In characterizing consciousness, it is often said that there exists a first-person perspective or point of view associated with having phenomenal experience. On some construals of this perspective, the subject gains knowledge of, acquaintance with, or access to certain categorically private first-person facts, the phenomenal ‘what it is like’ of experience, or qualia (Nagel, 1974; Jackson, 1982; Chalmers, 1995a). It is supposed that such facts about experience (the redness of red, the painfulness of pain) are not reducible or explicable in terms of third-person, objective facts about brains, neurons, patterns of excitation, and other researchable aspects of cognitive states. Nothing about such physical, functional, or representational states of affairs implies that qualia should feel precisely as they do to a particular subject, or that representational states should feel any way at all, in which case the particular qualitative looks and feels of sensory experience certainly seem a realm apart from what science can predict and explain. This difficulty is what David Chalmers dubbed the ‘hard problem’ of consciousness.
Despite such claims, there is of course a well-established and intuitive connection between consciousness and functionally essential cognition. First, it’s clear that organisms such as ourselves, by virtue of states and processes realized in our brains, represent or model various aspects of the world and the body. Various sensory maps embodying informational content reliably co-vary with the world as we change our location with respect to the environment (external content), and others reliably co-vary with states of the body (internal content). Such representational capacities, limited and shaped by our particular sensory systems, and modulated by top-down gating and filtering, are clearly essential for successful behavior. Second, cognitive processes involving conscious sensory experience also seem essential to guiding behavior. Despite the fact that, for instance, blindsight experiments show some rudimentary cognitive capacities remain intact with respect to the blindfield in the absence of phenomenal consciousness, the general rule is that if normal consciousness is curtailed, behavior is compromised, often radically (Weiskrantz, 1997; Marcel, 1988). Third, phenomenal consciousness certainly seems to carry information – intentional, representational content – in that sensory qualities are generally (although perhaps not exclusively) experienced as being of or about the world and the body: I feel a pain located in my back, I see that the apple is red, etc. Putting all this together, the natural conclusion is that conscious intentional content plays a key informational role in mediating behavior, such that cognition involving conscious processes is functionally essential.
Nevertheless, it is often pointed out that the particular way a given quality feels for the subject, or the fact that it feels any way at all, seem conceivably independent of the representational, informational processes that occur in tandem with consciousness (Chalmers, 1995a). Because such first-person, qualitative facts about experience aren’t obviously entailed by representational facts, the link between consciousness and behavior remains at bottom a matter of correlation, and the phenomenal qua phenomenal might be a non-functional property that’s causally inessential to cognition and action (Flanagan, 1992, ch. 7). As Jaegwon Kim (1998) points out, if consciousness doesn’t reduce to physical-functional states, then its role in governing behavior simply duplicates the work already being done by the brain, in which case it is causally otiose.
The challenge for those who seek unification of the apparently disparate realms of qualitative consciousness and scientific objectivity, therefore, is to show that, despite appearances to the contrary, the phenomenal is entailed by the functional-representational (if in fact it is), and that all actual facts about experience are third-person facts. In what follows, I will pursue such unification by suggesting, taking a page (or several) from Daniel Dennett, that what seem to be non-functional, categorically private facts about experience are indeed explicable as seemings, not facts, seemings generated by the way in which consciousness comes to be. The key to all this, I will argue, is that as subjects we don’t have a first-person perspective on experience, even though as persons we most certainly have a first-person perspective on the world and a unique cognitive connection to our bodies. To understand consciousness, we must extirpate any lingering notion that we witness experience, or to put it somewhat melodramatically, we must kill the observer.
I should mention that in what follows I will not critique variants of so-called higher order, 2-level theories of consciousness such as those developed by Rosenthal (1993) and Lycan (1987), which seek to explain conscious states by invoking some sort of ‘inner sense’, monitoring, or scanning. Although I have my doubts about such accounts, well expressed by Dretske (1995, pp. 104-116) and Guzeldere (1995), and although such theories may imply varieties of observerhood, I won’t address them here since it would unduly widen the focus of this paper. I’m only concerned with undermining the perspectivalism which supposes or implies that consciousness is some sort of subjective presentation involving a metaphysically distinct category of private, first-person facts about experience.
The basic theoretical context for what follows is the thesis, following representationalists such as Tye (1995) and Dretske (1995), that phenomenal qualities (qualia) are non-conceptual representational or intentional contents, instantiated by neural states and processes, that inform us about the world, including our bodies. Qualia are functional, not epiphenomenal, in that the information carried by qualitative states is essential for guiding complex, flexible behavior successfully. As representations, qualia more or less co-vary with features of external objects and internal bodily states, although of course they also depend on the representational capacities of the organism as well as top-down processes which influence perception. Note that I’ll be mostly concerned here with sensory qualia, which are arguably just a subset of the phenomenal or experiential, since after all there are aspects of experience that are not directly sensory, e.g., the construction of qualia into objects, the sense of being a subject, the feeling of familiarity, etc. (Van Gulick, 1995, p. 64; Mangan, 2001).
It’s not difficult, of course, to motivate a bare functional representationalism, the idea that certain brain states and processes, whether or not they instantiate consciousness, play an essential role in mediating behavior by standing in dependence relations of some sort to external states of affairs and bodily states, thereby instantiating content for the system (Van Gulick, 1980). The basic issue, nevertheless, is whether the existence of such representations entails anything at all about phenomenology. As Ted Warfield (1999) points out, if representationalism is true, it’s plausibly the case that only a certain subset of representational states comprise consciousness. In which case, he rightly asks, what is it about this class of representational states that makes them conscious, phenomenal states? If we can begin to answer this question the so-called ‘explanatory gap’ (Levine, 1983, 1993) will be narrowed.
The traditional approach to the issue of phenomenal facts has been via Jackson’s ‘knowledge argument’ (Jackson, 1982), in which anti-reductionists hold that someone conversant with all the neurophysical facts that correlate with an experience of, say, red (the philosopher’s archetypical quale) nevertheless learns a new, non-physical fact when she first experiences red. There are good replies to Jackson’s original argument in the literature (e.g., Van Gulick, 1993; Levine, 1993; Biro, 1993; Tye, 1995) and Jackson himself has abandoned it in favor of representationalism (Jackson, 2001), so I’ll for the most part avoid these well-worn paths and instead undertake a deliberate consideration of qualia and the subject to whom they might appear.
The notion of a first-person perspective, when construed in a certain sense, arguably helps to perpetuate the intuition that experience includes categorically private facts, facts that are inaccessible and unsubsumable by any sort of shared, objective, third-person understanding. However, what I think are more or less standard senses of such a perspective need not necessarily support such an intuition. One standard sense is, uncontroversially, that particular experiences are had or undergone by individual persons, such that no one else has, or could have, my particular token experiences. Although others might have experiences of more or less the same type, I have a special relation to my experiences since no one else undergoes them, and this relation constitutes my experiential first-person perspective or point of view. Such perspectivalism leaves open the question of what experience fundamentally is, since it could be the case that having experiences is an entirely physical phenomenon, as materialists are wont to claim. It also leaves open the question of the precise relation of the person to experience, and in particular it need not imply that persons have a literal perspective or point of view on their experience, as the expression ‘first-person perspective’ might suggest with its ocular, perceptual connotations (Metzinger, 1995a, p.14).
Since different creatures or systems possess different sorts of representational capacities, a representational account of experience must acknowledge that the sorts of experience undergone by different types of creatures will differ accordingly. Being a bat, armadillo, or 100th generation AI all likely involve different types of consciousness, perhaps with qualitatively very different experiences (Nagel, 1979, p. 171). The varieties of type consciousness, as I will call it, raise questions about the extent to which different sorts of creatures can understand what it’s like to be each other and share knowledge about the world, since their first-person perspectives as described above will involve different types or kinds of experience (Biro, 1993, pp. 180-3). But again, in considering such questions, the basic nature of the qualitative, of the phenomenal per se, is left open.
Another unproblematic sense of having a first-person perspective is that of being a motivated, cognitive agent with a particular point of view on and knowledge about the world, and a particular history within it. Having such a point of view and such a history – I’ll refer to it as the agent perspective – is a matter of having a unique physical trajectory through time and space, so there’s nothing particularly mysterious about this, although it’s a crucial part of being a specific person. No one else is me in the sense of having this particular perspective on the world or of having lived through this unique history, although others may have followed similar paths through life. My having had particular token experiences thus depends on this history, although it’s also dependent, of course, on the type of consciousness I instantiate.
A third sort of first-person perspective or point of view, to be discussed in section 9 below, is that associated with the phenomenal, experienced sense of being a self or subject. When conscious, we normally have the continuous experience of being a subject at the center of our experienced world, a more or less unitary self that interacts with and perceives the world from the subjective perspective of being located within a particular body that’s uniquely its own. The neural correlates of the phenomenal subject are the focus of an ongoing research program (Damasio, 1999, 2000; Metzinger, 2000a, 2003; Parvisi & Damasio, 2001; Vogeley & Fink, 2003). I’ll argue in section 9 that this experienced sense of being a self presented with the world can play a role in generating the problematic construal of the first-person perspective.
Finally, the problematic construal of the first-person perspective that I want to distinguish from all those above is that in undergoing experiences, the person might be in a literal perceptual relation to experience itself. The person somehow witnesses or observes experience such that it becomes a private presentation, involving a set of categorically private phenomenal facts, the qualitative facts about what experience is like. On this picture, phenomenal feels or qualia have a non-functional, non-representational, and by implication, non-physical aspect involving private facts about experience that only the individual can access.
