WEBTAKE

Tom Clark is thinking about the question: How essential is consciousness to highly evolved or intelligent life?

Think: is it possible that you could have a "...cognitively equivalent, but silicon-instantiated twin [that] might get around in the world as effectively as [you] do - with as much purposeful, self-protective acumen - but without feeling a thing?"

And if consciousness isn't essential for intelligence at or near our level, then what is its function, if any?

Read Tom Clark on consciousness. And bring your epiphenomenalist suspicions with you.

On the Cognitive Contingency of Consciousness

Thomas W. Clark

Center for Naturalism

twc@naturalism.org

Abstract

This paper attacks the claim that phenomenal consciousness is only contingentlyassociated with higher cognitive capacities, or, as it is often formulated,that qualia might be absent in creatures or artifacts with our level ofpractical intelligence. Since it is likely that neural processes correlatedwith phenomenal consciousness are functionally central in human cognition,then any equivalently intelligent creature or artifact will consist of functionallyequivalent processes that will necessarily instantiate consciousness. Onthis account, qualia are information-bearing representational states whichnormally dominate in the control of intelligent behavior, thus they cannotbe absent when such behavior occurs, no matter what the physical realizationof the intelligent system. Since conscious states are essentially functional,the question of why certain classes of functional processes are conscioushas no answer beyond the precise delineation of their functional role.

I

To some philosophers it seems plausible that phenomenal or sensory consciousness- raw feels, qualia, what it's like to be me - is only contingently associatedwith higher cognitive capacities. This is to say that, assuming I am a subjectof experiences, a creature who is "input-output" equivalent tome in cognitive abilities might not be such a subject. In his book, ConsciousnessReconsidered, Owen Flanagan allies himself with one version of this thesis:"It is conceivable that evolutionary processes could have worked tobuild creatures as efficient and intelligent as we are, even more efficientand intelligent, without those creatures being the subjects of experience.Consciousness is not essential to highly evolved or intelligent life"(Flanagan 1992, p. 129). Nor would it be essential, one assumes, to anyequally talented form of artificial intelligence. Minds and subjectivityare not deducible from mere behavioral sophistication or even from significantfunctional similarities to conscious creatures. My cognitively equivalent,but silicon-instantiated twin might get around in the world as effectivelyas I do - with as much purposeful, self-protective acumen - but withoutfeeling a thing.

This thesis about consciousness raises a difficult question for thosewho subscribe to it. If consciousness isn't essential for intelligence ator near our level, then what is its function, if any? Why should the subjectiveaspect of our cognitive capacities have evolved if it is indeed unnecessaryfor being efficient and intelligent? Unless we can discover a function forconsciousness then epiphenomenalism threatens: qualitative states may accompanyvarious cognitive goings-on, but they may not play a causal role in furtheringthe aims of creatures that have them. Chapter seven of Flanagan's book addressesthis issue, and although Flanagan suggests plausible functions connectedwith subjectivity, he nevertheless concedes that the "epiphenomenalistsuspicion" still survives.

There is a basic problem besetting anyone, like Flanagan, who on theone hand accepts the cognitive contingency of consciousness and then onthe other seeks to find a function for it. Whatever function one comes upwith, the reply can always be "Well, I can still imagine all that beingdone by a system or creature, for instance, Searle's Chinese Room, but thisdoesn't mean the system or creature is necessarily conscious." If functionsfor our conscious states were empirically established, then we might betempted to make ascriptions of consciousness based on finding those functionsinstantiated in whatever creature or system we are considering. But suchthird-person ascriptions would only be taken seriously by those who accepta functional characterization of consciousness in the first place. And theintuition about the possibility of absent qualia - the cognitive contingencyof consciousness - is just that consciousness cannot be functionally defined.

Since Flanagan does not mount an explicit defense of the contingencythesis (he takes it as obviously true) it's hard to say why he finds itattractive. Early in his book he says:

[I]t is crucial to emphasize that neither input-output equivalence nor even some richer form of computational equivalence is the same as strict mental equivalence. Our behavioral outputs are actions; that is, they are bodily movements identified and individuated in part by the intentions and motives that constitute them. When we drive, we signal the intention to turn. When we laugh at a joke, we are enjoying the joke. Our robotic equivalent that drives just like us but lacks a motivational economy merely moves the turn signal. Our robotic equivalent that laughs at a joke but lacks a complex cognitive and conative economy does not enjoy it or think it funny (p. 6, emphasis Flanagan).

