of obtaining intuitions by “drawing lines in thought”. (A162-3=B203)
The mathematics of extension (geometry) is based upon this successive synthesis of the productive imagination in the generation of figures. This is the basis of the axioms, which express a priori the conditions of sensible intuition …. (A163=B204)
Friedman argues, in contrast, that the Fregean logic of quantifiers for the first time makes possible a complete axiomatization of geometry, including a theory of dense linear order that entails the existence of infinitely many points and in terms of which the difference between enantiomorphs, e.g., a right-hand glove and its left-hand counterpart, can explicitly be formulated. Geometrical concepts — and indeed mathematical concepts in general — can in this way be implicitly defined. And that, in turn, creates logical space for a plurality of geometries and, correlatively, for our contemporary distinction between pure and applied geometry, and, more generally, between pure and applied mathematics.
On this picture, the Peano postulates, for example, collectively constitute the mathematical concepts zero, integer, and successor by fixing a truth value for every pure number-theoretic proposition in which they occur. In short, we also find in Frege’s work the basis for an entirely new understanding of concepts. Fregean concepts are products not of abstraction but rather of construction. A concept secures its content, that is, its predicative sense (Sinn), through the logical relationships among the propositions in which it non-trivially occurs, that is, through the inferences for which its (non-vacuous) occurrence makes a difference. Whether it is then possible to give such a concept “an object to which it may be applied” is, on Frege’s view, a further question, namely, the question of reference (Bedeutung).
One can perhaps concede that a grammatically well-formed expression … always has a sense. But that does not yet tell us whether a referent corresponds to that sense. … The expression “the least convergent series” has a sense; but one can prove that it has no referent by showing that, given any convergent series, one can find another less convergent that nevertheless converges. One’s grasping a sense, then, does not with certainty secure one a referent.[19]
Within a mathematical system, the question of the reference of a concept will receive a positive answer insofar as one can provide an existence proof,[20] a negative answer, if one can prove non-existence. Otherwise the question remains open. One can adopt various attitudes towards a system’s “primitive” concepts which occur in the axioms or postulates — e.g., zero, successor, and integer. Someone who emphasizes the difference between pure and applied mathematics, for example, might be inclined simply to equate the question of the existence of appropriate referents with that of the consistency of the axioms, and for present purposes, in fact, that will do. Our problematic concerns empirical concepts.
Applied mathematics is, however, a useful halfway house. We can contrast, for example, our contemporary notion of applied geometry with what we might call “operational geometry”, within which such terms as ‘point’, ‘line’, and ‘right angle’ would be definitionally introduced in procedural and observational terms (involving, e.g., stretched strings and lines of sight) and form the subjects of inductively confirmable empirical generalizations, e.g., that the diagonal of a square is always about 1.4 times as long as one of its sides. Contemporary physical geometry, in contrast, avails itself of an inferentially autonomous mathematical framework in formulating laws regarding the space of empirical objects, namely, (some form of) pure geometry, whose fine-grained richness of inferential texture, e.g., the fact that each of its one-dimensional “lines” contains a nondenumerable infinity of zero-dimensional “points”, precludes the direct translation of its theorem-sentences into generalizations framed in terms of operationally-definable concepts. Pure geometry becomes (applied) physical geometry when its mathematical vocabulary is appropriately correlated with the procedural idioms of observation, measurement, and experiment in a way which allows, for instance, for the inferential recovery (as theorems) of suitable counterparts to inductively confirmable generalizations, e.g., that the ratio of the diagonal of a square to one of its sides is Ö2.
Formally at least, we can repeat this trick by contrasting, for example, “operational” with “theoretical” mechanics. The primitive notions of operational mechanics, e.g., ‘mass’, ‘distance’, and ‘time’, are introduced by procedural definitions logically connecting them with observation predicates and techniques of measurement, and again form the subjects of inductively confirmable generalizations. “Theoretical mechanics”, in contrast, can begin as “pure mechanics”, i.e., by constructing a set of axioms whose logical relationships may be supposed collectively to fix the senses of the primitive non-logical terms, e.g., ‘mass’, ‘distance’, ‘time’, occurring in them. Like physical geometry, such theoretical mechanics avails itself of an autonomous inferentially fine-textured mathematical substructure in a way which precludes the direct translation of its sentences into the observational language of operational mechanics. Theoretical, but not operational, mechanics, for example, allows for point-mass interactions and instantaneous velocities, and for the operations of differential and integral calculus which relate them to the shapes and motions of massive bodies in space. We can again, however, empirically interpret the “pure” conceptual substratum of theoretical mechanics by correlating the basic expressions of its axiomatics with their procedurally-defined homonyms in a manner which allows for the inferential recovery of counterparts to inductively confirmable generalizations of the observational framework.
