Introduction

It’s perhaps an underappreciated fact about consciousness that no conscious experience has ever been observed. No one has ever seen a pain, a sensation of red, a feeling of elation, or any other instance of a phenomenal feel. What we can observe are the physical and behavioral correlates of such things, but experiences themselves are undergone or had, not observed. We don’t even observe our own experiences. Instead, as conscious subjects we consist of them as they succeed each other in consciousness. We might have the feeling (yet another experience) that we are in an observational relationship to our experience, but this an illusion. There is no observing self sitting behind experience that witnesses it.[1]

What we witness or observe is the world outside the head, and experience is the medium in and by which the world is presented. The world appears in experience, where part of that experience is the sense of being a self to whom the world appears. As Thomas Metzinger, Antti Revonsuo and some other representationalist philosophers suggest, the whole ensemble of experience, of a stable me looking out at a more or less coherent world, can be construed as a qualitatively expressed reality model, a virtual reality that normally corresponds at least to some extent with what’s outside the head. Each conscious being is possessed of such a model, and it constitutes for each of them a private subjective reality, in contrast with the intersubjective reality modeled by science and other empirical pursuits. Subjective realities are not visible from an external, 3^rd person perspective, only their physical correlates in the brain and body are visible. Consciousness itself is a categorically private affair, which generates the problem of other minds: you can't observe another creature's consciousness, you can only draw inferences concerning it from its behavior and internal structure.

I will argue that the privacy of phenomenal consciousness (e.g., experienced pain, red)  makes it impossible for it to play a role in 3^rd person explanations of behavior. Consciousness doesn’t appear intersubjectively, and it’s only intersubjectively observable entities and processes that can play ascertainable roles in scientific accounts of phenomena. This means that as far as science is concerned, consciousness isn’t even epiphenomenal. The whistle of Huxley’s steam engine exists along side the engine, but is causally inert with respect to its operations, hence is causally epiphenomenal. Since consciousness isn’t a publicly observable object or event it doesn’t exist intersubjectively; it doesn’t exist along side the brain and behavior, which are public objects and events. So qualitative conscious states don’t have the opportunity of playing, or not playing, a causal role with respect to the brain and behavior. Another way of putting this developed below is that consciousness doesn’t inhabit the same explanatory space as do the physical objects of scientific theories, including theories of human behavior which have at their disposal all the entities and processes intersubjectively described by biology and neuroscience.

This view does not entail that mental states such as beliefs and desires can’t play roles in 3^rd person accounts of behavior. Psychological explanations that invoke intentional states need not assume that the private phenomenal aspect of such states (when conscious) is causally effective.[2] We can ascribe beliefs and desires to agents and rightly think that as functional, information bearing states, with all the causal powers of their physical realizers, they participate in explaining behavior. But we need not suppose that the unobservable subjective experience of having such states (when conscious) adds anything to what the observable neural instantiations and physical manifestations of such states accomplish in controlling and constituting action.

The privacy constraint on consciousness

It might seem that in claiming the categorical privacy of consciousness, what I will call the privacy constraint on consciousness, I’m begging the question about what consciousness might be. After all, if conscious states turn out to be identical to some variety of physical states, for instance a certain collection of active neural processes, then wouldn’t it be the case we can observe conscious states just by observing those processes? Or if they turn out to be identical to physically realized functions, for instance neurally instantiated representational, information-carrying functions, wouldn’t we then actually see consciousness were we to observe the system carrying out those functions? The reply, it seems to me, is fairly straightforward. No: on those accounts what you’re observing isn’t conscious experience itself but its publicly observable sufficient conditions. If you were to observe in their entirety what turn out to be the sufficient conditions of my being in excruciating pain, that wouldn’t be to see the pain itself, since pain itself is only something that is had by the system (me, the human organism) that instantiates those conditions. My pain is an experienced element of a conscious reality model that appears for me alone, that exists for me alone as a particular locus of consciousness. Like other phenomenal feels, it is a categorically 1^st person reality. It is simply not in the cards for pains or other conscious experiences to become 3^rd person objects of public observation in the way that (what might turn out to be) their sufficient physical or functional conditions are public objects. No matter how hard you look at the sufficient conditions of consciousness, you won’t see consciousness itself. It is this fact, along with the qualitative nature of conscious states, that makes explaining consciousness so difficult. How does something categorically private and 1^st personal come to exist as a result of conditions in the 3^rd person world constituted by public objects? This, of course, is one way of stating the so-called “hard problem” of consciousness.