Such an observational perspective is suggested by locutions that crop up in the literature which place the experiencer in a privileged position of seeing, accessing, or directly apprehending facts about experience. For instance, Thomas Nagel remarks that ‘For if the facts of experience – facts about what it is like for the experiencing organism – are accessible only from one point of view, then it’s a mystery how the true character of experiences could be revealed in the physical operation of that organism’ (Nagel, 1979, p.172 original emphasis). Similarly, Nagel says that ‘It is difficult to understand what could be meant by the objective character of an experience, apart from the particular point of view from which its subject apprehends it’ (Nagel, 1979, p.173, original emphasis). Now, to apprehend an experience, or have special access to facts about it, might suggest that the subject is in a unique observational or perceptual relation to it, which in turn implies that certain facts about that experience might be privy to the subject alone. As John Biro (1993, p. 180) puts it, ‘A point of view, it is claimed, gives its owner access to a special kind of fact that is different from, and irreducible to any other fact or set of facts equally available to others’.
A more recent and literal expression of the observational perspective is found in Max Velmans’ idea that the first- and third- person perspectives offer two complementary views of experience, views that depend on what he calls ‘observational arrangements’ (Velmans, 2002, p. 11). The subject is in a position to observe experience in a way that no outside observer can, so that, for instance ‘Other people’s experience might be hypothetical constructs, as we cannot observe their experiences in the direct way that we can observe our own…’ (p. 22, original emphasis). The difference in observational arrangements produces two different sorts of facts or information about experience, the first-person information about phenomenology and the third-person information about the brain’s representational mechanisms (p. 15). These views or perspectives and the facts they support are, Velmans claims, mutually irreducible – no reduction of the phenomenal to the physical or functional is possible. Similarly, Steven Lehar (2003) proposes what seems a modified sense data theory of consciousness (see Dretske 1995, pp. 128-9), in which experience, from the first-person point of view, is the perception of what he calls ‘internal effigies’ or ‘internal percepts’. He says, for instance, that ‘We cannot … in principle have direct experience of objects in the world itself, but only of the internal effigies of those objects generated by mental processes’, and that ‘consciousness is indeed observable…because objects of experience are first and foremost the product or “output” of consciousness, and only in a secondary fashion are they also representative of objects in the external world’. Lastly, Antonio Damasio speaks, undoubtedly metaphorically, of experience as a ‘movie in the brain’ which is composed of mental images generated by neural processes (how neural processes do this, he admits, isn’t clear). But he slips from metaphor to literalism in supposing that we indeed observe such images, not the world, when we have experience: ‘The image we see is based on changes that occurred in our organisms…’ (Damasio, 2003, pp. 198-9, my emphasis). William Lycan calls the supposition that experience is a perceived object ‘the banana peel’ since, as he puts it, ‘anti-materialists typically…slip on it into the Movie Theater Model of the mind’ (1987, p. 17), as do, it seems, some unwary materialists.
This notion of having an observational perspective on experience, which takes experience, not the world, as that which we directly perceive or know, has helped fuel the belief in an ontological divide between the qualitative and the objective, the mental and the physical. It does this by setting up a realm of private phenomenal facts that one observes about one’s experience, facts that can never become shared, objective knowledge, for instance as specified by science, and that can’t be explained in terms of physically instantiated processes. But, if we consider what it means to observe and to be in possession of facts about things, and if we consider our phenomenological situation, including both the qualia of experience and the person who has them, I think it will become clear that we know the world, not experience, directly (as directly as we can know anything) and we know it via experience itself. I’ll argue that the observational construal of the first-person perspective is untenable, and it’s this construal which supports the existence of categorically private phenomenal facts about qualia. In proceeding, I will consider qualia first and what about them might constitute categorically private first-person facts, then conscious subjects and their possible relations to experience.
What precisely are qualia – the looks and feels of sensory experience – and why might they be supposed to incorporate private, non-representational, and non-functional facts? This is to take seriously the question, for instance, ‘what is it like to see red?’. In specifying what it’s like, we must pinpoint what about qualia is supposed to resist explanation in terms of representational, informational functions, for it turns out that there’s a good deal which is either explanatorily unproblematic, or at least unproblematically conceivable on a representationalist account. For instance, qualitative intensity, as in the subjective brightness of an experienced expanse of color, plausibly corresponds to the intensity of light striking the retina as coded by a particular level of neural activation. Likewise, the duration of a particular quality in my experience can often be linked to objective facts, such as the persistence of a particular object in my field of vision. For example, my particular experience of blue lasts as long as I continue to gaze at a cloudless sky at noon.
Most, if not all, qualia are occurently determinate, stable, non-conceptual values within the various modes of conscious sensation. By this I mean simply that each quale has a particular place in the phenomenal structure of my color experience as defined by its relations to other hues. The phenomenal character of the blue of my mouse pad is one of a huge number of particular hue values found in color experience whose specific place lies somewhere on the continuum of distinguishable blues. Moreover, particular qualities are stable values within a modality, in that, for instance, the particular look of my cream-colored door doesn’t change as I experience it from moment to moment, or from week to week (barring adornment with greeting cards or the slow accumulation of dust). And as Michael Tye, among others, has pointed out (1995, p.139), since the variety of qualia far surpasses my capacity to conceptually categorize them as unique qualities (e.g., I don’t possess distinct concepts for all shades of blue or off-white I might encounter), particular instances of qualia are perforce non-conceptual.
Let us provisionally adopt, on the basis of the representationalist paradigm, what I will call for convenience the informational hypothesis – that sensory qualia are representational contents embodied by world-responsive, neurally-instantiated, multi-dimensional representational phase or state spaces that feed information to higher-level perceptual representations (Churchland, 1988, pp. 455-7; Churchland, 1989, Ch 9; Tye, 1995, pp. 101-3, 119, 138). On this assumption, a particular phenomenal hue corresponds to the cluster of specific values of each component dimension of hue state space, and its relations to its cousins are fixed by the relative proximity or remoteness to other hues along these dimensions. Its stability – the reliable re-occurrence or moment-to-moment sameness of a phenomenal hue or other quale – is plausibly a function of the physical and functional stability of our sensory coding apparatus and the world it represents. As a non-conceptual sensory representation, it constitutes information fed to the sorts of categorizing, higher-level belief systems that constitute concept-based cognition (Tye, 1995, p. 104).
A singular, particular quale, or particular quality of experience, is just that which you can discriminate in a particular situation as admitting of no further detail; it’s non-decomposable and homogeneous, having no further phenomenal structure. For instance, from a distance a certain portion of my door looks uniformly cream-colored – I can discriminate no further color variation in that expanse of the door. Or imagine you’re listening to a good clarinetist playing a concert A 440 at a steady volume. That sound, at least for many of us, is experientially non-decomposable, hence counts as a single quale with a given duration and intensity. Whatever mode of sensation we choose to explore, we can find such singular, monadic, and homogeneous elements of experience. On the informational hypothesis, a particular quale, for instance the uniform cream color of that particular portion of my door, is non-decomposable since our color sensing systems have a finite resolution, such that (very crudely) when wavelengths within a particular frequency band impinge on a certain portion of the retina, the same set of color state-space values is assigned to all points within a corresponding portion of the internal visual map that contributes to an object representation and that in turn partially constitutes my experience. This account of qualitative homogeneity helps explain the seeming intrinsicality of qualia, often thought to be a barrier to their reduction in terms of relational processes and states (Clark, 1995; Feser, 2001).
Significant regarding sensory qualities is that, although we can think of them as basic, non-conceptual and non-decomposable bits of phenomenology, they are usually experienced as belonging to or associated with phenomenally complex objects and events. Qualia are ordinarily experienced not as features of experience or representations, but as characteristics of things that we perceive or sense such as the color of the door, the pain in my back, the taste of the coffee, the sound of a car going by (Van Gulick, 1993, p. 149; Metzinger 1995a, p. 11-12). They are essential in forming the often unexpressed, but expressible conceptual judgments that something is the case, e.g., that the door is cream-colored. (Of course in cases such as dreams, hallucinations, and afterimages it’s clear that phenomenal qualities experienced at a given time need not correspond to any actual state of affairs.) The informational hypothesis helps to explain this fact about qualia in terms of higher level binding processes: lower level, sensory state-space values contribute specific information – intentional content – to higher-level, occurrently bound, and behaviorally useful perceptual object representations. At the highest level, these representations are amalgamated into the experience of objects within a single, more or less coherent and predictable world with the subject at its center (Metzinger, 2000b). So qualia are all assigned, more or less, to the larger perceptual ensembles that figure in moment-to-moment experience.
Equally significant is that many sensory qualia matter to us, in that they are either directly affect-laden and emotionally valenced (e.g., the qualia of an injury or orgasm), or they can discriminate states of affairs that we tend to approach or avoid (red apples versus shiny green poison ivy). Such a motivational aspect has its informational basis in how states of affairs of the world and body are coded according to their consequences for drive satisfaction, homeostasis, and ultimately, survival. The point of tracking the world via sensory perception is to know (and approach) what’s good for me and know (and avoid) what’s bad for me. The fact that I’m rather attached to certain qualia, and strongly repelled by others, makes good sense on the informational hypothesis in the context of my being a well-honed survival machine shaped by evolution.
