Should we really regard self-acquaintance as a relational matter? Is it really a matter of some sort of thing standing in a relation to itself? On the one hand, there is no special problem either logically or phenomenologically speaking with the idea of something relating to itself in this way. Appearance is appearance-to. That’s relational. There is no a priori reason why something could not appear to itself. It does not lead to a regress.[38] One should put aside misleading and question-begging spatial analogies—consciousness is not like a knife trying to cut itself. Advocates of self-acquaintance will claim, opposing one analogy with another, that it is more like a candle’s flame illuminating itself by emission while it illuminates other things by the reflection of its light; it does not require another candle flame for it to be illumined.[39] Moreover, one must remember to exclude from one’s mind the sort of objectification and description-based cognition that normally overlays the phenomenal manifold. We are talking about the sphere of immanence, to speak Husserlian, and not about intentional objects or constituted objectivities given via Abschattungen. Again, we are talking about immediate self-acquaintance, not the representation of oneself as being such and such. It is indeed more like the emission of light than the reflection of light, if we must pick an analogy.
Nevertheless, even if we accept the relational construal and remember that it is an immediate and direct relation not mediated by concepts or descriptions, we still have a problem. It is not as if conscious episodes just happen to be self-manifesting. The property of being self-manifesting is not something that a thing can have and then not have—like changing coats of paint. It is of the very essence of a conscious episode. This is not an external relation to itself or one mediated by convention or history or anything else. Hence, it must have some set of intrinsic properties in virtue of which it is self-manifesting. Thus, the Heidelberg School, Michel Henry, and Dan Zahavi, I’ll concede, win on this ontological point. Dieter Henrich, Manfred Frank, Henry, and Zahavi have all maintained that self-manifestation could not be a relational matter (e.g., Henrich 1971, 1982; Frank 2002, 2007; Zahavi 1999; Henry 1973). And they are very close to being right. I think, however, that it is more accurate to say that even if it is a relational matter, it is not an external relation we are dealing with. So there must be something about the internal structure of consciousness that grounds the relation. In short, as Henrich and Frank have long said, there must be some intrinsic property in virtue of which episodes of consciousness (out of all other things in the world) are self-manifesting. What could this property be? Are we left with something that cannot be physical, or, even if it is physical, is nevertheless irreducible in some sense?
It may seem now that David Rosenthal is having his revenge.[40] In effect, I have been arguing against the extrinsicalist view—the view that something’s being conscious has to do with external relations the thing stands in—be those external relations to other mental states or external relations to historically distant states of affairs or to other parts of one’s cognitive apparatus. Now, to our chagrin, it seems we are left with something explanatorily basic. At this point we are left with two problematic strategies. We could go the panpsychist route (Strawson 2006): It’s no surprise that we're conscious if everything is! Or if, as I do, one thinks (after Locke in a similar context) that “every sleepy nod doth refute” this, we can hold that only certain physical complexes instantiate this particular property (or set of properties). This will mean either some form of property dualism or some form of identity theory (possibly with its “Harder Problem”; see Block 2002). If one does not want to be a dualist or a panpsychist, what can be said?
Here is the sort of approach that seems most attractive (to me, anyway). We want to hold that consciousness is indeed some sort of physical process. It’s not, however, just a matter of the satisfaction of some functional role. I think it also has a functional role. But it is not in virtue of playing that role that something is consciousness; rather, consciousness is suitable for that role because of its properties.[41] In principle, many different things could play that role (at least if we specify it entirely in behavioral terms). Or, at least, this is an open question. Consciousness has a functional role, but it is not to be identified with just any arrangement of elements that can play that role as causally and behaviorally specified. There is some special, distinctive physical process that is consciousness. It plays its functional role in virtue of its having the properties it does and not vice versa. But then does some version of Russellian Monism start to seem attractive (see e.g., Stoljar 2006, ch. 6 and Pereboom 2011, chs. 5 and 6)? Am I saying that the functional role is just being (contingently) satisfied by a (somehow) unified and self-manifesting group of qualia? Or something wild like that?
Here we play the same sort of trick we played when dealing with the Grain Problem. Consciousness is self-acquainted, but we are also, as Fumerton and Fales would say, acquainted with acquaintance; we are given givenness (Fumerton 1985, pp. 57–58 and Fales 1996, pp. 147–148). The relation does not seem complex or to involve many layers of relational structures. But we cannot infer from this appearance that it is in fact such a simple relation. Again, its not seeming complex does not, without controversial and implausible completeness assumptions, entail its being simple. Moreover, once we realize that normal consciousness involves a great many intricately related aspects—at least (non-contingently) differentiated unity and temporality, and (contingently) animation functions operating on a differentiated sensory manifold, iterations of these functions, pattern extractions, etc.—we have all the more reason to suppose that there is complicated machinery hidden from our introspective view. In fact, it will be noted in a Sartrean and Moorean vein, that consciousness, both as acquaintance relation and subject-relatum, seems mightily empty. Once we realize that Revelation theses fail, then we no longer need read this appearance as “consciousness qua acquaintance relation appears simple.” Rather, we read it as “consciousness qua acquaintance relation does not appear complex.” These are, in many cases, phenomenologically indistinguishable, but they are logically different.[42] The first reading, coupled with an infallibility thesis (or with just a strong presumption in favor of the deliverances of naïve introspection), leads to the view that acquaintance is simple. But the other requires a Revelation (or completeness) thesis to get the same result. Revelation is, again, totally implausible. And even if we were to assume infallibility, we have no a priori reason to favor one interpretation of the phenomenological data over the other—the “seeming non-P” vs. “not seeming P” formulation. We do, however, have plenty of a posteriori inductive reasons for preferring the latter: It does not seem complex, but it is (or at least could be for all we can tell phenomenologically).