Both
physical geometry and theoretical mechanics are “idealizing” theories. On the face of it, both admit of
straightforwardly “instrumentalist” interpretations as conceptual tools for
refining and extending generalizations framed in the observational vocabulary
of their operational counterparts.
Nevertheless, we can catch a glimpse of something important if we ask
how one system of theoretical mechanics, e.g.,
Although acknowledging the distinction between translation and correlation does introduce a conceptual distance between intuited objects and theoretical predicates, the empirical concepts which have so far found a place in our picture are still rather tightly chained to intuitions. On the one hand, we are no longer inclined to say that intuited items literally fall under the descriptive predicates of physical geometry or theoretical mechanics, in the sense that such items do literally fall under the corresponding operational concepts. On the other hand, however, the fundamental correlations requisite for the empirical applicability of the theoretical axiomatics crucially match theoretically primitive terms with primitive operational counterparts. In an important sense, then, although successive theoretical frameworks ultimately give us increasingly sophisticated representations of a system of lawfulnesses, they remain essentially the lawfulnesses of appearances. The objects with which such theories deal, that is, remain objects as they are for us.
The decisive step away from “abstractionist” concept formation is taken by theories which postulate new basic entities, and, indeed, unintuitable basic entities. The kinetic-molecular theory of gases, for instance, stands to the operational theory of gases and its “idealized” axiomatic-theoretical counterpart, ideal phenomenal gas theory, in essentially the same epistemological relationships as axiomatic Newtonian mechanics stands to operational and axiomatic Galilean mechanics. But whereas both operational and ideal phenomenal gas theory take gases as their fundamental entities, and temperature and pressure (for a spatially confined volume of gas) as fundamental parametric properties of those entities, related according to the ideal (Boyle-Charles-Gay Lussac) gas law, the axiomatic kinetic-molecular theory initially posits the existence of large numbers of imperceptibly small particles, called molecules, in continuous motion. In contrast to gases, the fundamental parametric properties of individual molecules are mass and velocity, and it literally makes no sense to attribute pressure or temperature to them.
A key difference between this family of theories and our previous examples, then, is that the successor axiomatics here cannot be empirically interpreted by correlating the terms designating its basic entities and their primitive properties directly with terms designating the fundamental entities and properties of its predecessors. Instead, the basic kind-predicate of the phenomenal theory, ‘gas’ is correlated with terms designating ensembles of molecules, and the primitive parametric predicates, ‘temperature’ and ‘pressure’, with defined predicates applying to such ensembles, viz., their mean molecular kinetic energy and the average force per unit area exerted on the walls of a container as a result of multiple impacts of their individual members.
I propose that the basic concepts introduced in this way by “constitutive postulational” theories are, in the relevant sense, genuinely novel, non-abstractive empirical concepts. In the requisite way, that is, they are “empirical concepts without intuitions”, and so suited to play a role analogous to random genetic mutations in the Darwinian model.
Now it will likely first be protested that the concept of a molecule is simply an extension of the intuitive concept of an object. Molecules, after all, are just like, for instance, billiard balls, only very much smaller. But it is crucial at this point to see that, although the concept of molecules is, indeed, governed by an analogy, the analogy according to which any such concept of unintuitable basic entities is introduced cannot be simply an analogy of proportion. This is strikingly clear when it comes to such quantum-theoretic entities as electrons or photons. But molecules already differ from billiard balls, not just in size, but fundamentally, for example, in being sorts of items that ex hypothesi not only are not individually colored but could not be.
The analogy according to which molecules are introduced is, in fact, an analogy of lawfulness. That is, molecules are introduced as non-intuitable items in the category of spatio-temporal continuant substances which resemble intuitable billiard balls insofar as they conform to the same (Newtonian) laws of motion and interaction, but as items whose intrinsic character is left unspecified. The analogy in question thus crucially supplies more than, for instance, the via negativa analogies of classical theology according to which we can have no positive conception of the infinite deity. It provides a positive, although in certain respects indeterminate, conception of its theoretical subject matter.
This theoretical intrinsic indeterminacy of molecules, however, does not stand in the way of our having a proper conception of them. An analogy of lawfulness no more requires that its pre-theoretical and theoretical subject matters share intrinsic, first-order non-relational, content properties than the analogy between the temporal order of events and the spatial order of points on a line requires that spatial points and temporal instants share any such properties.[23] This is not to deny that molecules must be conceived as having intrinsic content properties. In Kantian terms, given that molecules are posited in the category of spatio-temporal continuant substances, there must indeed be something which in relation to molecules is “the real in space” (cf. A173=B215), as color is the real in space in relation to visually intuitable objects. But it does not follow from this concession that the intrinsic content properties of molecules must themselves be intuitable, i.e., given in sensation.