Although it may not beg the question about what explains consciousness, the privacy constraint does seem to rule out certain answers to the hard problem. If consciousness really can’t be observed from the outside, but only had “on the inside,” then it isn’t likely that it will be a causal effect or product of what the brain and body are doing. Public objects generally produce effects that are themselves publicly visible. Observable neural states and systems, for instance those that precisely correlate with my excruciating pain, have many observable effects, including my grimacing, moaning, verbal reports of pain and the like. Were pain itself a further causal effect or product of such processes, we’d expect to see it sitting there in public space, but of course we don’t. As Daniel Dennett points out, there is no “second transduction” that takes us from neural states to non-physical mental states, or at least no evidence of such a thing; there is no phenomenal “juice” or “milk” that the neural correlates of consciousness secrete which could itself become a public object.[3] Light waves, sound waves, chemical odorants, pressure and heat on my skin, and other physical inputs to my sensory system all get transduced into neural activity, but neural activity doesn’t generate a further effect beyond more neural events and, sometimes, overt behavior. If a heretofore undetected effect someday comes to light, for instance a holographic aura that shows up on a yet to be devised consciousness detector, that still won’t be consciousness, such that my experience of pain suddenly enters the public domain. My pain is, and will always be, only present to me; it only exists for me. The same goes for all my experience.

This suggests that consciousness, a categorically private affair, might be some sort of non-causal entailment of whatever its publicly observable sufficient conditions turn out to be. I won’t pursue here what I think might be a fruitful line of inquiry into the hard problem: that, as Metzinger and other representationalist philosophers suggest, our being certain sorts of representational systems entails our being conscious systems, where that entailment is a matter of instantiation, not causation.[4] I only want to say that the privacy constraint on consciousness isn’t going away, and that it prevents us from claiming a literal and complete identity between public states of affairs on the one hand and consciousness on the other. If consciousness really is private, and we are realists, not eliminativists, about its existence,[5] then it can’t literally be the same thing as its publicly observable sufficient conditions, whatever those turn out to be.

First person explanatory space

Conscious, qualitative experiences participate in what I’ll call 1^st person explanatory space, such that for instance when explaining why I wince, I might very well tell you I’m in pain. As we’ve seen, my pain is a 1^st person state or event available only to me, and I experience it as playing a causal role in my behavior. It very much seems to me that my wincing is caused by my pain: were it not for the pain I wouldn’t be wincing. Other conscious subjects will likewise attest to the causal efficacy of their phenomenal states. Martha testifies that if she didn’t have the conscious experience of tasting crème brulee, she wouldn’t be eager to order it at every opportunity. Mike says that the conscious experience of terminal boredom keeps him away from staff meetings, and so forth. All such explanations invoke private, 1^st person, subjective phenomenal feels as causally active in generating behavior. Moreover, for conscious subjects the reality of such experiences can’t be placed in doubt. Although the veridicality of an experience as a faithful representation of the world outside the head can be doubted (is this a dagger I see before me?) the sheer existence of the experience cannot. Nor can we as subjects transcend experience. There is no non-experiential perspective we can escape to from which to inspect experience. As Metzinger puts it, as conscious beings we live entirely inside our “ego tunnels.”[6] So the phenomenal experiences of 1^st person explanatory space participate in an indubitable and untranscendable individual private subjective reality that exists for each conscious subject.

Although they are not available to public observation, conscious states are routinely recruited in 3^rd person, intersubjective explanations of behavior; they are ascribed causal roles in conjunction with, but distinct from, publicly observable brains, bodies and environments. We generally tend to believe that Martha, Mike and the rest of us do what we do as a result of the phenomenal aspect of conscious states, where that aspect has causal power over and above what the brain is doing. Discussing the role of conscious qualitative states (qualia) in explaining behavior, Avshalom Elitzur suggests that “Alice’s kiss may take a bit longer thanks to the additional effect of love’s quale, and the qualia of hunger and fear may add some speed to the rabbit’s and the fox’s race” (original emphasis).[7] And writing for the National Academy of Sciences, biologist Anthony Cashmore says “I suggest that … consciousness heightens our desire to listen to music, for example, or to watch or participate in sporting activities.”[8] The general idea is clear enough: the qualitative aspect of conscious states adds an extra causal “push” beyond what their neural correlates contribute. Although I haven’t explored the empirical research on folk beliefs about mental causation, I suspect that something like this idea might be widespread.