Thus far, we have the following picture of basic sensory qualia, which, to repeat, I’m not claiming is exhaustive of all phenomenal experience, since such qualia are just a subset of what constitutes consciousness. Qualia are homogeneous and non-decomposable; they exist as discrete, stable values in a non-conceptual sensory modality; they are ordinarily experienced as characteristics of objects or events in the world, including the body; they can vary in intensity and duration; and they often matter to the experiencer. And thus far there is nothing in this picture that doesn’t seem potentially explicable on the informational hypothesis, although of course nothing in the sketch above counts as a serious attempt at specifying the mechanisms that could instantiate these informational characteristics of phenomenal qualities (for a bit more on such explanations see section 10). Still, all these facts about qualia seem at least conceivably understandable as third-person facts involving representational processes.
But this picture of qualia has seemingly left out what anti-reductionists contend is the characteristic which most resists functionalization: the ‘way it is for me’ phenomenal feel of a quality. True, informational functions might specify world-tracking intentional content that contributes to representations which are necessary to guide behavior, but an explanation of why such content assumes a categorically phenomenal aspect seems nowhere to be found on the informational hypothesis. Saying what this categorically phenomenal aspect is, precisely, is notoriously difficult of course, but the mark of basic phenomenal particulars is more or less that they involve something simply and essentially qualitative, something that itself defies further qualification. It’s something quite specific, but having no internal structure, it can’t be further described and therefore seems arbitrary with respect to any informational function qualia might serve (Levine, 1983, p. 359; Van Gulick, 1993, p. 143-4). It’s not just that blue differs from red in consistent ways, thus permitting reliable discriminations and categorizations of objects, but that it has its own unique, seemingly intrinsic qualitative look that, apparently, could have been different while serving the same discriminative role. In fact, every discriminable quality in my experience (and there are thousands of such qualities) appears or feels to me a certain ineffable, non-relational way. Nothing about the informational hypothesis seems capable of specifying this rather large set of facts, facts about my particular experiences, not about the world that experience might represent. Nor does it seem capable of saying why represented content must appear any way at all to a perceptually and cognitively adept system.
The difficulty for representationalism, then, is what I will call the essential characteristic of a quality (as have others, e.g., Nagel, 1979, fn p. 175; Metzinger, 1995, p. 15): being a particular, ineffable, seemingly intrinsic and functionally arbitrary way for a particular subject. But before continuing, I want to reiterate that, as has been suggested above, there are many salient aspects of phenomenal experience (but perhaps not essentially phenomenal aspects) that are conceivably informational. Among these is the fact that qualia are particular values within sensory modalities. This particularity is a straightforwardly third-person fact, since as conscious creatures we all agree that my blue mouse pad (could you observe it) looks a definite distinguishable way. However, it’s not particularity in general that’s the explanatory target but the particular subjective look of blue to each of us as separate conscious individuals. This, it seems, is a fact not about my mouse pad, but about each subject’s particular experience. The question before us is whether or not the essential characteristic of a quality – that it is like this for me – involves categorically private first-person facts about experience over and above third-person facts about representation. If it doesn’t, then we can draw the conclusion that third-person facts fully explain the subjective qualitative character of consciousness.
On anti-reductionist accounts of consciousness, the essential characteristics of phenomenal qualities in an experience constitute categorically private facts accessible only to the person who undergoes the experience. Now, the complaint often lodged against reductionist accounts of consciousness is that no third-person description of a system that subserves consciousness, however elaborate, can capture these facts, for instance the way my blue mouse pad looks to me (Jackson, 1982; Nagel, 1986, p. 15; Flanagan, 1992, p. 117). As Chalmers (1999) puts it, ‘I also take it that first-person data can’t be expressed wholly in terms of third-person data about brain processes and the like… That’s to say, no purely third-person description of brain processes and behavior will express precisely the data we want to explain…’ (quoted in Dennett, 2001a). But, given the ineffability of qualia, it’s important to see that no first-person description does any better. Being a subject that has its own proprietary token experiences conveys no advantage to the subject in capturing or expressing what their qualitative elements are like, even though the subject would seem to be in the best possible position to know about such things. First-person descriptions are mute when it comes to qualia because ostensibly private facts about experience, by virtue of being ineffable, are unspecifiable. Like you, I can reliably recognize and report qualities as they occur in experience and place them in relations to their qualitative cousins, but I can’t descriptively specify what it is about a particular blue that makes it appear precisely the way it does to me. If I could, I’d be in a position to give a description of that blue, but of course I can’t. (Likewise for pain: I can’t specify what it is about pain that makes it painful). The fact that I can’t describe it, and compare that description with your description, helps drive the intuition that the way it appears is a categorically private, not public fact.
To see the difference between ostensibly private qualitative facts and public facts, imagine we are standing before a blue chair. That a blue chair is in front of us counts as a third-person fact, since we can agree on a description: it’s a chair, it’s blue, and it’s here. We’ve achieved inter-subjective consensus and now share collective knowledge about the world, so that chair is the same chair for both of us. But on the first-person-fact understanding of qualia, the what-it’s-like-to-you of your blue might be different than the what-it’s-like-to-me of my blue. Since neither of us can offer a description of our blues which could confirm or disconfirm a difference, they remain seemingly private, categorically subjective facts incapable of third-person verification.
However, the unspecifiability of the essential characteristics of qualia, though it may drive the intuition of privacy, is at least somewhat troublesome for their status as private facts. That the chair is blue is an indisputably informative fact about the chair. That my private blue is like this, and possibly unlike your blue, is a claim that sounds as if it had informational content, but as we’ve seen, apart from the story about the relations of a particular quale to its cousins (relations which are the same for all of us with similar representational capacities), there is no further informational story to be told about the essential characteristic of a particular quale, even to ourselves as the presumptive cognizers of such facts. If a purported fact (‘my blue is like this’) delivers no informational content to its possessor, then one wonders if it’s a fact at all. Is a fact that delivers no content and no knowledge indeed a fact? Remember, casting aspersions on what-it’s-like-for-me blue’s ambition to be a private fact is not to deny that blue looks a particular way to me, or that I can reliably recognize and name it on the basis of such a look, or that it plays an essential informational role in my cognitive economy. It’s only, and strictly, to suggest that the way it looks to me doesn’t involve a private fact, and (generalizing to all qualia) that there is no categorically private first-person information, available to the subject alone, to be gleaned about consciousness. Put another way, it’s to say that the demonstrative ‘this’ refers strictly to informational content about the world available to other creatures with similar representational capacities (my type of consciousness), content that gets misconstrued as a world of private facts accessible only to the subject.
To flesh out the story of why qualia might not involve private facts, it will be helpful to discuss briefly our situation as cognitive creatures. As organisms (or more generally, intelligent systems), we observe, in a straightforward and unequivocal sense, the world around us. (I’ll stick with the visual modality in what follows, although we could tell the same story about observation in terms of hearing, touch, echolocation, or any world-responsive sensory modality that participates in perception of the world outside the body.) Light reflects off objects, enters the eye, and produces a cascade of neural events that ends up contributing to a freshly updated, neurally-instantiated representation of those objects which helps us get around successfully in our environment. This is what constitutes directly seeing the world, being in a direct perceptual or observational relation to it, and, ultimately, possessing facts and knowledge about it: the creation of informational, intentional content by incoming stimulation which helps determine values within the various representational state spaces that make up the visual system, values which (crucially) get integrated with other content into higher-order, suitably bound object representations that subserve behavior. To be in possession of facts is for a cognitive system to have access to (and consist of) such behaviorally useful representations, useful because they reflect both the way the world is in some respect, or regularities it manifests, and the needs of the organism.
Observing the world and coming to know facts about it are a matter of getting the intentional content of one’s representational system to be constrained (mostly) by the world, although of course it’s still a highly selective, type consciousness-modulated and agent-specific take on the world. Thus constrained, the content ends up being about the world, and less about how you’d like it to be, always a good strategy for survival. Such content ranges from first-order, conceptually indeterminate content of sensory experience (e.g., my door is that color) all the way up to higher-order, abstract conceptual content (e.g., knowledge is justified true belief). But whether we’re considering non-conceptual or conceptual knowledge, the story about the organism observing the world and knowing facts about it ends here, in that there is no further inner observer of representational content, although there might seem to be such an observer (of which more below in section 9). To take Dennett’s line, representations don’t have to be literally witnessed or appreciated by anyone or anything outside the network of representations to be efficacious (Dennett, 1991, 2001a).
This means that when it comes to the content delivery systems of the various modes of sensation, these too are unwitnessed and unappreciated. We don’t have, for instance, a built-in representational apparatus that represents the process of color-coding which then delivers content about that process to help guide behavior with respect to that process. Although meta-representation of representational processes often occurs in various higher-level cognitive contexts (as in applying phenomenal concepts when reporting experience, e.g., ‘I’m in pain’, or as exampled by this paper) we simply don’t have a direct behavioral need for information about how we go about the basic sensory work of representing the world. We just need sense modality-encoded information about the world, not information about how we encode such information.