Now Kant himself, of course, objects in principle to the notion that we can form such concepts of unintuitable categorial entities.
We cannot define any [category or principle] in any real fashion, that is, make the possibility of their object understandable, without at once descending to the conditions of sensibility, and so to the form of appearances — to which, as their sole objects, they must be consequently be limited. For if this condition be removed, all meaning, that is, relation to the object, falls away; and we cannot through any example make comprehensible to ourselves what sort of a thing is to be meant by such a concept. (A240‑1=B300)
But once we properly appreciate that similarity need not be understood in terms of shared properties, and that properties as well as particulars can resemble and differ from one another in determinate respects, nothing stands in the way of making comprehensible to ourselves by means of a model “what sort of a thing is to be meant” by the concept of a molecule.[24] As Sellars put it:
[Models] provide a basis for a more or less vague and open-textured reference to a framework of propositional functions which the predicates of a theory are to satisfy. They are specified as the functions which hold, with certain qualifications, of the predicates which apply to the entities of the model. I say “with certain qualifications” because the reference to a model is accompanied by … a “commentary” which eliminates specific functions from the analogy and modifies others. [These] relatively precise qualifications operate within the context provided by an intuitive grasp of the framework of the model which it would be difficult if not impossible to articulate in terms of an explicit list of postulates. (SRII, 168)
In contrast to “abstractionist” concepts, such constructed concepts of unintuitables are not constituted by features (Merkmale).
A feature is the aspect of a thing that constitutes a part of its cognition; or — equivalently — a partial representation, insofar as it is regarded as a ground of the total cognition. — All our concepts are accordingly features, and all thinking is nothing but representation through features. (JL, 64)[25]
Precisely because they consist of shifting clusters of features, argues Kant, empirical concepts “cannot be defined at all, but only made explicit”,
for since we can have in mind only a few features of a certain kind of object, it is never certain whether someone using a word to designate the same object is thinking, on one occasion, of more of its features or, on another, of fewer. Thus, within the concept of gold, one person can think, besides its weight, color, and ductility, its property of not rusting, of which another perhaps knows nothing. We make use of certain features only as long as they suffice for making distinctions. New observations, however, remove some features and add others, so that the concept is never contained within fixed boundaries. (A727-8=B755-6; cf. JL 153-4)
On Kant’s view, in other words, an empirical concept functions to pick out a kind of object by reference to its features, i.e., in the first instance, its intuitable properties. Such empirical concepts are essentially “open-textured”; no set of features can be fixed as necessary and sufficient for an object’s belonging to the correlative kind. Specific features can serve to distinguish objects of that kind from others, but which features are relevant to such distinctions can vary significantly from occasion to occasion. The “look” of gold, for instance, is notoriously insufficient to distinguish it from certain iron pyrites. Where we are entitled to assume that we have a suitable exemplar of the kind to work with, however, we can attempt to enhance our correlative concept by experimentally discovering further of its features, in particular, its lawful powers and propensities, a service that, for example, Archimedes is reputed to have performed for King Hiero II in aid of distinguishing pure gold crowns from those composed of gold and silver alloys. But such progress presupposes that we sometimes do know that we are dealing with a representative of a natural kind.
Kripke has notoriously argued[26] that the Merkmale of such Kantian “phenomenal” empirical kinds as gold are known only a posteriori and consequently do not form “part of the [corresponding] concept”. What he does not stress, however, is that it follows that we are always free to treat such phenomenal kinds as merely nominal. If there is no reason to suppose that what we take to be gold’s empirical features belong to it essentially, that is, then there is no empirical reason to suppose that gold has an essence at all, i.e., that all instances of what we take to be gold belong to one natural kind. That gold is a natural kind is not a consequence of empirical observation and experiment, but rather a presupposition of certain interpretations of such investigations, e.g., of treating such results as Archimedes’ as criterial for a presumptive sample to count as (pure) gold.[27]
As Kant recognizes, however, the constructed concepts of mathematics do admit of definition, and so, too, will our “arbitrarily” constructed empirical concepts. The concept of element 79, Au, for instance, is explicitly definable in terms of the fundamental concepts of the contemporary atomic theory of matter — concepts of such unintuitable entities as protons, neutrons, and electrons and their primitive properties (mass, charge, spin, etc.), in terms of which the periodic table of elements can be systematically built up. Such constructed theoretical kind-predicates will obviously have a core of non-defeasible descriptive contents. We can know a priori, for instance, that the nucleus of an atom of Au will contain 79 protons, that its electron configuration will be 2∙8∙18∙32∙18∙1, and so on. Constitutive postulational theories whose basic entities are introduced as unintuitable, that is, explicitly posit natural kinds as such, by specifying, in theoretical terms, (some of) their essential features. And here, precisely, we do not have to do with appearances. The objects with which such theories deal are not objects as they are for us.