But of course the claim that phenomenal consciousness has its own proprietary causal role generates the problem of mental causation, or more specifically, the problem of phenomenal causation: how do private phenomenal feels like pain, taken as non-identical to their public neural correlates, help to cause the physical events we call behavior?  What is their special causal contribution to behavior over and above what the physical brain and body accomplish? Although most of us are realists about the existence of conscious states and although we generally suppose they are causally efficacious, integrating them into 3^rd person explanations of behavior is notoriously problematic. No viable accounts of phenomenal-physical interaction or influence have been forthcoming, despite centuries of speculation on this problem.[9] Of course, if, contrary to the privacy constraint on consciousness, phenomenal feels turn out to be identical to some set of physical processes, then consciousness doesn’t have a special causal contribution to make since its causal powers are exactly those of the physical processes it’s identical to. This neatly solves the problem of mental causation, but as explained above, I think it's difficult to sustain the claim that consciousness is identical to its physical correlates.

Third person explanatory space

The difficulty of integrating phenomenal feels, taken as non-identical to physical processes, into 3^rd person explanations of phenomena results from the fact that such explanations deal only in public observables: what can be seen, measured and confirmed to exist by any suitably placed observer, using whatever equipment might be necessary to carry out the observation. Observables are thus the basic elements of 3^rd person explanatory space, and they constitute public intersubjective reality: what we all generally agree exists independently of the private subjective reality of each person’s conscious experience. Observables having to do with human persons – their brains, bodies, and physical and social environments – are uncontroversially real elements of the 3^rd person explanatory space having to do with human behavior.

Of course we observe and routinely take into account publicly available reports of conscious experiences when constructing explanations of behavior. But experiences themselves aren’t in 3^rd person explanatory space since they aren’t intersubjectively available. They cannot be seen, measured, weighed or otherwise captured from an outside perspective. As a result, their causal connection to or influence on what is observable, namely bodies and their constituents, is irredeemably obscure (hence the puzzle of phenomenal causation as explained above). But for each experiential report adduced in explaining behavior, we can in principle observe the accompanying physical states and processes, and it is these, not experiences, that can play well-substantiated, unequivocal roles in 3^rd person accounts of behavior. Indeed, without them there would be no behavior to account for.  

In explaining my pain behavior – wincing, moaning, reports of pain, learning to avoid exposed live electrical wiring, etc. – we can in principle trace the course of physical events that occur between my contact with an object (the live wiring, let us suppose) and all subsequent neurally realized internal states and overt behaviors. There are successive neural and/or behavioral events for each stage of the process, however we carve it up in terms of ordinary language attributions of mental states, including beliefs, desires and phenomenal experiences. In explaining and predicting my behavior, you will of course attribute to me a desire to avoid painful episodes and a newly acquired belief that contact with live wires causes pain. You will also assume, correctly, that I undergo experiences of pain. But a properly intersubjective, as opposed to subjective, explanation need not and cannot suppose there are unobservables which ground or back up these attributions, on pain (so to speak) of trafficking in explanatory posits which lack any public evidentiary basis.

Beliefs, desires and other intentional and motivational states – a subset of mental states – are perfectly real, observable attributes of the behaving system. After all, we attribute beliefs and desires on the basis of physical, functional and behavioral criteria constituted by 3^rd person observables: how I act in the presence of exposed wiring; my reports of intentional states related to avoiding contact with them; the neural circuitry subserving my harm-avoidance and damage-minimizing functions, including C-fibers and other specialized neural networks; and my brain systems subserving memory, motivation, deliberation, speech and action. Most important for the present argument, the causal efficacy of intentional mental states in explaining behavior need not appeal to anything over and above the causal powers of their observable physical realizers, whatever those turn out to be. So 3^rd person explanations of behavior can uncontroversially include this subset of mental states as physically and functionally real, as causally effective elements within intersubjective explanatory space.