Here Flanagan seems to beg the question about consciousness since heassumes, without argument, that our input-output and computationally equivalentrobot twin would lack motivational and conative economies, and be less cognitivelyadept than we are. But why assume this? Without expecting "strict mentalequivalence" we should at least leave open the possibility that computational,input-output equivalence, which, after all, gives the robot all our behavioralflexibility and goal-directedness, might entail consciousness. To turn theexample around, why would a robot laugh unless it found the joke funny?It is initially at least as plausible, to paraphrase Paul Churchland, thatanything input-output equivalent to us must contain a "computational-executiveorganization" which would be a "home for qualitative states"(1989, p. 38). Or as Lycan puts it: "Is it really possible to imaginesomething's sharing my entire many-leveled functional organization and stillnot being conscious in the way that I am?" (1987, p. 24 his emphasis).Of course one can imagine this, but the likelihood of absent qualia in theface of functional equivalence need not be the default assumption.

Searle, in The Rediscovery of the Mind, does take pains to defend thecontingency thesis, and a look at his arguments will show some of the difficultiesinvolved. Searle believes that "Causally, consciousness serves to mediatethe causal relations between input stimuli and output behavior; and froman evolutionary point of view, the conscious mind functions causally tocontrol behavior" (Searle 1992, p. 69, his emphasis). So Searle issaying that normally, for us, consciousness does play a central functionalrole. But Searle refuses to identify consciousness generally with a functionalrole1, since he thinks it's in principlepossible for such a role to be carried out by processes which don't resultin consciousness. He supposes that it's possible my silicon-instantiatedcognitive equivalent doesn't have qualia: "...the silicon chips didnot duplicate the causal powers of the brain to produce conscious mentalstates, they only duplicated certain input-output functions of the brain"(68). The conclusion Searle draws is what he calls "the principle ofthe independence of consciousness and behavior," which is pretty muchwhat I have called the contingency thesis:

The capacity of the brain to cause consciousness is conceptually distinct from its capacity to cause motor behavior. A system could have consciousness without behavior and behavior without consciousness (69, my emphasis).

So what's wrong here? Well, on the one hand Searle believes that consciousnessis an effect, or product of brain processes which is not essential for thecontrol of complex behavior: the brain has causal powers "to produceconscious mental states", states which might well be absent in a creaturewho is our cognitive and behavioral equal. On the other hand, he believesthat for us such effects or products of the brain normally are functionallycentral to behavioral control. But, anything functionally central to suchcontrol must, in us, be a neural process, in which case consciousness can'tbe just a "product" or an "effect" of that process whichcould somehow fail to be present as the process plays its functional rolewhen differently instantiated. Searle can't claim causal efficacy for consciousnessand at the same time imply that it "branches off" the causal chainas a possibly epiphenomenal or possibly absent by-product of whatever itis that actually controls behavior.

And if we do grant Searle the characterization of consciousness as aneffect or product, he still ends up in contradiction, since he holds tothe "rough and ready principle that we use...in science and in dailylife: same causes-same effects, and similar causes-similar effects"(75, emphasis Searle). Roughly, if my silicon twin is functionally equivalentto me, then by this principle he should be conscious, since, according toSearle, in me consciousness plays a central functional role. Searle couldsay, I suppose, that two different sorts of equally intelligent creaturesmight have differently instantiated routes to the more or less equivalentcontrol of behavior. If these routes are equivalent in their causal powers,then, by his principle, if one creature is conscious, then so will the other.If they are not causally equivalent, even though the functional controlof behavior is equivalent, then Searle says, one might be conscious, theother not. But here again the proprietary role of conscious processes canreally only be to cause or produce consciousness, since when non-consciousprocesses take their place behavior is just as intelligent. Thus it seemswrong, on this account, to claim a truly functional, behavior-controllingrole for consciousness in the first place. Since Searle strongly disavowsany form of epiphenomenalism, there is an underlying inconsistency and ambivalencein his book about just what consciousness is and does. This sort of inconsistency,I think, is bound to come up in any account of consciousness that triesto give it a functional role and includes the contingency thesis.

