1. Wilfrid Sellars and John McDowell both present themselves as latter-day Kantians. They both believe that Kant's rich conception of experience as a unity of sensibility and understanding is a key to a proper view of the mind's relation to and knowledge of the world. But they differ about almost everything else: how to think about sensibility, how to think about understanding, and how to think about their unity in experience. The issues are complex; a single paper can treat only a part of their dispute. My principal goal here is to examine John McDowell’s attempt to undermine Wilfrid Sellars’s two-component analysis of perceptual experience.[1] McDowell offers us a competing vision in which perceptual experience is said to be a matter of the “conceptual shaping of sensory consciousness.” Without worrying about who is the better interpreter of Kant, I will argue that McDowell’s proposal leaves us with an ill-specified and possibly empty notion of sensory consciousness itself and that it cannot be attractive to anyone who believes in the possibility of an empirical science of the mind, even if they acknowledge a transcendental dimension to our knowledge.
2. Kant’s distinction between the sensory and the conceptual is a major landmark in our understanding of experience. We are, however, still in the process of digesting it. Wilfrid Sellars makes the Kantian distinction crucial to his own analysis of experience, further refining and developing what he thinks is not yet satisfactory in Kant. Sellars’s conception of the relation between sense and concept requires a substantive notion of the sensory as, in some important sense, a contributor to and an aspect of experience independent from the conceptual. This means, at very least, that sensory episodes are non-conceptual in some significant sense and can exist in us without being “brought to concepts” — that is, without being conceptualized and without being conscious states. It also means that conceptual episodes lacking any sensory aspect must be possible. His “two-component” analysis of experience is central to Sellars’s impressive systematic philosophy.
3. Robert Brandom, one of the Pittsburgh philosophers influenced by Sellars, has expanded admirably the inferentialistic analysis of the conceptual that in Sellars remained largely in nuce[2]. But whenever possible, Brandom avoids talking about experience and the sensory. He seems to hope, I think erroneously, that employing the notion of a reliable differential responsive disposition frees him from having to say anything further about sentience. According to Brandom, I have reliable differential responsive dispositions [RDRDs] to a large number of things in the world and in my self which, when actualized, position me within an inferentially articulated conceptual system, which position can then be exploited to change, sustain, elaborate, or specify other positions in the system that I hold. In short, Brandom permits us to enter the inferentially articulated conceptual realm directly via appropriate RDRDs, and that is where his concerns end. But possessing such dispositions does not entail sentience, even remotely. A piece of iron has a reliable differential responsive disposition with regard to water: under normal conditions, it rusts in the presence of water. Brandom apparently thinks that the fact that experience actualizes RDRDs that are dispositions to believe or enter into some other intentional state is sufficient to distinguish perception from the rusting of an iron bar. He allows no room for sense impressions as occurrent states and, thus. no room for a robust notion of sentience per se. Brandom’s neglect of the sensory is, in my opinion, a serious shortcoming of his work, and it is certainly a significant departure from Kant and Sellars.
4. John McDowell, another Pittsburgher influenced by Sellars, cannot be accused of giving the notion of experience short shrift. He devotes numerous pages to it, and has taken significant inspiration from Sellars’s notion that experiences can, in some sense, “contain claims.” Yet beyond this and the idea that intentional states are standings in the logical space of reasons, McDowell does not accept much of the architecture that Sellars developed. In particular, McDowell has always questioned Sellars’s interpretation of the sensory. At one point, for instance, he labels Sellarsian sensations “idle wheels”.[3] McDowell seems to have several concerns about Sellars’s belief that sensation makes an independent contribution to and constitutes a distinguishable aspect of experience. He is, first of all, afraid that Sellars’s view makes it impossible to account adequately for the unity of experience,[4] and second, that permitting the sensory such independence is a way of submitting to the myth of the given.
5. Here I want to defend Sellars’s position concerning the nature of the distinction between sense and concept. I will focus on McDowell’s arguments, leaving Brandom for a later time.