Barring phenomenal feels from 3^rd person explanations

What 3^rd person explanations cannot appeal to, however, is the causal power of the conscious phenomenal aspect of mental states construed as something non-identical to their physical realizers. This is because that phenomenal aspect, unlike the physical realizers, isn’t observable – it only exists for particular subjectivities, so is a denizen of 1^st person, not 3^rd person, explanatory space. Matters would be different if we could claim a literal identity between the publicly observable conditions sufficient for phenomenal consciousness and consciousness itself, since in that case consciousness would have exactly the same causal powers as its observable realizers. But as seen above we can’t claim such an identity: the privacy constraint on consciousness rules that out. So although we commonly appeal to phenomenal feels like pain when explaining behavior from a 3^rd person perspective, I suggest we are making an unwarranted causal claim in doing so. We can’t see, observe, measure or otherwise directly ascertain the existence of phenomenal feels from an external vantage point, only their physical and behavioral correlates. Those correlates do all the causal work in properly formulated 3^rd person explanations, and such explanations can appeal to functionally efficacious intentional (mental) states as realized by some of those correlates. But an appeal to phenomenal experience as a causal factor in behavior violates the primary epistemic constraint on intersubjective accounts of phenomena: that the elements participating in such accounts be publicly observable, that they provably exist in 3^rd person explanatory space. To allow unobservable experiences to play explanatory roles in 3^rd person explanations is no more justifiable than allowing ghosts, spirits, souls or other empirically unevidenced entities to play such roles: all unobservables are properly ruled out as explanatory non-starters.

But of course the response comes: we have good evidence that phenomenal feels exist and that they play causal roles, namely that we undergo unequivocally real conscious experiences of pains, pleasures, tastes, sights, sounds – the whole rich ensemble of our qualitative lives – and we experience these phenomenal states as causally central to our lives. Doesn’t that count as good evidence that conscious experience, as distinct from the neural states that correlate with it, plays a role in 3^rd person explanations of behavior? No, it does not. I am not denying the reality of conscious experience for the experiencing subject, including the experience that such states are causally effective. Consciousness is undeniably real and exists in nature, since subjects exist in nature. But however vivid the experience of phenomenal causation might be, it doesn’t prove that qualitative experiences such as consciously felt sensations and emotions are in fact working along side physical systems in generating behavior, which is what 3^rd person explanations would require. To prove that, we’d have to observe experience operating as an additional causal factor over and above the brain, body and environment, which is exactly what can’t be done since experiences are intersubjectively invisible. Phenomenal consciousness doesn’t exist in 3^rd person explanatory space, in reality as modeled by intersubjective, empirical disciplines, so is necessarily barred from playing a role in 3^rd person explanations of behavior.

As persons, we are both conscious subjects and observable physical entities, so we are simultaneously situated in both subjective and intersubjective realities. As a result, we are well acquainted with the elements that participate in their respective explanatory spaces, 1^st and 3^rd person: private phenomenal experiences on the one hand and public observables such as physical objects on the other. We are, therefore, naturally inclined to combine elements from both explanatory spaces in our 3^rd person accounts of behavior. But good explanatory practice requires we resist this temptation, since 3^rd person accounts can’t traffic in unobservables. Resisting it, I suggest, effectively solves (or dissolves) the problem of mental causation, at least construed as the problem of how phenomenal feels could possibly influence physically instantiated behavior control mechanisms. The answer is straightforward: they don’t, and we should stop trying to concoct metaphysically extravagant, conceptually convoluted, or empirically unwarranted accounts by which they do.

In his PNAS paper, Anthony Cashmore insists that "...there must be a mechanism by which consciousness does influence behavior. There must be a flow of information from consciousness to neural activity" (p. 4). But according to my analysis, the best hypothesis is that there is no such mechanism. There is no causal interaction between consciousness and neural activity since they inhabit different explanatory spaces, those connected with subjective and intersubjective reality respectively. These spaces aren’t the sort of things that could exert either mutual or one-way causal influence on one another. Cashmore also asks what seems to be a straightforward question: “…what is the evolutionary selective advantage of consciousness?” But from a 3^rd person perspective it isn’t consciousness that was naturally selected for, but the neurally instantiated higher-level behavior control systems that we’ve recently found to be associated with consciousness.[10] Again, in rigorous scientific practice we have to segregate the unobservables from the observables in intersubjective accounts of the world, both as things to be explained and as things that do explanatory work.