In what follows I want to undermine the contingency thesis by makingplausible its opposite, that phenomenal consciousness is the necessary concomitantof sophisticated cognitive systems. Subjectivity, to put it baldly, is simplyto exist as such a system when it's up and running to full, or near full,capacity. I will argue that qualia are what it is to be the sorts of representationalcreatures we are, and that their functions are simply those of the neuralprocesses empirically found, as neuroscience proceeds, to instantiate them.These functions, which apparently involve (among other things) the selectivetransformation, enhancement, and integration of peripheral sensory information,are likely to be crucial to "efficient and intelligent" cognition,however it is realized.2 Thus subjectivityqua subjectivity turns out not to have a special causal role; it's justthat the representational functions which instantiate higher cognition alsoinstantiate subjectivity.

What sorts of functions these are in us can only be precisely determinedby neuroscience. My general proposal here, however, is that these functionsare normally dominant in the control of intelligent behavior. The empiricalfact is that whatever is conscious tends to directly and preponderantlyinfluence the course of our actions, even though unconscious motives andinformation can have their effects. To be a subject is to exist as thosesorts of representational states and processes which normally predominatein the control of complex, adaptive, flexible, and novel behavior, and itis the current situation of being a particular organism in a particularenvironment which determines the content of those states and processes.

By identifying the causal role of consciousness with the causal roleof cognitive processes essential to intelligence, this thesis undercutsthe epiphenomenalist suspicion. But it has its own associated problems.If certain classes of neural activity are empirically shown to instantiatesensory consciousness, why just these classes and not others? What is itabout the functions of these activities that make them instantiate subjectivity?Why are other neural processes, for instance those which transpire whenI'm in a dreamless sleep, unconscious and not conscious? The empirical (andonly) answer to this will consist of contrasting those functions operativewhen I'm asleep or unconscious with those normally present when I'm awake.It will simply be the case that some functions, those that predominantlycontrol behavior, constitute consciousness, while others don't. There won'tbe any deeper answer to the question, still looming, "But why justthese functions?". Neural activity in the brain during dreamless sleepjust isn't the sort (it doesn't involve the "right" kinds of informationaltransformations, enhancements, and integration) to instantiate consciousness.Beyond seeing what the functions of conscious states actually are thereis no answer to why they constitute consciousness.

The question also arises of just how much information and what sortsof transformations, enhancements, and integration are required to "achieve"consciousness. Horseflies have representational states which dominate indetermining their behavior, but does this make them conscious? The answerto this may prove equally unsatisfying. There may be no principled way todraw precise lines in saying just how much representational sophisticationinstantiates subjectivity and thus in granting or withholding ascriptionsof subjectivity to other species. To establish that certain functions instantiateconsciousness cannot rule against the possibility that some analogous, butmore primitive, functions might not also. My position thus leads to an agnosticliberalism: when in doubt, which is much of the time, assume subjectivity.

II

At various points in his chapter on the function of consciousness, Flanagansuggests that it plays a central integrative and coordinating role in managingour interaction with the world:

What is consciously accessible is primarily just what we have the mostneed to know about: conditions in the sensory environment, and past facts,and events. (134)

Consciousness is multimodal, involving the discrimination of many differenttypes of states, things, and events... (135)

[C]onsciousness facilitates performance on many activities, despite beingnot absolutely essential for these activities... (139)

Consciousness, it seems, is usually involved with most, if not all, higherorder cognitive functions, including memory, anticipation, speech, learning,planning, and complex motor activity. Why, then, should there be any doubtof its causal efficacy and its necessary connection to cognition? The doubtarises when it is seen that some of the same functions can be carried out,to some extent, nonconsciously, for instance in the better than chance responsesof blindsighted subjects in guessing about objects in their blindfields.And it arises when systems ordinarily not credited with subjectivity (computers,robots, Chinese Rooms, Block's Chinese nation, beer can ensembles, etc.)are imagined to have achieved intelligence comparable to our own. This sortof artificial - or just bizarre - intelligence need not or cannot, manysuppose, give rise to an inner life. Such considerations seem to show thatdespite the apparently central role of consciousness in our cognitive lives,it might 1) play no such role in us, hence, be strictly unnecessary ("notabsolutely essential" as Flanagan puts it) for our intelligence, orperhaps 2) play a central role for us but, when that role is instantiatedby a different sort of system, it doesn't result in consciousness.