6. What are the major considerations advanced in support of the idea that we need a substantive, independent conception of the sensory if we are to make sense of experience and human cognition? The literature contains two principal kinds of arguments: those based on fairly straightforward empirical considerations, and those arguments of a more transcendental hue.
7. The first, empirical kind of arguments arise “in the attempt to explain the facts of sense perception in scientific style” (EPM §7, in SPR: 132-33; in KMG: 211). In order to explain illusions, delusions, hallucinations, etc., it is thought necessary to impose a layer of internal states that mediate—causally, not epistemically—between the world and our perceptual beliefs. Thus, when it appears to one that there is a red ball over there, but there is no such object in the relevant location (even, perhaps, no object at all), the experience is accounted for by the posit of a set of sense impressions that would normally be associated in standard conditions with the presence of a red ball over there, but are now present despite the absence of the red ball. One’s taking the red ball to be there is a direct reaction to these sense impressions rather than to what is over there. It would be unconscionably odd if these internal states occurred only when appearances were deceptive, so once we’ve posited them to explain perceptual illusion, we have to believe that they are always present in perception.
8. But if they are always present in perception, wouldn’t such states constitute a “veil” behind which the world is sequestered? Some philosophers have bought this inference, but there is no need to accept it. A window or a lens can be a causal intermediary between me and the object I see. In proper conditions, the window or lens is transparent and permits, even enables, my direct visual access to the object I see. But glass can get dirty or break, lenses can be out of focus, and under such deviations from the norm, we become aware of the glass as an intermediary. In cases of veridical perception, sensations normally fade from view; they are then as if transparent, allowing the world to reveal itself to us.[5] Hallucination and illusion are the deviant cases that reveal the complexity of our normally direct perceptual access to the world.
9. It is important to note that not all perceptual error is to be accounted for by reference to sense impressions. Sometimes we misperceive: we do not look, listen, or taste carefully or closely enough, we do not notice something palpably there, or we jump to a perceptual conclusion, as it were. The idea is that there are two forms of perceptual error — one on the conceptual side, the other on the sensory side. Reference to sense impressions need not be the general, or even the usual, strategy employed to explain perceptual error. This is important to understanding the substantive nature of the inference to sense impressions; we are not led to posit them by some general consideration of perceptual error, so that finding some other way to account for perceptual error invalidates the posit. Sense impressions seem forced upon us precisely when there is perceptual error without apparent violation of the epistemic virtues.
10. Blindsight is a recently discovered phenomenon that twists things in the opposite direction: veridical reports in the absence of sensory consciousness. Thus we seem to have to acknowledge both (1) conceptual episodes that are not “shapings of sensory consciousness” yet are both true and directly responsive to the environment, and (2) shapings of sensory consciousness that are both false and yet not errors of immediate conceptualization.[6] If the empirical facts of experience were different—if perception were generally but not perfectly reliable, but was never subject to illusion, hallucination, or blindsight—we would have no argument of this empirical kind for believing in the presence in us of internal sensory states that are essential parts of experience. In that case we could locate all perceptual error in the exercise of our conceptual capacities. All illusion would be misunderstanding.
11. Transcendental arguments support a substantive notion of an independent set of sensory states or sense impressions, not as elements in a causal story about certain facts about experience, but as required for the objective validity of experience. Such arguments claim that sense impressions, though themselves non-cognitive, are necessary conditions for experience to have the objective purport without which it would not be experience, but merely something lived through. It is important to understand how such transcendental considerations differ from the causal, empirical considerations we have just addressed. Transcendental arguments concern, in Kant’s phrase, a quid juris question, a question of right, justification, or normative status, not of causation. They point to a condition that must be in place if something else is to have a certain normative status.