Phenomenal feels as convenient explanatory fictions

Of course, declaring such a policy isn’t likely to change our everyday non-rigorous practice of citing qualitative conscious states in 3^rd person explanations of behavior, nor should it.[11] So long as we don’t imagine there’s an accepted evidence-based account of phenomenal-physical causation that backs up such claims (there is no such account), no great harm arises: phenomenal feels will continue to serve as very convenient 3^rd person explanatory fictions even when the physicalist and functionalist story behind behavior gets completely filled in. After all, their co-variance with behavior and brains states is very reliable, so their predictive power remains what it was. Since there will always be accompanying physical correlates that from an intersubjective standpoint reliably get the behavioral job done, discovering the intersubjective truth about phenomenal causation, namely that there is none, does not disempower or diminish us. We remain just as rational, smart, empathetic, sensitive and emotionally and interpersonally skilled (or unskilled) as we were before. It’s just that, strictly speaking (as we try to do in philosophy and science) the 1^st and 3^rd person explanations of these virtues don’t causally interact.

Although explanations of action that give qualitative experience a causal role are misplaced from a 3^rd person perspective, they pay important tribute to the undeniable 1^st person subjective reality of having phenomenal feels and experiencing them as causally effective. The experience of phenomenal causation is not going to go away, whatever our intersubjective, 3^rd person theory says about the causal role of consciousness. Moreover, conscious states won’t ever stop mattering to us as individual subjects, so we have no choice but to take phenomenology seriously in human affairs and the affairs of other sentient systems. For instance, we must continue to obey the ethical injunction to minimize unnecessary conscious suffering when considering the effects of individual actions and public policy.

Consciousness isn’t epiphenomenal

A perhaps surprising and paradoxical entailment of this view is that consciousness can’t be construed as epiphenomenal. For something to be epiphenomenal – for it to fail to play a causal role – it has to exist in the same explanatory space as that which it might have played a role in causing. But as we’ve seen, the stubborn and strange fact is that consciousness, although undeniably real (unlike ghosts or souls), doesn’t exist in the 3^rd person explanatory space in which its physical correlates exist. It can’t be found in the intersubjective reality inhabited by its correlates, which is limited to public observables. This means it can’t interact with its correlates – it can’t cause them, or be caused by them – so isn’t in a position of being causally inert with respect to them. When operating from a 3^rd person, intersubjective perspective it’s a category mistake to suppose consciousness is behaviorally epiphenomenal, where the categories are two mutually non-interacting explanatory spaces. Consciousness isn't epiphenomenal with respect to physical goings on, rather it's causally orthogonal.

The worry about consciousness being epiphenomenal is that it gets demoted to something extraneous or second-class in comparison with the physical, a useless accompaniment to what’s doing the real work in controlling action. But since consciousness doesn’t exist in 3^rd person explanatory space, none of these things are true of it. We can relax about the fact that scientific explanations won’t find a role for phenomenal feels like pain and pleasure in explaining what we do, since that isn’t to dethrone consciousness, but only to accept that 3^rd person explanations are limited to observables. As suggested above, however much science advances we’ll continue to conveniently explain behavior by appealing to our phenomenal states, paying tribute to 1^st person subjective reality, to the fact that we are conscious creatures. Consciousness will still be central to our lives – it cannot be dethroned since it’s the inescapable, untranscendable condition of being subjects in the first place.

Some might worry that if consciousness doesn’t play a behavior-controlling role in the same causal, explanatory space as the brain, we therefore lose power and control. But, simply put, this worry fails to give enough credit to the marvels of neurally instantiated cybernetics. The brain and body, in collaboration with their environment, don’t need consciousness to get the behavioral job done, and that, from a 3^rd person perspective, is that.

Looking at the world intersubjectively we can ask: where is consciousness? It isn’t anywhere in the sense of being an observable phenomenon directly available to science, but nevertheless it is  empirically associated with certain sorts of complex physical systems that do cognitive, world-representing work. Human beings are one instance of such systems. The philo-scientific project of doing full justice to the world in our descriptions – what we think of as attaining maximum objectivity – can’t responsibly declare consciousness non-existent, since after all here it is for each of us, an ineluctable, non-illusory 1^st person reality that each of us sincerely attests to, a fantastically rich quality space within which we as phenomenal subjects and our phenomenally presented worlds both exist. Moreover, we have made great strides in pinning down the neural correlates of consciousness (see note 10), so it’s to some extent an empirically tractable phenomenon, albeit invisible to intersubjective observation. Why, therefore, should we suppose that the private subjective reality of consciousness is somehow suspect, or plays second fiddle to the public intersubjective reality of what’s observationally available? An objective account of what’s real must encompass both realities, along with the elements that play roles in their respective explanatory spaces. How these realities arise and interrelate is the puzzle facing those seeking a unified account of consciousness in nature.[12]

TWC, 8/2010

Notes