Flanagan's discussion of blindsight and other cognitive deficits (inwhich he critiques some of Ned Block's (in press) recent proposals) showsthat the first of these claims is empirically well worth doubting. In thisdiscussion, Flanagan distinguishes between what he calls "experientialsensitivity" and "informational sensitivity". Subjects withblindsight are not experientially sensitive to objects in the blindfield,that is, they deny any conscious awareness of them. Their better than chanceperformance in various forced choice situations, however, suggests thatthey are informationally sensitive to what's in the blindfield. That is,some information comes through the visual system which is available to controlbehavior (the forced choice or guess), even though they aren't consciousof seeing anything.

Flanagan notes, significantly, that if I am phenomenally aware of somethingthen in general I have access to information about it as well. I can reportmy phenomenal states and reason and act using the information availablefrom them. In defining experiential sensitivity Flanagan says that "weare experientially sensitive to what we are phenomenally aware of"(148), but in addition we are, to use Flanagan's terminology, informationallysensitive to what we are phenomenally aware of. In fact, to be phenomenallyaware of something - to be experientially sensitive to it - is just to bemaximally informationally sensitive to it, or so I propose. According tothis gloss, Flanagan's distinction between informational and experientialsensitivity marks out a continuum of purely informational salience or dominance.Subjectivity is constituted by those more central representational processeswhich transform and enhance peripheral sensory information to the pointwhere it normally dominates in the control of behavior.

As amazing as blindsight abilities may first seem to those who firsthear of them, the prosaic bottom line is that they don't really compareto the phenomenally informed abilities of normal subjects. This is hardlysurprising since most of our everyday experience shows that to be phenomenallyaware of something is to be in a far better position to act appropriatelyand effectively with regard to it. If cognition is defined, partially, asthe ability to reliably discriminate objects in one's environment (and one'sown internal states), and discriminate these in the service of one's currentdesires and in the light of one's relevant beliefs, then according to thisdefinition those with blindsight are at a crucial cognitive disadvantagein their blindfields. Whatever blindfield information gets through for thesesubjects only contributes to behavior when they are prompted to respond;it never generates the spontaneous goal-directed activity typically associatedwith being conscious of an object that one might wish to use.

The difference in performance between blindsighted and normal subjectscan best be explained as a representational deficit: no sufficiently structuredrepresentations of objects placed in the blindfield are available to contributeto the sorts of behavior that characterize the responses of normal subjects.I suggest that this representational deficit precisely constitutes the absenceof any phenomenal awareness of the object. Flanagan virtually reaches thisconclusion when he says

"The information the blindsighted person has may be degraded. This may be part of the reason that the information does not reach phenomenal awareness. Or it may be that the information is not degraded but simply untransmissible due to the nature of the lesion in [the visual cortex]. In either case the inference to the best explanation is that if the information were to become phenomenally conscious, performance would improve. (148)"

I would offer a slight clarification, or amplification. The reason performancewould improve is not because a certain chunk of information would becomeconscious, but because better integrated, more structured, and more appropriatelytransformed information would be present to control behavior. It is thisincrease in informational integration and structure which best accountsfor better performance and which constitutes becoming conscious of an object.Information becomes phenomenally conscious when it is appropriately organized(into discriminated objects, for instance) so that it can more efficientlyand intelligently control behavior. To have a phenomenal experience of anobject is to be maximally informationally sensitive to it in the sense thatthe informational states constituting the phenomenology are those optimallystructured and placed to produce effective action. A little further on Flanagansays

"[T]here is some plausibility in inferring that the lack of phenomenal awareness of the blindfield at every stage of processing partly explains the inability to bring the knowledge the system possesses ... into normal, high-quality play in inference, reporting, and action" (149).

Again, the lack of phenomenal awareness just is the inability of thesystem to generate sufficiently enhanced, integrated representations ofthe sort which ordinarily contribute the most to inference, reporting, andaction. Having qualitative feels, and hence phenomenal awareness of objects,"explains" improved cognitive performance in the sense that suchawareness just is the functioning of normally dominant representationalprocesses. Information can, and often does, contribute to behavior withoutconstituting phenomenology, but not nearly to the extent of phenomenologyitself under ordinary conditions. (Exceptions might be subjects acting underpost-hypnotic suggestion. They find themselves behaving in ways which seembizarre because their actions are dominated by an unconscious intentionplanted earlier. But even here the conscious sensory surround remains theprimary constraint on behavior.) The upshot is that although phenomenologymay not be absolutely essential to the rudimentary performance of some activities,the crucially enhanced performance made possible by the processes whichinstantiate it is essential to being the sorts of creatures we are.