12. Such a transcendental approach to the sensory has gotten significant attention lately. In his Woodbridge Lectures, John McDowell claims that there is a significant shift in Sellars’s arguments for sense impressions from EPM to Science and Metaphysics, in fact exactly a shift from the causal/empirical justification for belief in sense impressions to a transcendental justification.[7] McDowell’s claim has focused a great deal of attention on this transcendental aspect of sensation from, inter alia, David Forman[8] and Susanna Schellenberg.[9]
13. The central issue here is whether sensations or sense impressions play an indispensable role in justifying the claim that our intentional states and, indeed, our experiences are of the world. One interpretation of this thought is this: If there is no external constraint from the world guiding our thought, our thinking could be only a “frictionless spinning in the void;” we would have no right to claim objective knowledge about or experience of the world in which we live. Indeed, if there is no outside epistemic constraint, then we would have no right to believe in any metaphysical externality: we would be left in an obnoxious subjective idealism. There seem to be two alternatives concerning the form of such constraint here: Either perceptual thought about the world is guided directly by the world itself, and veridical perception is just a case of our being transparently and directly open to the world itself, or perception is not a transparent and direct openness to the world, and there is some intermediary that guides thought and perception and ties our thought to a world which thought cannot access directly.
14. McDowell wants to persuade us of the former alternative, that it is the world itself that guides our thought in perception, a world which, in veridical perception, is transparently and directly open to us. Sellars, McDowell argues, espouses the alternative, the path of mediation. Again, there are two paths open to those who choose the mediated route. In the one case, the intermediary is an epistemic intermediary, something that is itself known immediately and already imbued with a normative status that it can transmit to contentful thoughts about the world. As McDowell well knows, this path is simply the Cartesian or givennist path that Sellars spent so much time working against.[10] The second mediated route denies that the intermediary itself possesses epistemic or conceptual status. It is a given only in a causal and innocuous sense. But McDowell then applies the pressure here: since these immediately given sense impressions have no epistemic or semantic status, how could they play any role in the justification of the normative status of our thoughts as true to the world? Sense impressions would be, McDowell thinks, idle wheels in the transcendental story about human knowledge.
15. We might ask why the notion favored by McDowell — that in veridical perception the world itself is open and transparent to us — is not itself a form of the given. McDowell’s answer is clear: “The Myth of the Given is the supposed idea of an availability for knowledge [...] that presupposes nothing about the knower except, perhaps, natural endowments, for instance sensory capacities—an availability for knowledge that presupposes no learning or acculturation.”[11] We do not have to learn to sense, according to McDowell, but we do have to learn to see, hear, and think about the world. Having learned to see and hear the world, however, what we see and hear is precisely the world. As McDowell says, our thought does not stop short of the facts. Perhaps, then, we should say that there is, even for him, a form of mediation in one’s current and occurrent grasp of the world that plays a transcendental role, but it is not the causal mediation of sense impressions: it is the historical mediation of our learning history. This history cannot be narrated entirely in the language of causal laws, for it is the history of our ascendence into the logical space of reasons, and a story that itself concerns issues of right, justification, and normative status. David Forman has argued that even so, we cannot tell that story without alluding to sense impressions, thereby providing sense impressions with an indirect transcendental role.[12]
16. My own argument is a variation on this theme. First, let me reiterate that causal intermediaries in the perceptual process need not be epistemic intermediaries. This is, I hope, simply obvious at this point. Second, Sellarsian sense impressions play no justificational role in the story to be told about particular empirical claims we make or the generalizations we base upon them, but there is a transcendental role for sense impressions in that other logical dimension in which observation reports rest on other empirical propositions. Employing the transcendentally important category of observation report is tantamount to making a commitment to being able to unpack the story of the causal underpinnings of empirical knowledge. The best story we have of the causal underpinnings of empirical knowledge include sense impressions as part of the story of how we perceive, so the empirical arguments for sense impressions have transcendental significance.
17. Let me try to spell this out in a bit more detail. Any system of empirical knowledge has to have room for items that play the role of observation reports or beliefs (I’ll call them ORBs). Roughly, this is the role of assertions or beliefs that purport to be
(Introspective reports or beliefs are not, on this characterization, observation reports or beliefs, if only because no sensory organ seems to mediate their occurrence. Appropriate adaptions would have to be made to characterize the role of introspection in a system of empirical knowledge.)