This entails that we are representationally blind to the various multi-dimensional state spaces that constitute our vehicles of sensory representation, and a fortiori we are representationally blind to the dimensional aspect of the content they code for. We are not in a position to see or observe – that is, directly, non-conceptually represent – the fact that a particular bit of sense-derived content is instantiated by such and such a set of dimensional values in a state space. Now, proceeding on the informational hypothesis, if that content were, for instance, the blue of my mouse pad as it phenomenally appears to me, this would explain the fact that I can’t see or say why – in virtue of what further set of facts – my blue looks a particular way to me, even though, on the informational hypothesis, it is a particular way by virtue of it’s being a certain set of dimensional values. I’m simply not in a position, vis a vis the experienced content blue, to know anything about the way it looks in and of itself because I don’t observe the process of sensory representation, rather I consist of it (along with many other representational and non-representational states and processes) as an organism that observes the world. Since I don’t have a perspective on the informational goings-on of experience, my blue will necessarily seem arbitrary with respect to its informational function and it will necessarily seem intrinsic, that is, non-relational and sui generis. It will seem as if some other apparently intrinsic phenomenal hue could have served as the particular color of my mouse pad, in which case it will seem eminently conceivable that someone with the same representational set-up could be experiencing at least a slightly different phenomenal hue from mine under the same perceptual conditions.
I may recognize and report blue as a feature of my experiences when applying the phenomenal concept ‘blue’ (Tye, 1995, p. 167-70, what Loar (1999) calls a recognitional concept) but such recognition shouldn’t be confused with cognition or observation of blue as an object of my experience. I’m simply not in a position to experience my experience, that is, to take experience itself, in its basic qualitative particulars, as an observed, represented object about which I possess facts (Dretske, 1995, pp. 100-3). Apart from being able to specify its relations to its hue cousins, I don’t perceptually cognize the particular look of blue as a specifiable private fact since I don’t have the representational capacity to do so. Its ineffability, unspecifiability, and seeming intrinsicality as being that color are entailments of my representational limitations.
Since I can’t assume a perceptual observer relation to the basic informational contents of sensory perception or the representational vehicles that encode them, sensory qualia don’t include privately given or directly observed facts about my experience, rather they are represented facts about what is represented, what I as an organism directly observe, namely the world. Instead of cognizing (representing) blue, I cognize (represent) the world about which that particular blue is properly a specifiable fact: it’s the specific color of my mouse pad as represented by the sensory system that I am, partially, as a cognitive creature. Although they are instantiated by me as a representationally adept organism, and so are informational properties of representational states, qualia aren’t about a special subjective world that only I have access to, rather they are informational content that represents the world as being certain ways, content that (as I will argue below in section 10) participates in those higher-level, integrative, complex behavior-guiding functions that it turns out are specific only to consciousness. The essential characteristic of a quality, that it feels or looks like this to me, is thus a fact about the world-as-represented-by-me, not something I observe about my experience. There is, therefore, nothing factually available to me about experience over and above its intentional content, i.e., the relational, extrinsic, informational characteristics of qualia outlined in section 4 above (plus other non-sensory content) that I can successfully report, as can others. Consequently, the essential characteristic of an experience doesn’t involve a further, private fact, something distinct and separate from such content. It seems to involve such a fact only because we’re not in a position to directly, non-conceptually represent the fact that the values in various sensory state space dimensions constitute the look and feel of experience.
There isn’t, it turns out, anything metaphysically essential about the essential characteristic of a quality; instead, the ‘mark of the qualitative’ for basic sensory qualities is just that they can’t be further described or specified because they are first-order, non-conceptual discriminative representations whose dimensional structure is not itself represented by the system, and therefore cannot be a fact for the system. The concrete, unanalyzable qualitativeness of sensory particulars is simply the fact that, as Thomas Metzinger puts it, the state space dimensions of qualia are ‘impenetrable to cognitive operations’. All that’s directly available to us is the content delivered by sensory representations, not the fact that such content is being represented or how it’s represented. Another way to express this is that sensory representations are ‘transparent’, that is, cognitively invisible to us (Metzinger, 2003, p. 387), even though they are, of course, neurally instantiated in our heads. This makes it easier to see how the ‘trick’ of experience is accomplished by processes that at bottom are quantitative and computational, that is, not irreducibly qualitative. To see exactly how this trick is accomplished, which will involve a complex story about representational functions and their neural instantiation (see section 10), is to close the explanatory gap. And as Dennett (2001a) rightly insists, nothing that is itself irreducibly qualitative or subjective can figure in the story if it’s to count as an explanation of consciousness.
The absence of categorically private first-person qualitative facts entails that qualities given in sensory experience all represent third-person, objective facts about the world as represented via a particular type of consciousness. To the extent to which others who share the same type of consciousness assume more or less the same perceptual, observer relation to the world, or the extent to which they can imagine being in such a relation, such qualities are shared, non-conceptual knowledge about the world. For creatures of a particular type of consciousness, phenomenal qualities pick out (represent) the same properties and regularities in the world in service to more or less the same set of cognitive needs.
My first-person point of view of the world as a cognitive agent – my agent perspective – consists of the unique set of third-person, objective facts (conceptual and non-conceptual) that I represent about the world as a particular type of consciousness, a type shared by all humans. But my view of the world is also conditioned by the rather special, but nevertheless third-person fact that only I am constituted by this body. On the view I’m recommending, the most private, subjective experience, e.g., the current feel of my toothache, represents a third-person, objective fact that only I happen to know by virtue of the fact that only my brain is recursively hooked up to itself and my body (Flanagan, 1992, p. 94). My toothache isn’t private by virtue of incorporating categorically private phenomenal facts about my experience, although experience indeed consists of phenomenally represented facts about the world, in this case my body. Were a yet-to-be-devised (but perhaps not so far off) brain/body scanner trained to recognize the neuro-physical correlates of my toothaches, we’d see that the privacy of such facts is contingent, not categorical or metaphysical.
This is to say that only I, as this particular representational set-up, am now representing the objective state of my tooth (assuming that the tooth is in fact the problem) in this particular non-conceptual, affectively-laden, behavior-guiding manner, which is simply to say it’s my experience, not anyone else’s. This constitutes, uncontroversially, the first-person perspective of my undergoing my particular token experiences. It is this individually tokened, non-conceptually based mode of representing particular objective facts about my body that others can’t share by virtue of not being me – only I instantiate this mode vis a vis third-person facts about my tooth, a mode which, on the informational hypothesis, is what it’s like to have a toothache. As Tye says, to fully understand or know what it is to experience a toothache, one has to have exercised similar non-conceptual representational capacities such that one can imagine what it’s like to be in that representational state (Tye, 1995, pp. 54-6, 169-70), which is to say that one must in this respect instantiate more or less the same type of consciousness. While only I instantiate this particular non-conceptual representation, you, as someone who likely has undergone similar experiences, know what it’s like to instantiate representations of analogous third-person facts about yourself. But neither of us is privy to subject-specific, first-person, unshareable facts about our experiences of a toothache. As Dretske puts it, ‘What we are conscious of when we feel pain (hunger, thirst, etc.) are not the internal representations of bodily states (the pains), but the bodily states that these representations (pain) represent. Though we can be – and most often are – aware that we are in pain, pains, like visual experiences, are awarenesses of objects, not objects of which we are aware’ (1995, p. 103). In experiencing red for the first time, Frank Jackson’s Mary comes to know what it’s like to instantiate a non-conceptual representation of a certain complex property of normal human environments; it’s a new (for her) representational take on a set of third-person facts that she, a well-trained neuroscientist, already knew via high-level conceptual representations. The privacy of sensory experience that follows from the unproblematic sense of a first-person perspective is not, therefore, a matter of access to special, categorically private data about experiences as objects, but just to be the only person who now instantiates this instance of representing a particular fact about the world in this particular non-conceptual mode.
Despite such considerations, it might still seem intuitively the case that we are in a privileged observer-like relation to our experience, and indeed I think that our cognitive situation can provoke and maintain the illusion that experience is literally a presentation, not a representation. To help explain this illusion we must consider not only the purportedly private facts about experience but the observing subject that supposedly has access to them, since they stand or fall together.
As an organism (directly) observing the world by means of representing it, crucial objective facts for me to represent about the world are that I am in a particular position in it and that I’m interacting with it. As Antonio Damasio (1999, 2000), Thomas Metzinger (2000b), and others have pointed out, to survive I have to have a continuously updated egocentric reality model of my overall situation, a situation that includes me-the-organism contained in a larger environment that, from the organism’s survival-driven perspective, is not-me. The most basic, essential distinction to be drawn within that model is that between self and non-self. Thus, a more or less veridical and functional self-model (Van Gulick, 1993, p. 150; Metzinger, 2000a, p. 289) must participate in a larger global representation that represents the self as being in, but other than, a world that is not-self. At all times the organism must have a sufficiently robust representation of its bodily boundaries, its location and trajectory, and its plans and purposes for its self-interested agenda to bear fruit. This constitutes a functionally essential, self-in-the-world-modeling representational architecture.
As a further elaboration of the reality model, the system must also represent itself as interacting with – being affected by and affecting – the world; it must model what Metzinger calls the ‘intentionality relation.’ As he puts it, it’s functionally essential that the system have an ‘ongoing, dynamical representation of the system as currently interacting with an object-component’ (2000b, p. 296, original emphasis). On the informational hypothesis, this represented object-component consists of the intentional content about objects in the not-self world as delivered by perceptual processes fed by sensory representations (and modulated by top-down processes), while the self-model is built from internal representations of homeostatic functions that preserve bodily integrity (Damasio, 1999, 2000). Although I won’t defend it here, Metzinger’s explanation of how the phenomenal self arises from a representational architecture involves somewhat the same (but far more elaborate) considerations I adduced with respect to qualia in section 7 above: by virtue of the fact that it can’t represent the self-model as a model (which is to say the self-modeling process becomes transparent or invisible to the system), the system inevitably falls into naïve realism with respect to the model, which therefore becomes an experienced reality of centered subjectivity (2000b, pp. 298-301). Such subjectivity involves the untranscendable feeling that we are seeing and perceiving the world around us as it’s presented to us by our senses. This feeling of self-as-observer to whom the world is presented thus models, in experience, the objective situation of being an organism in a perceptual relationship to its environment. The way in which our experience of the world changes as we experience ourselves acting with respect to it is the more or less reliably co-varying and behavior-controlling phenomenal analog of the organism’s interaction with the actual world. In describing this situation, Metzinger and others, such as Revonsuo (2000, pp. 65-6), have suggested that experience constitutes, in effect, a virtual world with the virtual subject at its center.