Since we have an extremely limited resolution when it comes to penetrating into the nature of consciousness by introspective means, we are quite free to adopt another strategy. We can accept an a posteriori identity theory. Consciousness is identical to some sort of recurrent physical process unfolding in the brain. Fundamentally, what we get from introspection is a sort of structure and some irresolvables—the sensory qualities—that are like reflections of the materials in which the form or structure is instantiated. Since we have rejected Revelation (completeness) theses, we can accept that sensory qualities (and the acquaintance relation itself) are complex and involve layers of relations even though they do not seem this way (just as the headless woman[43] in the famous illusion does not seem to have a head—absence of appearance is transformed into the appearance of an absence; see Armstrong 1968 and 1973).
We can use what structure we are aware of, however, to build models to guide our search for the neural correlates of consciousness. One thing we see is that the (only apparently simple) acquaintance relation involved is such that whenever xRy, xRx; while it is not the case that if xRy, then yRy (in the case where y is a sensory quality or manifold thereof). And we have some idea of what the qualities in the manifold could be—e.g., limits of resolution or irresolvables operated on by a spatializing filter. We can also see that spatial projection, integration of multimodal information, temporality, and the modulation of attention are involved (along, of course, with more advanced things like intentional animation, cognitive filtering and reprocessing, and poise for action). We have a self-manifesting totality containing a unified and spatialized but differentiated manifold. Consequently, we do need to look for processes that can do information integration and binding, but that is only necessary, not sufficient. We need to look as well at processes that spatialize the multimodal (and multidimensional) information (see Williford et al. 2012).
This does not at all mean we are looking for a little room in the brain that has patches of red, yellow, blue, and green mental paint in it. Rather we must look for more abstract correspondences. In the case of the sensory qualities, we are possibly looking for higher-order relations between fairly complex structures, structures that can transiently be pulled into and “rendered” by the core process. Basically, this panoply of contrast-related irresolvables gets generated in a real-time and transient fashion, now occupying this virtual “location”, now occupying that, depending on a whole host of input factors (head orientation, background, conceptualization, etc.). These “locations” map onto (we hope) real physical space at a certain scale, but it is not a matter of finding a “bubble within a bubble.” It is a matter of an abstract correlation of structure. The isomorphisms (or homomorphisms) could be there even if the internal “space” of experience is entirely virtual, a kind of computational “movie in the brain” to use another phrase of Damasio’s. Assuming the principle that the positive and critically evaluated set of phenomenological descriptions gives us not just the way consciousness seems, but the way it in fact is, along with our identity postulate, we can be sure that something in the brain has a structure corresponding to this, no matter how transformed by “layers of abstraction” it may be.
What is more, self-acquaintance will demand that we explore models in which real reflexivity can be encoded. Hofstadter’s model is one of these.[44] But following D. Rudrauf and further encouraged by D. Bennequin, I have moved in the direction of considering projective geometrical models. There is no space to go into this here, but suffice it to say that there exist mathematical frameworks that allow us to conceptualize and investigate more deeply the self-acquaintance-related features of consciousness by considering the interplay of the space we project and the origin of the projection (see Williford et al. 2012 and Rudrauf et al. ms).
The goal of such work would be the refinement of mathematical models of the structure of consciousness. Upon the achievement of that end, we would then try to determine how such models could possibly be physically realized in the brain. Once we can say what the physically detectable signatures of such a realization might be, then we could one day meaningfully test such theories. Were we to verify the existence of such a structured process in the brain, explaining consciousness would reduce to explaining how the process is realized—what parts have to be in what order doing what and at what time scale.
It will always seem to be a brute fact, at some level, that consciousness is physical process X, however X gets fleshed out. But we’ll just get used to it, as long as there is some somewhat intelligible bridge (in this case provided by mathematical models) from the lived phenomenon to its brain correlate. We’ll get used to it just as we’ve gotten used to water being H2O. It could be that there will be multiple ways to implement such a process. Suppose, just for example, that it has to do with generating certain types of fields and that multiple substrates, not just brains, can generate and support the relevant sorts of fields. Then consciousness will be, to that degree, multiply realizable. Suppose it is a matter of realizing a certain computational organization. Then, in effect, implementing a certain program will be equivalent to being conscious; and if machines made from different substrates or with different architectures can run the program, consciousness will be multiply realizable in the sense of computational functionalism. Your particular consciousness then, as you know and love it, would be just the concrete running of the program in your particular brain.