18. Now, there can be a meaningful role for ORBs in a system of empirical knowledge only if (1) there is a meaningful distinction between ORBs and other assertions and beliefs, and this can be the case only if there are some non-arbitrary standards or ways of justifying which assertions or beliefs are to be considered ORBs and which are not; (2) there are also non-arbitrary standards for distinguishing true or at least trustworthy ORBs from potential but defeated or unusable ORBs. There is no reason to think that such nonarbitrary standards for distinguishing ORBs from nonORBs or successful ORBs from unsuccessful ORBs can be specified apriori. We have to discover what is observable and which observations are reliable in the course of acquiring and refining our empirical knowledge.
19. The theory of sense impressions helps us make the distinction between the observable and the unobservable by specifying a set of base-level sensory states, the sense impressions. Sellars sometimes characterizes these in terms of what we sense of the object. We see the pink ice cube, and, ceteris paribus, we see of it its very pinkness. But we cannot see of it its very coldness, though we can feel of it its very coldness, for cold is a sensible proper to touch. We can neither see, nor feel, nor smell its disposition to melt. Sellars doesn’t think that we can observe only the proper and common sensible properties of objects, but these notions set limits to the observable: an object that possesses no sensible properties (e.g., a number), is not observable, and objects and complex properties are detected via combinations of the proper and common sensible properties.[13] The theory of sense impressions also helps us elaborate a theory about which ORBs are how reliable under which conditions, as we explore empirically the operations of our sensory mechanisms and their connections to our conceptual mechanisms. The theory of sense impressions in general fills a transcendental role in accounting for our generic ability to relate to an independent but causally related world, which we can then fill out with empirical content.
20. In order to ensure that we cannot mistake the sensory for an epistemic intermediary between us and the world we perceive, McDowell wants to reject Sellars’s two-component view of experience in favor of a conception of experience as unitary. According to McDowell, experience is conceptually shaped sensory consciousness. Thus, rather than Sellars’s notion that our perceptual thinking is guided by the sensory, McDowell says, “The thinkings that provide for the intentionality of perceptual cognitions are not guided by sensory consciousness, as it were from without. They are sensory consciousness, suitably informed”.[14]
21. McDowell claims that this is not an attempt to eliminate sensations. He insists we can still say everything we want to say about sensations, but that it would be a mistake to reify such talk and take sensations to be independent, substantive elements of experience — that would just return us to Sellars’s two-component view. Rather, McDowell understands the concept of sensation to be a product of abstraction from our conception of the unitary conceptualized sensory consciousness that is experience. We conceive of sensations by abstracting from the spatiality of intuitions, say, by starting with the concept of an intuition of a translucent pink cube, abstracting from its spatiality, and thus generating the concept of a sensation of translucent pink.[15] Thus, if we abstract from the representational aspect of experience, we can isolate the sensory aspect, but we should never lose sight of the fact that the concept of a sensation is essentially abstractive. I think McDowell’s notion can be approximated by an analogy: the motion of an object in space can always be analyzed into three distinct mutually perpendicular vectors, but there is no presumption that any of these vectors possesses distinct material or causal reality as real forces. The sensory can always be analyzed out from perceptual experience, but it would be a mistake to attribute it a distinct causal reality.
22. McDowell, however, owes us a bit more explanation. An intuition of a pink cube is an intuition of something in physical space. It is not hard to abstract from that intuition the reference to physical space, but that still leaves behind a reference to the spatiality of the visual field. Our commonsense conceptions allow sensations themselves to have spatial properties, such as shape or spatial relation to other sensations. A sensation of pink could be to the right or the left of a sensation of blue, even if both are afterimages without objective spatial presence. When, as McDowell would have it, we abstract from the spatiality of an intuition to generate the concept of a sensation, do we abstract from physical spatiality or from all apparent spatiality?[16]
23. Let me return, however, to trying to understand Sellars’s reasons for thinking there is a substantive conception of sense impressions implicit in the manifest image. There is a distinctive move that Sellars makes when thinking about ostensible perceptual states. I call this move the “sensible presence inference,” and it is exemplified by inferential moves like this:
X ostensibly sees a pink ice cube.