Being a subject to whom the world is presented – the subjectivity of experience – is thus on Metzinger’s account a construction within experience. This phenomenal first-person perspective is analyzed and explained in representational terms that are themselves neither perspectival nor phenomenal (again, although I find Metzinger’s account persuasive, I won’t attempt to defend it here). But despite this type of theoretical reduction, the strong, perhaps untranscendable feeling of being an observer or witness to the world can, I think, mislead some into supposing that experience itself is witnessed. That is, some, such as Velmans, Lehar, and Damasio (section 3 above), will make the mistake of modeling their analysis of experience on the phenomenal sense of observerhood conferred by our representational architecture.
To see this, first note that the felt sense of being a self can vary from person to person, and over time for the same individual. Some persons (me, for instance) routinely experience the phenomenal subject as a more or less unitary, point-like self, nearly or totally devoid of any intrinsic character of its own. This self looks out at the not-self world, with its felt location more or less between and behind the eyes (most likely a function of the fact we are such visually oriented creatures). On this version of the phenomenal subject – call it the bare phenomenal self – the body is most certainly mine, but perhaps not essentially me. By contrast, others (who don’t live ‘in their heads’ as much as I do) may have a more distributed sense of the essential self, constituted by some combination of bodily sensations, occurrent feelings, and thoughts, such that the phenomenal subject isn’t bare, but consists of a set of characteristics. And of course it’s possible and likely that the felt sense of self can vary within individuals over time, depending on their circumstances, for instance when meditating or under the influence of neurological disorders affecting the self-model (Metzinger, 2000b, pp. 295-6).
What I want to suggest is that the experience of being a self presented with the world, particularly when the self is experienced as the bare phenomenal subject, can create the impression that not just the world is presented to us, but that experience is presented to us. The bare phenomenal self is experienced as being essentially other than what is presented to it, and for such a subject even bodily sensations, occurrent thoughts, and emotions (‘internal content’), along with the external world, are experienced as presentations to the essential me. This phenomenal situation involves, then, what non-conceptually feels to be a quintessentially witnessing subject, something that stands apart from and observes all of what is presented in experience. I’m suggesting that it’s this extreme separation of the self from other contents of experience (for the phenomenal self is itself such content) that helps generate the problematic intuition of having an observational perspective on experience. For example, Baars (1996) says, ‘You are the perceiver, the actor and narrator of your experience’ (original emphasis), and Deikman (1996) identifies the ‘I’ – what he describes as bare, contentless awareness – as the observer of the contents of consciousness. Now, when this notion of observerhood is combined with the conceptual (and veridical) understanding that the world is represented to us partially via experience (and partially via unconscious processes), the temptation might be to conclude that the representations themselves, the ‘images’ of experience as Damasio and Lehar call them, are observed from some sort of vantage point or perspective. Thus, instead of properly construing the phenomenal subject as constructed within experience, the subject might be construed as the witness to experience, so that experience is thought of, finally, as a literal presentation to an observing self, not what it truly is, a virtual world-creating representation that includes the self.
In the grip of such an observational picture, the conviction might arise that the intentional contents of sensory experience are witnessed as private, first-person facts about experience, e.g., as the look of my experience of blue, the feel of my experience of pain, when in fact, as we saw in section 7, we are not in a position to directly represent basic representational processes and thus know facts about particular, considered-in-themselves sensory representations that on the informational hypothesis constitute qualia. While the phenomenal situation of our virtual worlds involves the functionally useful sense that the experienced self is being presented with the world and the body via sensory experience, no one and no thing is literally presented with experience; as Dennett says, experience is ultimately and literally unwitnessed and unappreciated.
This, then, is my psychological-cognitive diagnosis of the philosophical mistake underlying the modern descendants of sense data theories such as Lehar’s, which duplicate, in the purportedly direct perceptual relation of the person to experience, the actual direct perceptual relation between the person and environment. Naïve realists, since they don’t suppose representations are involved in witnessing the world (the world is simply presented) can’t fall prey to this mistake. It’s only those sophisticated enough to realize that the world is known via experience who might be seduced by the strong (and functionally useful) sense of phenomenal observerhood into supposing that experience itself is a presentation involving a first-person observational perspective on private facts about experience. But, contra Velmans, no observational or representational arrangements exist which could provide this sort of perspective or give access to such facts.
Although my approach so far has been largely negative, seeking to undermine intuitions that qualia involve categorically private phenomenal facts, a positive account of sensory consciousness as informational states is emerging from neuroscience and neurophilosophy (see for instance Dehaene, 2002; Metzinger, 2000a, 2003). Defined methodologically, consciously available information is just that embodied in representations that participate in functions subserving the empirically discovered capacities conferred by conscious states as opposed to unconscious states (Baars, 1999). For instance, conscious states have the capacity to make information available over extended time periods in the absence of continued stimulation; they permit novel, non-automatized behavior; and they allow spontaneous generation of intentional, goal-directed behavior with respect to perceived objects (Dehaene & Naccache, 2001). Studies of neural activity which contrast conscious and unconscious capacities indicate that phenomenal experience is associated with widely distributed but highly integrated neural processes involving communication between multiple functional sub-systems in the brain, each of which plays a more or less specialized role in representing features of the world and body (Kanwisher, 2001; Dehaene & Naccache, 2001; Jack & Shallice, 2001; Parvizi & Damasio, 2001, Crick & Koch, 2003). Such processes, it is hypothesized, constitute a distributed, ever-changing, but functionally integrated ‘global workspace’ (Baars, 1988; Dehaene & Naccache, 2001).
The recruitment of sub-systems into the global workspace suggests that the neural correlates of phenomenal states integrate represented features into exactly those sorts of bound, coherent, object-level representations which dominate in subjective awareness and that seem necessary for most high-level, flexible behavior and cognition. The various aspects of qualia covered in section 4 – their relative salience, their particular sensory value, their emotional valence, etc. – may well correspond to various lower-level representational processes that intercommunicate during conscious episodes and thus are representationally linked to constitute conscious contents. But it’s critical to see that we aren’t normally aware of any of these aspects independently or alone, but always in a represented world of self, objects, processes (relations and interactions of objects over time), all in more or less stable arrangements (Metzinger, 1995b, p. 448; Metzinger & Walde, 2000). It takes a myriad of qualia to constitute a phenomenal, virtual world, and if we could ‘experience’ just one quale in isolation, it wouldn’t be experience at all; it would simply be the unconscious existence (part of no represented object and present to no represented subject) of one representational component of what together constitutes the phenomenal. Consciousness constitutes (and is constituted by) a represented holistic context within which discriminable elements are embedded (Kanwisher, 2001, pp. 107-8). The neural sub-processes responsible for each component of conscious representations, when acting in an unbound fashion, are perforce unconscious, but when they participate in the global workspace that unifies them into sensory percepts within such a context, they contribute their informational content to consciousness: the amalgamated, contextualized, object and self-creating content that controls complex, flexible behavior. Although sections 4 through 8 concentrated on explaining qualia – the basic sensory elements of experience – the twist in the plot (flagrantly anticipated in section 9) is that phenomenology nearly always instantiates a virtual world, in which qualia participate as bound elements in globally representing the self in its environment of objects and events. To be phenomenal is (at least) for representations to be cognitively impenetrable and thus phenomenally transparent on the level of sensory elements (see section 7), and for these components to participate in an integrated, behavior-controlling, and (on Metzinger’s account) transparent self-in-the-world model.
What Dennett (2001a) suggests we should conclude from the study of the neural correlates of consciousness, and I concur, is that experience just is that behavior-controlling information represented in the brain which is sufficiently globally available, i.e., not segregated in a lower level schema or in one sensory modality (see also Jack & Shallice, 2001, pp. 185-7). Shared and utilized by multiple functionally integrated sub-systems, this information largely ends up driving behavior (largely but not exclusively, since unconscious representations may have their effects). On Dennett’s ‘fame in the brain’ gloss on the global workspace model, conscious representations are just those that dominate in the workspace by winning out over competing sets of less processed sensory information and incompletely bound higher-level cognitive content, and they win the competition via a staggeringly complex function of how the system’s drive states, represented goals, and self-model interact with perceptual input (Dennett, 1991 and 2001a; Desimone & Duncan, 1995; Cooper & Shallice, 2000; Franklin, 2000). We know such information has won out since, after all, it’s what we largely act on and can remember, report, and include when formulating intentions and plans. Again, it’s not that unconscious representations, e.g., masked stimuli and post-hypnotic suggestions, don’t have their influence on behavior, but typically it’s conscious representations that rule as we go about our cognitively flexible business.