We might wonder, in such a case, what it is to “run” a program or to “have” a certain structure or to “instantiate” such an arrangement or system of fields or whatever. Of course, this is a quite general metaphysical problem that we should not confuse with any problem specific to consciousness. However, given that we are acquainted with our own individual existence, it seems that somehow its instantiation makes its very instantiation available or manifest in some non-representational way. This is rather peculiar. If we are going to be physicalists who are nonetheless responsible to the phenomenology, however, this is what we have to accept, or so I have argued. Something is conscious if it has a certain internal structure and attendant dynamical profile. Being conscious is having that structure and profile. We will never be able to explain why that is the case because it is simply a confusion to think that identities like this admit of explanation; they can only be discovered (Papineau 2002, ch. 3). We must, of course, give evidence in favor of the relevant identity claim; uncovering such evidence is the goal of scientific research on consciousness. Our choice is between this sort of view and the view that there is something else, something non-physical that just is consciousness. Of course, we’d never be able to explain why that is the case either. So in the absence of compelling arguments for dualism or panpsychism, Occam’s Razor would lead us, as Smart pointed out so long ago, to embrace an identity theory.
The identity theory only adumbrated here would be neither a crude type-type identity theory nor a causal-role functionalist token-token identity theory where the realizers do not matter at all. Since any concrete consciousness is a marriage of form and matter (and the self-appearance of that marriage), and since there no doubt are physical constraints on what sorts of materials can be put into that form, we want to identify consciousness with neither a specific type of material (or “wonder tissue” in Dennett’s phrase) nor with an abstract, disembodied form that seems trivially realizable by practically any set of elements—since purely abstract isomorphisms may be a dime a dozen.[45] In other words, we need a non-eliminativist and non-idealist account of what it is to really realize a structure, instantiate a form, or, as the case may be, to really run a program or compute a function. To my knowledge, no one currently has such an account.
At bottom, this is just the old metaphysical problem of the Methexis—the relation of universals to particulars or of form to matter. When I am feeling optimistic, I imagine that I’ve reduced the problem of consciousness to another, more general (as well as ancient and probably insoluble) metaphysical problem. We may not know what it is for matter to really and mind-independently take on a certain form, but it is hardly an implausible metaphysics that says that this happens. It is arguably this type of metaphysical view that would best explain the success of applied mathematics, engineering, and the sciences: they are successful because the world really does have (or approximate) the relevant mathematical structures—these are in re structural universals. This seems to be a commitment of scientific realism. But perhaps we will never get beyond a rather crude operationalism when we empirically investigate such matters; perhaps the metaphysical nature of property instantiation will forever remain obscure to us. That should not, however, discourage us from carrying on such empirical investigations in the case of consciousness. Even if there will be a residual metaphysical mystery, it is a general one, not one specific to consciousness.
The main point here, and the concluding one, is that consciousness could be self-acquainted, where this is not a matter of external relations, and still some form of relatively non-mysterious (hylomorphic) physicalism could be true. One might balk at the idea that this would not be a matter of external relations, especially if we go the computational functionalist route. But think of it like this: If we are realists about the implementation of computational structures, then even though the structures involve parts and elements, there is still a unity to the pattern as implemented. It is, in a certain sense, an indivisible whole that is not just the mereological sum of its parts. Analogously, the circle has its own structure and characteristic properties even though it is made of points. What we really need, and may never have (but who knows?) is a theory that tells us when we have a real, concrete unified whole, (where this is not simply a functional or conventional characterization but is a matter of more basic physical relations) and when we have unities and wholes (and instantiations of structures and properties) that are only conventionally real.
Suppose then that we adopt a sort of realism about computational (or otherwise structural) wholes, which we have some independent reason to do. Circles have remarkable properties, qua circles, even if they are made up of points. Concrete circular things approximate these. Simultaneous cycles have certain number-theoretic properties just qua cycles regardless of what they are cycles of (e.g., reproducing cicadas and cicada predators, see Baker 2005). Likewise, for the concrete implementation of consciousness, it is surely the case that certain elements must be put into a certain arrangement, realizing a certain structure and dynamics. This would not mean, however, that consciousness as such is to be identified with either those elements or the arrangement abstractly conceived. Rather it is the concretely implemented organization of those elements qua whole. In virtue of being an instance of that form or structure, it has certain properties. One of these could be the property of being self-manifesting. That property could itself be a complex relational property having a certain unity. The account sketched here presupposes a certain realism about the instantiations of mathematical and computational structures—that there are determinate, mind-independent facts of the matter about this. We cannot go further into this rather large and complicated metaphysical hornet’s nest. Suffice it to say that a real, unified, concretely instantiated structure could, in a certain sense, be relational and have components even if it is, in another sense, an intrinsic property.