Therefore, something in some way pink and cubic is in some way present to X other than as thought of.
Sellars thinks this is a good inference.[17] Its exact status for him is a bit unclear, though. In one place, this connection between ostensible perception and presence-other-than-as-thought-of is characterized as a matter of “phenomenological assurance” (FMPP, I 88), and I am not sure just what that means for him.[18]
24. Sellars thinks, of course, that the context sensation of translucent pink only superficially resembles the context thought (or, more interestingly: perception) of a translucent pink cube. Perceptions are thinkings, and they possess their logical and cognitive powers because they exhibit the full-blown intentionality of thought. Or rather, it is because they have logical and cognitive powers that they possess full-blown intentionality. In Sellars’s view, perceptions also possess a different kind of directedness (or presence), and it is precisely this that the sensible presence inference tries to capture. Ostensible perceptions differ from mere thinkings and believings-in by virtue of the presence to the subject of something in a way that is other than as thought of. And Sellars, of course, admits that the question of what this something is and how it is present to us has been a thorn in the side of philosophy for centuries.
25. McDowell, however, has to deny the validity of this connection, because he thinks that the only form of presence is intentional, being believed in. He thinks that as soon as we talk of directedness (and talk of presence is just the complement to talk of directness), we have introduced intentionality. There is no other form of directedness. So the inference that in ostensible perception something is present to us other than as thought of cannot be a good one.
26. McDowell therefore offers us a different story. Sensation is not a separable element in perceptual experience, present to us in a different way from thought; it is a mere abstraction from such experience, arrived at by abstracting from the full intentional directedness of perception. Abstract from what perception ostensibly relates us to, and what is left is a description of “a mere modification of a subject’s state”,[19] that is, a sensation. Thus, there is no reason to believe that the directedness (or mode of presence) of sensation is different from that of perception or thought.
. . .[I]f we reach “of translucent pink” by dropping “a . . . cube” from “of a translucent pink cube” in what was a specification of the intentional content of an intuition, why should “of” change its character? Why not suppose this form for describing sensations of colour exploits—in a vestigial form—the apparatus of intentionality?[20]
The advantage of this view, at least as an interpretation of Kant, is that spatiality is essentially foreign to sensation, since the sensory is conceived of in abstraction from the objectively spatial. But, as I have noted, we need greater clarity from McDowell about the status of subjective spatiality and the shapes and relationships of sensations themselves. Kant treats spatiality and temporality alike; I am, frankly, simply puzzled about what this means for McDowell’s treatment of sensations.[21]
27. McDowell’s talk of sensation as a vestigial form of intentionality seems misleading. The vestigial is usually a trace or degenerate form of something, often left behind in some developmental process. As McDowell himself notes, sensation is something we share with the animals,[22] but to the extent that brutes do not have the capacity to move about in the logical space of reasons — the location of full-fledged intentionality — it does not seem appropriate to describe the directedness of which they are capable, sensational directedness, as a vestigial form of intentionality, for it is not a left-over or degenerate form of something achieved earlier, but a forestage or condition for something yet to be achieved. Calling it a kind of proto-intentionality would seem more apropos, but proto-intentionality is not yet intentionality, so that presumably would give too much to Sellars, who would not reject that description, given proper commentary. By calling it ‘vestigial’, McDowell hopes to get us to lump sensational presence in with intentional presence as of a kind, but when thought through carefully, it turns on him and seems to encourage the idea that there must be some form of presence prior to full intentionality.