Once we discount the seeming observation of private phenomenal facts, we can begin to see that specifically phenomenal aspects of information – the highly integrated, self-and-objects-in-a-world aspect, and the resistance-to-further-representation aspect – involve nothing beyond the functioning of the global workspace in which representations of the self and environment dominate in controlling behavior. Such integrated representations dominate because we need a global, integrated, reality model to behave effectively, and, for reasons of cognitive efficiency, we don’t need to immediately know – that is, non-conceptually represent – facts about the representational processes that contribute sensory content and build the integrated model. Crucially, consciousness isn’t something extra over and above the ever-fluctuating processes that instantiate such representations; it doesn’t ‘arise’ out of them or get ‘generated’ by them, rather it’s a property of those processes (Clark, 1995; Churchland, 1999; Dennett, 2001). The highly integrated nature of conscious content is both phenomenally apparent (as phenomenally integrated selves, we experience coherent objects in a coherent space-time context) and neurally realized (in the high degree of sub-system communication manifested in the global workspace). Likewise, the cognitive impenetrability of qualia is phenomenally apparent (indeed, it’s the mark of qualitative particulars) and an empirical fact about the limits of representation.
Ongoing research will eventually answer in detail Warfield’s question (section 2 above) about the characteristics of representational states that endow them with phenomenally conscious content: they participate in as yet obscure higher-level binding processes that instantiate a global reality model, a model which in turn supports abilities whose hallmarks are behavioral flexibility and the use of robust, multi-modal information about the world, in contrast to the more discrete, modular capacities subserved by unconscious processes. This picture of what constitutes the phenomenal contrasts markedly with Block’s (2001), in which qualitative consciousness is something categorically other than informational representation, integration and dominance in behavior control. According to Block, what this something is remains deeply mysterious, barring some major new conceptual or empirical development. My claim is that there may seem to be some mysterious, extra, non-informational, categorically private phenomenal component attached to qualitative experience, but there isn’t. Once we’ve subtracted the seemings of private phenomenal facts, and accounted for cognitive impenetrability and informational integration, we’re well on our way to capturing and explaining the phenomenal.
Given this account of consciousness, what are its implications for the causal role of phenomenal experience and some of the standard puzzles about qualia? The causal efficacy of consciousness is secured on this account, since, if phenomenal states are nothing over and above the highly integrated, workspace-monopolizing representational content necessary for certain types of higher-level, flexible behavior, then phenomenal states themselves necessarily play a causal role in such behavior. The specifically phenomenal aspects of experience I’ve covered here – cognitive impenetrability and content integration – are entailed by the representational architecture necessary for our behavioral capacities. Although Dretske says ‘Maybe something else besides experience would enable us to do the same things…’ (1995, p. 122), it seems plausible that the representational functions empirically associated with consciousness necessarily instantiate experience. In reducing consciousness to a certain variety of behavior-controlling content (but not eliminating it), the ‘epiphenomenalist suspicion’ is laid to rest (Flanagan, 1992, ch. 7; Kim, 1998). Nor can qualia be absent on this account, and for the same reasons. Any free-standing, independent system that does what I do, cognitively, will necessarily instantiate a self-in-the-world representational architecture and thus (on Metzinger’s account) a subjective phenomenal world of self and objects in which qualia can be picked out as sensory elements. Unfeeling zombies are doomed (Dennett, 1999) as soon as we stop supposing that phenomenology is something other than the integrated information required for the control of flexible behavior.
As for the intuition that there might exist inverted or shifted qualia between similar or even identical representational systems, the informational hypothesis suggests that since phenomenal qualities are fully fixed and accounted for by representational processes, the subjective feel of phenomenal information is perforce the same for similar systems. We will most likely continue to have the intuition that ‘my’ blue might be different from ‘your’ blue, precisely because of the seeming functional arbitrariness of qualia, a seeming explained by our representational limitations. However, the intuition may start to lose its grip if we keep in mind that there is no first-person fact about the essential characteristic of qualia over and above the variability that distinguishes qualia from each other, a variability common to similar systems. The feel of information is simply to non-conceptually represent things as being a certain way in the service of higher-level behavior control, and to suppose that similar systems might harbor different feels is just to be in the grip of the private fact picture of consciousness.
Systems which differ from us in their representational capacities, for instance in the stimulus range of a modality (seeing ultraviolet colors), or in the type of sensory modality (a bat’s echolocation), support differing qualitative phenomenologies; that is, they instantiate different ranges or types of non-conceptual first-order representations of the world that directly represent different facts about it (what a nice ultraviolet scarf you have on). As Nagel (1974) has pointed out, as sensory representational capacities diverge, so the ability to non-conceptually imagine what it’s like to be another creature, in terms of sensory experience, diminishes. Not having instantiated such representations, we can’t reproduce them in our mind’s eye, that is, elicit them in experiential simulations. But although we are barred from directly representing these sorts of facts about the world (that is, barred from certain sorts of sensory experience), we can have conceptual knowledge of such facts. Differences in phenomenology, therefore, do not constitute unbridgeable differences in what creatures know about the world, only differences in how they know about the world. By categorizing the regularities among non-conceptual content into abstract cognitive schemes, concepts permit the development of a more and more objective, in the sense of shared, understanding of the world, in that they abstract away from both the individual’s and type consciousness’ takes on the world, both of which are heavily selective takes due to the motivated, survival-driven nature of cognition (Biro, 1993, pp. 184-9). Such objectivity, however, is not limited or thwarted by categorically private subjective facts about experience (there are no such facts), only constrained by the pace of conceptual development.
But how will we know if a given sort of creature or system is phenomenally conscious or not? Do fish feel pain? Do ants? Once we have the necessary and sufficient neural correlates of sensory experience in hand, plus a complete account of content that connects the physical, functional, representational, and phenomenal levels (the really hard problems), the empirical test for phenomenal sentience will be to compare the target system, in its representational architecture, to systems such as our own which we know instantiate phenomenology. There may be no clear, bright line to be drawn between systems that likely embody experience and those that don’t (Metzinger, 2003, pp. 558-62), or between systems that instantiate different types of consciousness, just as there may be no clear line in one’s own case about what’s conscious and what’s not (consider the edge of your visual field, or the sensation in your foot you were pretty much (but not totally?) unaware of until just this second). But the representational account will give us a heuristic to impute consciousness: look for the minimal required representational set-up, and then be liberal in granting sentience, just to play it safe.
The lingering suspicion for many, perhaps most, readers is likely that what’s essentially qualitative or phenomenal is still being left out on this representationalist account. But what is essentially phenomenal or qualitative about consciousness? If it’s the character of particular qualia, that’s been accounted for as the informational content of multi-dimensional sensory representations that contribute information to higher level, bound, object representations. If it’s the seemingly intrinsic, ineffable, not further specifiable thusness of qualia, that’s been accounted for by our representational limitations – we can’t directly grasp the dimensional aspect of sensory content. If it’s subjectivity, that’s been accounted for as the construction of the subject/world distinction within experience – a functionally essential, robust piece of non-conceptual representational content. If it’s that consciousness gives us an entire, immediately present, coherent world, with the self at its center, that’s accounted for in terms of the representational architecture of the reality model which is neurally instantiated by the global workspace.
But, you persist, what about conscious experience itself? Well, what about experience do you have in mind? If you feel with regards to qualia that there’s still something this account doesn’t capture, but you can’t quite specify it, that’s to be expected. That, after all, is the mark of the qualitative: it’s something specific, but you can’t say what it is, and we’ve explained why it’s both specific and unspecifiable. Some might object that what’s been left out on this account is, for instance, the vivid, concrete, subjective interiority of experience (Chalmers, 1996; Nagel, 1974, 1986, and 1998). This objection employs what are in fact adjectival descriptors that pick out variation among experiences and suggests, wrongly, that they are essential aspects of all experience which a representational account can’t capture. In fact, the informational hypothesis explains why some (but, importantly, not all) sensory experiences are vivid, that is, relatively more intense; why the elements of sensory experience (but not experience as a composite phenomenal whole) are phenomenally concrete and monadic, i.e., seemingly intrinsic; and why experience (usually, but not necessarily always) seems an internal or inward subjective phenomenon, i.e., presented to a phenomenal subject felt to be here inside behind my eyes.
But how can mere information, however integrated and impenetrable, be the marvelous, multi-faceted, completely engrossing world of qualitative subjective consciousness? The answer is that for each of us, this isn’t mere information, it’s personally crucial behavior-guiding information without which we wouldn’t last for long. It’s information that’s represented as inescapably significant and that, precisely because it’s neurally instantiated, has ineluctable effects on behavior. It is alternately lovely, terrible, exhilarating, saddening, and of course imbued with extraordinary detail and variety far beyond what we can express in mere concepts (not to disparage these very useful and recent cognitive innovations). Since we can’t transcend this representation and its connections to behavior – since we consciously and cognitively consist of it – there’s no escaping the motivational hegemony of experience, for good or for ill.
Such considerations might help convince those like Kanwisher, who still wonder ‘why perceptual awareness feels like anything at all’ (2001, p. 90, original emphasis), that the ‘hard problem’ has been considerably attenuated as a philosophical conundrum. To feel is to non-conceptually, sensorily represent the world in the context of a sufficiently articulated and ramified self-world model. Both Nagel (1998) and Block (2001, pp.198, 212) suppose that some major empirical discovery or conceptual revolution perhaps in the far-flung future (Nagel: ‘long after we are all dead’) is needed to solve the problem of the fundamental nature of the phenomenal, but I think the revolution is well underway, driven by current research. If, as I and other functionalists and representationalists suppose, there is no extra-functional or non-representational private aspect of the phenomenal to account for, then as Dennett has long maintained, we’re home free, but, I would add, without having quined – that is, eliminated – qualia (Dennett, 1990). Once we’ve quined the seemings of private first-person facts about experience presented to a non-experiential observer, there’s no reason that qualia, conceived of as the reported basic particulars of experience, can’t persist in our theories as non-conceptual representations in good standing, with a relatively clear-cut neural, functional, and behavioral basis.