That qualia are behavior-controlling informational states seems not aterribly farfetched hypothesis. After all, it is the phenomenal experienceof the various qualities associated with objects, including our bodies,that serve normally and for the most part as the cues which govern action.If I desire an apple, it can only be an object of a certain color, shape,feel, texture, and smell which starts (and keeps) me eating it. As PaulChurchland describes it, my conscious sensory responses are covariant mappings,in various neural state spaces, of my situation vis a vis the apple (Churchland1989, pp. 102-110). As a subject I consist, partially, of such mappingsas they develop through time since they constitute my sensations of red,heft, sweetness, wetness, etc. Any defect in my representational abilities(e.g. closing my eyes, being color blind, having a cortical lesion, beingasleep) means that I am that much less informed about what I'm eating andhence at a cognitive disadvantage in getting what I want. Any gap in normalphenomenal awareness, whether the result of peripheral or more central dysfunction,represents an representational deficit which potentially compromises theeffectiveness of my behavior.

To identify qualia with functionally central representational statesmay seem to beg the question at issue: the possibility of their absencegiven such states. But the claim here is only that if we, uncontroversially,have inner lives, then our inner lives consist of existing as such states.Unless we are ontological dualists, or unless we decide with Dennett thatqualia and phenomenology don't really exist (Dennett 1991, p. 365), thebest candidates for qualia are the neural processes correlated with reportsof awareness in each sensory domain. To have a conscious sensation is toexist as such a process. Such processes, I have tried to show, are cognitivelycentral for us since it is empirically the case that when they are disruptedor abridged, as in blindsight, performance deteriorates dramatically.

Another objection might be that the qualitative state "red herenow" can't serve a functional role because it's possible to imaginethat my normal sensation of red (my cue to eat the apple) could have beensomething quite other, say my normal sensation of blue, and still have contributedjust as effectively to my behavior. Don't Dennett's thought experimentsabout changed or inverted qualia (Dennett 1990) show that the phenomenalway the world looks to us just drops out when behavior is at issue? Allthis may be so, but nevertheless, qualia, as discriminative sensory states,are highly informative. It isn't that red qua my current, ineffable, privatesensation of red tells me a thing (Lewis 1988, Lycan 1987, pp. 78-9), butthat as a response to a frequently encountered, if as yet obscurely definedensemble of wavelengths, reflectances, luminances, and so forth, it servesas a reliable indicator of particular objects or situations. Conscious sensorystates are informative not because they are, subjectively, a certain intrinsicway, but because they constitute a spectrum of differential representationalresponses (Shoemaker 1975, p. 400 in Rosenthal 1991, Churchland 1989). Onceagain, the function of consciousness isn't in subjectivity per se, but inthe function of those states, which, if one happens to be those states,constitute one's subjectivity.

A third objection (waiting in the wings for quite some time now) runsas follows. We may find empirically that conscious processes play a crucialfunctional role in controlling behavior, but, how can we be certain thatwhat constitutes consciousness is their playing that role and not some non-functionalaspect or property they possess or produce? To go back to my elaborationof Searle's position, why couldn't it be the case that there are two differentlyinstantiated routes to the equivalent control of behavior, one (neural)which just happens to "carry" consciousness with it as an epiphenomenalby-product, the other (silicon) which does not? The reply to this is toask in return why, if consciousness is just a contingent accompaniment ofneural processes, does it seem only to accompany processes which are ordinarilydominant in the functional control of complex, adaptive, novel, and flexiblebehavior.3 If consciousness is identicalto particular sorts of neural properties that aren't functional in thisrestricted, cognitive sense, why aren't we conscious of digestion, liverfunction, hormone production, etc., etc.? If we try to take the tack thatconsciousness is just a by-product only of cognitively functional processesthen this is really equivalent to saying that it isn't cognitively contingentat all, but a necessary, if non-functional, concomitant to such processes.I am saying, of course, that consciousness is not such an epiphenomenalby-product, rather it's to be these processes in action.