28. Verbal points aside, McDowell is committed to there being two kinds of thoughts: those that are shapings of sensory consciousness, and those that are not. Veridical perceptions or Kantian intuitions are shapings of sensory consciousness; thinkings or believings without phenomenal qualities are not. But isn’t there a third category that would include illusions, hallucinations, etc.? Most people consider these to be states of, that is, shapings of sensory consciousness. For McDowell, however, they are thinkings that are not shapings of sensory consciousness; they merely seem to be such shapings. We have to conclude that sensory consciousness is shaped only in veridical perception. I earlier claimed that there are two possible locations for perceptual error: in our sensory system itself, or in the conceptual uptake of our sensory state. McDowell appears committed to there being really only one location of perceptual error: it must be in the conceptual uptake, because he does not recognize a sufficiently independent sensory system for mistakes to be located in it.
29. When thinking does not shape sensory consciousness correctly, it does not shape it at all; it only seems to shape it. What, then, are we to make of sensory consciousness itself? According to McDowell, it can be shaped by our concepts, but only in veridical experience; in other kinds of experience it only seems to be shaped by our concepts. But I have some difficulty with the idea of something that can be shaped, but only correctly. This would be like a kind of clay that can be molded only into aesthetically pleasing forms. My concern is that McDowell thus leaves us with an essentially empty conception of sensory consciousness. He invokes sensory consciousness to distinguish perception from abstract thought, but I don’t see that he leaves us a way to investigate it in its own right any more than we can investigate the horizontal or vertical components in the motion of objects in their own rights.
30. According to McDowell, Sellars’s interpretation of the Kantian notion of sensations takes them to be “inner episodes (or states) exhaustively characterizable by descriptions that relate them solely to the subject as modifications of its state”.[23] I am not quite sure what to make of this: McDowell seems to be attributing to Sellars an individualistic interpretation of sensation. McDowell gives a different formulation moments later: “for Sellars an episode (or state) that is a sensation is completely describable—so far as concerns what it is for consciousness—by descriptions that use ‘of’ only in [a] non-intentional way”.[24] This at least restricts the kinds of descriptions in question to those purporting to specify sensory content, but it seems to assume that sensations are always something “for consciousness,” which is also not a Sellarsian thought. McDowell’s individualistic interpretation of Sellarsian sensations immediately raises for him a question: why should we think of sensations as representations, since, as sensations, they relate us to nothing. McDowell chides Sellars for offering a conception of a kind of representation that represents nothing.
31. But it is, of course, just false that “In Sellars’s view, . . . the whole truth about an item that is a sensation is captured by a characterization that relates it solely to the subject as a modification of its state”.[25] Kant, indeed, describes sensations this way, but in Sellars’s view that is because Kant has not yet seen how to embed minds in nature as products of evolution. Now, in Sellars’s view the generic notion of sensation is itself a functional concept: sensations are states that play a particular kind of role in the behavioral economy of organisms, and that role cannot be adequately characterized without alluding to objects in, and states of, the world independent of the subject of the sensations. This claim should not be confused with the different claim, rejected by Sellars, that particular sensory states are functional states. Particular sensory states do typically have functional roles, but they are such that they play those roles in virtue of possessing some intrinsic, that is to say, non-functional property. The point is that the non-functional, intrinsic, qualitative character of a sensory state is not itself anything like “the whole truth” about the state. The episodes and states that are sensations are what they are in virtue of their participation in sensory systems that themselves have a long evolutionary history that installed those systems to produce such states to play that role in the mediation of the behavior of such organisms. Furthermore, we conceptualize our sensory states by locating them within a field of similarities and differences, our conception of which is modeled on the field of similarities-and-differences characterizing the sensible qualities of physical objects. For example, physical objects resemble and differ from each other in color; our visual sensations resemble and differ from each other in analogous ways. This homomorphism is an important part of the whole truth about sensations. There is abundant relationality in the Sellarsian conception of sensation.