The ontological status of sensory, phenomenal consciousness as being a certain variety of representational content is, on the face of it, obscure, since although the processes that instantiate such representations (the representational vehicles) are clearly physical, their content – the represented facts about the world – seems less obviously so, since after all many different physical realizations of more or less the same representational content are possible. So, in locating phenomenal consciousness in the world described by science, we are led to say yes, consciousness is physical (since fundamental entities described by science involved in neurally realized conscious representational processes, entities such as proteins, molecules, and electrical charge, are what’s physical), but nevertheless it’s essentially representational. Conscious phenomenal content is ‘in the head’ since that’s where the representational vehicles are, but you won’t see it there, since content isn’t the sort of thing that can be seen, rather it’s the currency of seeing, sensing, and understanding. So which is ontologically more basic, the physical (body) or the phenomenal (mind)? One answer is neither, since on the one hand physical processes are, it seems, essentially involved in conscious representations, and on the other it seems that the very notion of the physical is a derived component of the elaborate, abstract conceptual scheme we call science that, while it constitutes the route to maximal shared objectivity, is itself built on sensory, non-conceptual representational capacities that when integrated into a self-world model constitute phenomenal consciousness.
Given deeply entrenched folk-psychological intuitions about consciousness, e.g., that it simply can’t be instantiated by physical processes, it is predictably counter-intuitive that sensory qualia might be nothing over and above informational contributions to integrated representations that dominate in the control of higher-level, flexible behavior. If you suppose there’s still something left out on this account, then my guess is you’re thinking of consciousness in terms of presentations of private facts about experience, not the world, given to a non-experiential subject that has observational access to experience. But if, as I hope to have made plausible, no such animals exist, and we simply consist of experience as represented subjects-in-a-world, then there is no theoretical or empirical obstacle to reconfiguring our concept of the phenomenal to exclude the notion of private facts, and thus to fully naturalize qualia and consciousness.
© Thomas W. Clark, June 2005
Many thanks to the anonymous reviewers for their insightful comments on an earlier draft of this paper.
Baars, B. (1988), A Cognitive Theory of Consciousness (Cambridge, MA: Cambridge University Press).
Baars, B. (1996), ‘Understanding subjectivity: global workspace theory and the resurrection of the observing self’, Journal of Consciousness Studies, 3 (3), pp. 211-16.
Baars, B. (1999), ‘Contrastive phenomenology: a thoroughly empirical approach to consciousness,’ in Block et al., 1999.
Biro, J. (1993), ‘Consciousness and objectivity’, in Davies and Humphreys 1993, pp. 178-196.
Block, N. (2001), ‘Paradox and cross-purposes in recent work on consciousness’, Cognition, 7, 197-219.
Block, N., Flanagan, O., Guzeldere, G., eds. (1999), The Nature of Consciousness: Philosophical Debates (Cambridge, MA: MIT Press).
Chalmers, D. J. (1995a), ‘Facing up to the problem of consciousness’, Journal of Consciousness Studies, 2 (3), pp. 200-219.
Chalmers, D.J. (1995b), ‘Absent qualia, fading qualia, dancing qualia’, in Metzinger, 1995.
Chalmers, D.J. (1996), The Conscious Mind (New York: Oxford University Press).
Chalmers, D.J. (1999), ‘First-person methods in the science of consciousness’, Consciousness Bulletin, Fall, pp. 8-11.
Churchland, P.M. (1989), A Neurocomputational Perspective (Cambridge, MA: MIT Press).
Churchland, P.S. (1988), Neurophilosophy (Cambridge, MA: MIT Press).
Churchland, P.S. (1999), ‘Can neurobiology teach us anything about consciousness?’, in Block et al., 1999, pp. 127-140.
Clark, T.W. (1995), ‘Function and phenomenology: closing the explanatory gap’, Journal of Consciousness Studies, 2 (3), pp. 241-54.
Cooper, R. and Shallice, T. (2000), ‘Contention scheduling and the control of routine activities’, Cognitive Neuropsychology, 17, pp. 297-338.
Crick, F. and Koch, K. (2003), ‘A framework for consciousness’, Nature Neuroscience, 6 (2), pp. 119-26.
Damasio, A. R. (1999), The Feeling of What Happens: Body and Emotion in the Making of Consciousness (New York: Harcourt Brace).
Damasio, A. R., Grabowski, T., Bechara, A., Damasio, H., Ponto, L.L.B., Parvizi, J., Hichwa, R.D. (2000) ‘Distinctive patterns of subcortical and cortical brain activation associated with self-generated emotions and feelings’, Nature Neuroscience, 3 (10), 1049-56.
Damasio, A.R. (2003), Looking for Spinoza (New York: Harcourt).
Davidson, D. (1987), ‘Knowing one's own mind’, Proceedings and Addresses of the American Philosophical Association, 60, pp. 441-58.
Davies, M. and Humphreys, G.W. eds. (1993), Consciousness (Oxford: Blackwell).
Deikman, A. (1996), ‘“I” = awareness’, Journal of Consciousness Studies, 3 (4), pp. 350-6.
Dehaene, S. and Naccache L. (2001), ‘Towards a cognitive neuroscience of consciousness: basic evidence and a workspace framework’, Cognition, 79, pp. 1-37.
Dehaene, S. ed. (2002), The Cognitive Neuroscience of Consciousness (Cambridge, MA: MIT Press).
Dennett, D. (1990), 'Quining Qualia', in Marcel and E. Bisiach, eds, 1988, reprinted in Lycan 1990, pp. 519-47.
Dennett, D.(1991), Consciousness Explained (Boston: Little, Brown and Company).
Dennett, D.C. (1999), ‘The zombic hunch: extinction of an intuition?’, Royal Institute of Philosophy millennial lecture, http://ase.tufts.edu/cogstud/papers/zombic.htm.
Dennett, D. (2001a), ‘Are we explaining consciousness yet?’, Cognition 79, pp. 221-237.
Dennett, D. (2001b), ‘The fantasy of first-person science’, debate with David Chalmers, http://ase.tufts.edu/cogstud/papers/chalmersdeb3dft.htm.
Desimone, R. and Duncan, J. (1995), ‘Neural mechanisms of selective visual attention’, Annual Review of Neuroscience, 18, pp. 193-222.
Dretske, F. (1995), Naturalizing the Mind, (Cambridge, MA: MIT Press).
Feser, E. (2001), ‘Qualia: irreducibly subjective but not intrinsic’, Journal of Consciousness Studies, 8 (8), pp. 3-20.
Flanagan, O. (1992), Consciousness Reconsidered (Cambridge, MA: MIT Press).
Franklin, S. (2000), ‘Deliberation and voluntary action in “conscious” software agents’, Neural Network World, 10, pp. 505-21.
Guzeldere, G. (1995), ‘Is consciousness the perception of what passes in one’s own mind?’, in Metzinger 1995, pp. 335-357.
Humphrey, N. (2000), ‘How to solve the mind-body problem’, Journal of Consciousness Studies, 7 (4), pp. 5-20.
Jack, A. I., Shallice, T. (2001), ‘Introspective physicalism as an approach to the science of consciousness’, Cognition, 79, pp. 161-196.
Jackson, F. (1982), 'Epiphenomenal Qualia', Philosophical Quarterly, 32, pp. 127-36, reprinted in Lycan, 1990, pp. 469-477.
Jackson, F. (2001), ‘Mind and illusion’, Royal Institute of Philosophy Lecture, ‘Minds and Persons’ series, 2000-2001, forthcoming in Philosophy, supp. vol. 53, pp. 253-273.
Kanwisher, N. (2001), ‘Neural events and perceptual awareness’, Cognition, 79, pp. 89-113.
Kim, J. (1998), Mind in a Physical World: An Essay on the Mind-Body Problem and Mental Causation (Cambridge, MA: MIT Press).
Lehar, S., (2003), ‘Gestalt isomorphism and the primacy of subjective conscious experience: a gestalt bubble model’, Behavioral and Brain Sciences, 26 (4). Online at http://cns-alumni.bu.edu/~slehar/webstuff/bubw3/bubw3.html.
Levine, J. (1983), ‘Materialism and qualia: the explanatory gap’, Pacific Philosophical Quarterly, 64, pp. 354-61.
Levine, J. (1993), ‘On leaving out what it’s like’, in Davies and Humphreys, 1993, pp. 121-36.
Loar, B. (1999), ‘Phenomenal states’, 2nd version, in Block et al., 1999, pp. 597-616.
Lycan, W. G. (1987), Consciousness, (Cambridge, MA: MIT Press).
Lycan, W. G., ed. (1990), Mind and Cognition (Cambridge, MA: Blackwell).
Mangan, B. (2001), ‘Sensation’s ghost: the non-sensory “fringe” of consciousness’, Psyche, 7(18), http://psyche.cs.monash.edu.au/v7/psyche-7-18-mangan.html.
Marcel, A. J. (1988), ‘Phenomenal experience and functionalism’, in Marcel and Bisiach, 1988, pp. 121-58.