III

Assuming now that our consciousness plays a central functional role incognition, the question becomes whether it's possible that any representationalsystem which controls behavior as intelligent as ours does not also instantiatea subject. Could creatures cognitively and behaviorally equivalent to usbut with silicon "brains" just be unfeeling zombies? I hope thatnow it seems less plausible that they could be. If, as I think, my qualitativestates are simply those (sufficiently) enhanced and integrated representationswhich normally dominate in controlling behavior, then my silicon-instantiatedcognitive equivalent also has qualitative states. After all, it must beas representationally sophisticated as I am in order to behave equivalently,and if representations which are functionally indispensable constitute consciousnessfor me, those serving the same sorts of functions will constitute consciousnessfor it. The same sorts of functions will be found in my silicon equivalentsince, for it to manifest intelligent behavior at my level, it must be functionallyisomorphic to me to a significant degree. Thus, unless we are biologicalchauvinists (an increasingly scarce species), we must suppose it has aninner life.

If qualia are token identical to certain neural states, and if thesestates are functionally central to our sort of cognition, then any statesthat play an equivalent role for a creature or system will constitute qualia.Any creature or system that, operating on its own, can do more or less whatwe do and do it appropriately, that is, with mostly successful outcomes,should be considered phenomenally conscious. The teleological constraintsimplicit in our fundamental (if fuzzy) notion of intelligence require thatany system which achieves that sort of intelligence consists partially ofrepresentations which themselves constitute subjectivity.4 When pondering the consciousness of various imaginaryformal systems like Searle's Chinese room, or bizarre systems such as Block'shomunculi head, or our future companions, the robots, we must ask whetherthe behavior supported by the system is intelligent and appropriate, forif it were this would be the best evidence that informational states instantiatingconsciousness existed in the system. Of course we can't rule out consciousnessif the system just sits there, but an ascription of subjectivity depends(or should depend, I am arguing) on our estimate of the system's cognitivecapacities whether or not these are being overtly expressed. If we imagineonly the system itself, in all its strangeness, disconnected from an environmentin which its intelligence can be made manifest, then indeed our intuitionsmay well favor the possibility of absent qualia. But get the system actingintelligently, i.e. with some evidently self-generated purpose, then itsstrangeness will take second place to its behavioral appropriateness. Suchappropriateness, at least when it reaches our level, can only be the productof representational capacities which instantiate consciousness.

The intuition that consciousness is only contingently connected to functionalintelligence loses its plausibility once we identify subjectivity with thecomplex, synthesizing transformations and structuring of information thatseem crucial for sophisticated cognition. Confirmation of this identitydepends on finding, as Flanagan suggests (p. 11), the right sorts of empiricalcorrelations among reports of phenomenology, intelligent human behavior,and concurrent brain processes. It's possible, of course, that neurologywill show that the brain processes instantiating phenomenal experience justrun "in parallel" to those processes which do the system's realwork, so that consciousness is indeed functionally disconnected from thesystem. But it's prima facie implausible, although not impossible, thatsuch freely spinning neural "wheels", taking up valuable cranialspace and metabolic energy, would have evolved. It's far more likely thatthe neural activity instantiating consciousness plays the causal role itso plainly seems to, that of providing sufficiently structured and integratedinformation to coordinate intelligent behavior. Any representational systemthat is equivalently intelligent will consist of equivalently structuredand integrated informational states, and hence is equivalently and necessarilyconscious when those states are operative.

Thus I doubt, with Dennett, and against Flanagan, that the notion ofan intelligent but unsentient zombie, whether artificial or naturally evolved,makes sense (Dennett 1991, p. 313), but unlike Dennett, and siding withFlanagan (pp. 61-85), I don't think we need jettison the idea of qualitativephenomenology as illusory or ill-founded. Qualia are quite real, being,as they are, what it is to instantiate (or be instantiated by, take yourpick) certain representational processes. Pain, is not, as Dennett claims,best conceived as an abstract entity like a center of gravity, a haircut,or a person (1991, p. 460). It's as real as the bodily, neurological processesthat instantiate it. To have an inner life is simply to be such processes,so there is no ontological divide between the objective and the subjective,it's just that these are, so to speak, two different perspectives, first-personand third-person, on the same things.5From the outside a brain state is described in neurological terms, fromthe inside it's reported as being a particular sensation, emotion, or whatever.