32. Nevertheless, perhaps McDowell has his finger on something important about Sellars’s conception of sensation. Sellars is committed to what I call the restricted validity of “of”-elimination in sensory contexts. That is, Sellars is willing to make the move from
John has a sensation of red
to
John has a red sensation.
This move is of restricted validity because the move from
John has a sensation of a computer screen
to
John has a computer screen sensation
is an entirely different move. The idea is that there is a restricted set of terms, pretty much the terms for the proper and common sensibles, that, although developed with respect to sensations by analogy to their application to physical objects, nonetheless apply literally to sensation, not merely metaphorically. We have two isomorphic sets of predicates, one of which applies to physical objects and the other to sensory states. The analogy between the predicates of physical objects and the predicates of sensations, taken seriously, becomes a proto-theory that enables us to think about sensations in their own right. This is important, because sensations are entitative, that is, independent and substantive, precisely to the extent that they have characteristics in their own right. Thus, it is easy to see that “of”-elimination is not a trivial matter. Such a proto-theory can inspire empirical work investigating the how of our experience and cognition. On the flip side, “of”-elimination never works in intentional contexts. Intentional objects as such cannot be presumed to be entitative.
33. The sensible presence inference mentioned earlier is thus connected to “of”-elimination: the somehow redness of sensation is not supposed to be a merely intentional redness, a merely represented redness. It is an actual redness, a present redness, even if not the same redness present in physical objects. Sellars takes it that the manifest image is committed to the actuality of color and sound in sensation, and thus to both the sensible presence inference and restricted “of”-elimination.
34. As I understand him, McDowell rejects both the sensible presence inference and all “of”-elimination. There is no substantive reality to sensations. They are an abstraction from the unified complexity of experience. Abstractions that they are, sensations can play neither a causal nor a transcendental role in knowledge. They may seem to be elements of sensory consciousness, but McDowell’s view must surely be that sensory consciousness is prior to the sensations we isolate within it. Unfortunately, as we have seen, he says little about the sensory consciousness that is shaped by our conceptual episodes, and I fear that he empties the notion of any real content. All the interesting distinctions seem to be made in terms of conceptual form rather than the sensory clay it shapes. How, then, would an empirical scientist go about trying to understand the contribution of sensation to experience and knowledge?
35. I find the sensible presence inference and restricted “of”-elimination attractive in their own rights. This is one reason that I prefer Sellars’s picture. Furthermore, Sellars sketches a view that makes it sensible to think that the sensory could be a legitimate object of empirical investigation. Episodes and states that help us understand why we have the particular conceptual responses to ourselves and our environment that we do at a particular moment, that come in families that are homomorphic to the quality spaces inhabited by physical objects, and that can be used to account for certain kinds of perceptual errors are things that empirical science can go looking for. Why is it important that our transcendental reflections lead us to items we can envisage cropping up in empirical science as well? It is a principle of Sellars’s transcendental naturalism that transcendental structures must be reflected in causal structures, even if there is no reduction of the transcendental to the causal. I share the impulse to be as naturalistic as is compatible with recognition of the transcendental dimension of human being. Presuming a transcendental right to believe that we experience an objective world of causally connected objects independent of us is tantamount to undertaking a commitment to construct an objective story about our causal participation in the world that details the how of our relation to the world. As long as we are making progress on the objective world-story, we can retain our presumption of transcendental right. McDowell’s learning-plus-cultural-transmission story gets us part of the way by giving us the large historical canvass, but he denies our ability to penetrate the story of the how of knowledge at the sub-personal level. Sellars’s picture includes the large historical view, but permits us a research program aimed at understanding the how of experience at the sub-personal level. This is consonant with the naturalistic commitment that transcendental principles are reflected in causal structures. Given the vagaries of human experience, accepting a conception of substantive, independent sensory states that play a distinctive causal role in perceptual experience helps us to understand how cognition is possible and why it takes the shape it does. Such entities pose significant difficulties for us, but a coherent naturalism demands that we accept the challenge[26].