But why suppose that such theories deal with actual objects at all? For what is crucially characteristic for constructed (Fregean) concepts, we may recall, is that their inferential relationships secure for them a content, that is, a sense, that is independent of questions of reference, i.e., of whether the concepts so constituted apply to or are true of anything. Earlier, too, we remarked that there was no obstacle to treating “idealizing” theories purely instrumentally, i.e., as devoid of literal ontological import. Why, then, not treat constitutive postulational theories in a similar way, saying only, for instance, that gases behave as if they were ensembles of molecules? Absent corresponding intuitions, in other words, why should we suppose that there actually are any such “objects” as molecules, atoms, protons, electrons, and the like?
Surprisingly, it requires only a conservative extension of Kant’s own story to validate this ontological step. For what, after all, according to Kant, is criterial for empirical judgments of actual existence? The Second Postulate gives a clear answer:
That which is bound up with the material conditions of experience (sensations) is actual. (A218=B266)
The crucial question now is surely, Bound up in what way? Kant’s own discussion is surprisingly liberal. The correct answer should not depend, for instance, on contingent facts about “the coarseness of our senses”:
The postulate regarding the recognition of things as actual does not, indeed, demand immediate perception … of the object whose existence is to be recognized. But it does require that the object be bound up with some actual perception according to the analogies of experience, which elucidate all real connection in an experience in general. … [For then] we can, in the series of possible perceptions under the guidance of those analogies, make our way from our actual perception to the thing. … Our recognition of the existence of things thus reaches where[ever] perception supplemented by empirical laws extends. (A225-6=B272-3)
Interpreting Kant’s allusions to the analogies of experience here broadly as a way of highlighting the causal-explanatory role of empirical laws allows us to achieve closure on our dialectic. For constitutive postulational theories are precisely in this line of work. The kinetic-molecular theory of gases, for instance, explains the lawfulnesses captured in operational or ideal gas theory — and, crucially, some of the failures of phenomenal gases to conform in certain conditions to such lawfulnesses — by telling us what such a gas is. That is, as Sellars put it, constitutive postulational theories
explain why individual objects of various kinds and in various circumstances in the observation framework behave in those ways in which it has been inductively established that they do behave. Roughly, it is because a gas is … a cloud of molecules which are behaving in certain theoretically defined ways, that it obeys the empirical Boyle-Charles law. (LT, 121)
Similarly, it is because gold is element 79, Au, that it is, for instance, a ductile, malleable, electrically-conductive metal. That gold is Au, that is, explains its empirical Merkmale, and, crucially, explains as well the belonging-together of those features as observational and operational indices of a single natural kind. The atomic theory of matter, in short, earns its ontological claim inter alia by explanatorily securing the natural kind status presupposed for the phenomenal counterparts with which its defined “ensemble” predicates are correlated.[28]
Constitutive postulational theories, in other words, are in the business of explaining appearances as such. They explain why perceivable objects are as they are for us — in terms of their characteristic features, inductive phenomenal lawfulnesses, and occasional departures from such lawfulnesses — by telling us, defeasibly, what those perceivable objects are in themselves. Unchaining concepts from their abstractionist dependence on intuitions thus allows us to accept Kant’s transcendental idealism without endorsing his incognizability thesis. And nature remains a thoroughgoingly disenchanted realm of law. And once it becomes clear that we can genuinely discover that our freely-constructed empirical concepts do the causal-explanatory work necessary to earn empirical ontological import, I suggest, McDowell’s objections to picturing our relationship to the world “from sideways on” also simply evaporate. Unchaining our rational spontaneity from its abstractionist dependence on intuitions is precisely what is needed to allow it full cognitive access to a genuinely independent reality, and nothing in principle stands in the way of its thereby ultimately achieving an adequate causal-explanatory understanding of the empirical conditions of its own operations as well. Empirical concepts without intuitions, then, need not be empty after all — provided that they can earn a cognitive standing with respect to such intuitions. And how they can do that, I have suggested, is something that we are already dialectically well situated to understand.*
 BACK |
Notes