Marcel, A. J., Bisiach, E., eds. (1988), Consciousness and Contemporary Science (Oxford: Oxford University Press).
Metzinger, T., ed. (1995), Conscious Experience (Thorverton, UK: Imprint Academic; Paderborn: mentis).
Metzinger, T. (1995a), ‘The problem of consciousness’, in Metzinger 1995, pp. 3-37.
Metzinger, T. (1995b), ‘Faster than thought’, in Metzinger, 1995, pp. 425-61.
Metzinger, T. (ed) (2000a), Neural Correlates of Consciousness: Empirical and Conceptual Questions (Cambridge, MA: MIT Press).
Metzinger, T. (2000b), ‘The subjectivity of subjective experience: a representationalist analysis of the first-person perspective’, in Metzinger, 2000a, pp. 285-306.
Metzinger, T., Walde, B. (2000), ‘Commentary on Jakab’s “Ineffability of Qualia”’, Consciousness and Cognition, 7, 353 – 362.
Metzinger, T. (2003), Being No One: The Self-model Theory of Subjectivity (Cambridge, MA: MIT Press).
Nagel, T. (1974), ‘What is it like to be a bat?’, Philosophical Review, 83, pp. 435-451, reprinted in Nagel 1979, pp. 165-80.
Nagel, T. (1979), Mortal Questions (Cambridge: Cambridge University Press).
Nagel, T. (1986) The View From Nowhere, (New York: Oxford University Press).
Nagel, T. (1998), ‘Conceiving the impossible and the mind-body problem’, Philosophy, 73 (285), pp. 337-52.
Noe, A. (2002), ‘Is the visual world a grand illusion?’, Journal of Consciousness Studies, 9 (5-6), pp. 1-12.
Parvizi, J. and Damasio, A. (2001), ‘Consciousness and the brainstem’, Cognition, 79, pp. 135-59.
Rosenthal, D. (1993), ‘Multiple drafts and higher-order thoughts’, Philosophy and Phenomenological Research, 53 (4), pp. 911-18.
Revonsuo, A. (2000), ‘Prospects for a scientific research program on consciousness’, in Metzinger, 2000a, pp. 41-56.
Tye, M. (1995), Ten Problems of Consciousness: A representational theory of the phenomenal mind, (Cambridge, MA: MIT Press).
Van Gulick, R. (1980), ‘Functionalism, information, and content’, Nature and System, 2, reprinted in Lycan, 1990, pp. 107-29.
Van Gulick, R. (1993), 'Understanding the phenomenal mind: are we all just armadillos?', in Davies and Humphreys 1993, pp. 137-54.
Van Gulick, R. (1995), ‘What would count as explaining consciousness?’, in Metzinger, 1995, pp. 61-79.
Velmans, M. (1996), ‘Consciousness and the “causal paradox”’, Behavioral and Brain Sciences, 19 (3), pp. 538-42.
Velmans, M. (2002), ‘How could conscious experiences affect brains?’, Journal of Consciousness Studies, 9 (11), pp. 3-29.
Vogeley, K. and Fink, G. (2003), ‘Neural correlates of the first-person perspective’, Trends in Cognitive Science, 7 (1), pp. 38-42.
Warfield, T. (1999), ‘Against representational theories of consciousness’, Journal of Consciousness Studies, 6 (1), pp.66-9
Weiskrantz, L. (1997), Consciousness Lost and Found: A Neuropsychological Exploration (New York: Oxford University Press).
 Thomas Metzinger (1995a) p. 20
 For a fact to be irreducibly first-person, private, and subjective bars it from scientific categorizations or descriptions that are prototypically objective. As Dennett (2001b) suggests, there can’t be a first-person science of consciousness.
 Similarly, in an earlier paper (1996) Velmans says that ‘Once one accepts that consciousness and its contents (viewed from a first-person perspective) provide valuable psychological data, one can get on with the business of working out how given conscious states relate to given forms of processing in minds or brains (viewed from a third-person perspective) in a way that does not prejudge either the ontological nature of consciousness, or its causal status’ (original emphasis).
 Of course differences in experienced intensity can sometimes be the result of top-down processes, e.g., in visual illusions which result from hard-wired perceptual ‘assumptions’ about light, shade, boundaries, and objects.
 Although they get short shrift in my discussion, I do not mean to minimize the role of internal schemas that provide default representational assumptions for the system, nor am I making a claim about the moment-to-moment completeness or detail of internal representations as discussed, for instance, in Noe (2002).
 Because the structure of subjective color space is asymmetric, it’s plausible that simple or wholesale inversions of color qualia are ruled out, and indeed that carefully imagined intersubjective qualia differences are very tightly constrained (Van Gulick 1993, pp. 144-5). Chalmers has argued that significant differences in qualitative experience are nomologically impossible for functionally identical systems. However, he allows that this ‘does not refute the possibility of a very mild spectrum inversion’ (Chalmers 1995b, p. 325).
 Another way to see this is that we can’t change our perspective on qualia, only on represented objects in experience. As I move around the blue chair, my represented perspective on it changes, and so my experienced qualia change. But I can’t achieve a perspectival, observational relationship with (that is, ‘move around’) any of the qualia involved in the experience (Clark, 1995).
 As Jackson (2001) puts it, ‘I know only too well the residual feeling that redness could not be got out of the physical picture alone, but that is nothing more than a hangover from the conflation of instantiated property with intensional property [the property of representing that something is red]. That “redness” is not a feature one is acquainted with, but instead is a matter of how things are represented to be’ (original emphasis).
 Metzinger asks rhetorically, implying an answer in the affirmative: ‘Does the extremely high-dimensional form of a bodily state, which I do not recognize as such, lie concealed behind Yves Klein’s “dimensionless depth” of the subjective sensory quality International Klein Blue?’ (1995b, p. 449, original emphasis).
 Of course this world need not, as when we’re dreaming, correspond in real time to anything in the actual world, although when we’re awake it does so sufficiently to keep us out of trouble. The vertiginous strangeness of thinking of consciousness as a virtual world is evoked when Metzinger observes: ‘…a fruitful way of looking at the human brain, therefore, is as a system which, even in ordinary waking states, constantly hallucinates at the world, as a system that constantly lets its internal autonomous simulational dynamics collide with the ongoing flow of sensory input, vigorously dreaming at the world and thereby generating the content of phenomenal experience’ (2003, p. 52).
 This section seeks to implement Nicholas Humphrey’s suggestion (2000), that in explaining consciousness we must work from both the philosophical, conceptual side and from the neuroscientific, empirical side, bringing them into mutual accommodation.
 Metzinger & Walde: ‘Not only is it impossible to experience hue without saturation or brightness, but it is also impossible to experience hue plus saturation plus brightness without an integrated percept – typically segregated from a background. Conscious experience seems to start on the object level, and elementary states in the true sense of the word do not exist’ (2000, p. 6).
 Indeed, to realize and believe in our gut that the reality model was ‘merely’ a model would undermine its effectiveness; naïve realism is arguably a necessary illusion for survival. As Metzinger puts it, ‘The evolutionary advantage of the underlying dynamical process of constantly confusing yourself with your own self-model is obvious: It makes a selfless biological system egotistic by generating a very robust self-illusion’ (2000b, p. 301, original emphasis).
 For general theoretical constraints on the notion of content see Van Gulick (1980), and for a well-articulated, comprehensive theory of phenomenal content, see Tye (1995). More recently Jackson (2001) has suggested that phenomenal representations have distinctive features of conveying contextual richness, immediacy, and causal origination, while playing a special functional role in determining beliefs.
 On our way, but with far to go, since there are further aspects of consciousness to be accounted for, e.g., other global properties of experience adduced by Metzinger (‘presence’, ‘dynamicity’, and ‘convolved holism’, 2000b, p. 286), and the sorts of non-sensory phenomenal content mentioned by Mangan (2001).
 Externalists about content such as Dretske claim that conscious content is instantiated only by historically grounded representational functions (e.g., evolutionarily-derived sensory capacities), so that a creature microphysically identical to me that sprang into existence a minute ago (Donald Davidson’s Swampman, 1987), would be a non-conscious zombie since it has no such history (Dretske, 1995, pp. 141-51). But on my account, my microphysical duplicate would behave just like me in the same circumstances, in which case such functional, representational equivalence would entail its consciousness. To entail consciousness, representational functions need not have an actual history, they only need to perform the proper cognitive, behavior-controlling role, even if such a role is normally historically derived.
 It is controversial whether such non-conceptually represented sensory facts are indeed facts in any good standing about the world, since, to take color for an example, there seems to be precious little in common, in terms of reflectance properties, among objects in the world we represent as being blue (Dretske, 1995, p. 89). Nevertheless, color constancy must track reliable and real consistencies in the world, however obscure, for if it did not, we couldn’t rely on colors as behaviorally useful discriminations, which they obviously are. Sensory facts are the most system-relative sorts of facts, but facts nevertheless.
 Not fully accounted for here, of course, but I think by Metzinger (1995b, 2000a, 2003).
 Regarding vehicle vs. content, Metzinger (2003, p. 166) observes that ‘The vehicle-content distinction is a highly useful conceptual instrument, but it contains subtle residues of Cartesian dualism in that it always tempts us to reify the vehicle and the content, by conceiving of them as distinct, independent entities. A more empirically plausible model of representational content will have to describe it as an aspect of an ongoing process and not as some kind of abstract object’ (original emphasis).