If, as a result of the identity of consciousness with sophisticated representationalfunctions, a creature cognitively equivalent to us is necessarily conscious,then what about less complicated creatures? At what point would sensoryconsciousness cease to be instantiated by behavior-controlling representations?If we can pin down those neural functions in us that instantiate phenomenology,this gives us some confidence that any creature instantiating those functions,more or less, is likewise conscious. But since the complexity of behavior,function, and neural structure all vary more or less continuously acrossspecies, there may be no firm line to be drawn separating those favoredwith phenomenology from those not so lucky. At what point does the discriminativeand representational capacity of neural processes become so obtuse thatno "inner life" remains? We can't say, even though we have exampleson either side (bacteria, ourselves) about which we feel comfortable makingsuch judgments. Sensory consciousness is to exist as active representationalprocesses, so unless you happen to be the creature in question, doubts aboutyour consciousness (on the part of others) can always arise if your behaviordoesn't seem sufficiently sophisticated, flexible, or adaptive, or if yourinternal structure doesn't seem sufficiently ramified to support the sortsof representations normally thought to be required for such behavior. Butwhat's judged as sufficient here is, of course, necessarily biased by ouranthropocentric view of consciousness, in which being human (or primate,or mammalian, if we are more liberal), often seems a necessary condition.Thus the maxim again is, when in doubt, assume subjectivity. After all,how would we like being judged, and treated as, unconscious brutes by systemsvastly more representationally sophisticated than we are? (Of course, beingthat bright they would know enough to try to avoid any sort of chauvinismwhen attributing consciousness.)

In conclusion, the functional view of qualitative, phenomenal consciousnessis pretty much that consciousness is as consciousness does, and that thereare no special sorts of instantiations of cognition that alone confer subjectivity.If a determinate set of cognitive, representational functions are foundto be what consciousness is, then this identity will be necessary, holdingin all possible worlds.6 There won't,however, be a further fact which explains why such functions constituteconsciousness. This is because consciousness is nothing over and above beingthose functions, however they may be realized.

Notes

See Lycan1987 pp. 10-21 and Humphrey 1992 pp. 220-224 for discussions of identityclaims, especially as related to Kripke's (1970) arguments against the possibilityof any identity, necessary or otherwise, between the mental and the physical.

References

• Block, N. 1978. "Troubles With Functionalism," in Minnesota Studies in the Philosophy of Science, volume IX, C. Wade Savage, ed. Minneapolis: University of Minnesota Press. Reprinted in Rosenthal 1991, pp. 211-228.
• Block, N. forthcoming. "What Does Neuropsychology Tell Us about a Function of Consciousness?," in Debates About Consciousness, N. Block and O. Flanagan, eds. Cambridge, MA: MIT Press, in press.
• Churchland, P. M. 1989. A Neurocomputational Perspective. Cambridge, MA: MIT Press.
• Dennett, D. C. 1990. "Quining Qualia," in Consciousness in Contemporary Science, A. Marcel and E. Bisiach, eds. Oxford: Oxford University Press. Reprinted in Lycan 1990, pp. 519-547.
• Dennett, D.C. 1991. Consciousness Explained. Boston: Little Brown.
• Flanagan, O. 1992. Consciousness Reconsidered. Cambridge, MA: MIT Press.
• Humphrey, N. 1992. A History of the Mind. New York: Simon and Schuster.
• Kripke, S. 1970. Naming and Necessity. Cambridge, MA: Harvard University Press.
• Lewis, D. 1988. "What Experience Teaches," Proceedings of the Russellian Society, University of Sidney. Reprinted in Lycan 1990, pp. 499-519.
• Lycan, W. G. 1987. Consciousness. Cambridge, MA: MIT Press.
• Lycan, W. G. ed. 1990. Mind and Cognition. Cambridge, MA: Blackwell.
• Rosenthal, D. M. ed. 1991. The Nature of Mind. New York: Oxford.
• Searle, J. R. 1992. The Rediscovery of the Mind. Cambridge, MA: MIT Press.
• Shoemaker, S. 1975. "Functionalism and Qualia," Philosophical Studies, XXVII, 5. Reprinted in Rosenthal 1991, pp. 395-407.
• Sober, E. 1985. "Putting the Function Back into Functionalism," Synthese Vol. 64, No. 2, pp. 165-93. Reprinted in Lycan 1990, pp. 97-105.

---

Go to the home page

Questions, comments to:
ken@nodogs.org
(Last Updated: August 1, 1995)