This article was sent off to the Journal of Consciousness Studies almost two years ago, and I have never received either a rejection or an offer to print it. Some of the characterizations of Dretske here I now know to be unfair to him, but I stand by the basic line of argument that evolutionary explanation won't close an explanatory gap where reduction fails us.

Evolutionary Explanation and the Hard Problem of Consciousness


Steven Horst

Wesleyan University



David Chalmers (1995, 1996) has recently done philosophy the favor of distinguishing the "hard problem of consciousness"&emdash;why it is that conscious phenomena appear in the world at all&emdash;from the "easy" problems such as the ability to discriminate, categorize, and react to enviornmental stimuli and the focus of attention. (One assumes that the choice of the word 'easy' for these problems is intended to be somewhat droll, and true only by comparison with the hard problem in which he is primarily interested.) Chalmers argues that the hard problem cannot be solved in physicalistic terms, and suggests that consciousness be viewed as being or involving a distinct kind of fundamental property in addition to those required for basic physics. Chalmers' arguments are directed against all attempts to explain the phenomenological, experiential, first-person side of consciousness in physical terms&emdash;those that do so by way of neuroscience as well as those that would try to do so directly; those that appeal only to properties internal to the conscious being and those that appeal to relational (physical) properties.

I happen to think that his arguments are successful, as are those of Kripke (1971), Nagel (1974), Searle (1992), Jackson (1982), and Horst (1996). However, that is not the subject of my paper. Instead, I wish to examine whether a different form of naturalistic explanation&emdash;in this case, explanation in teleofunctional, evolutionary terms&emdash;can succeed where explanation in physical terms is seen to fail. If one agrees that we cannot answer the hard problem of consciousness in physical terms, this can be viewed as an examination of whether evolutionary explanation can save naturalism about the mind. If one is not yet convinced that consciousness cannot be explained in physicalistic terms, what follows may be viewed more modestly as an examination of whether evolutionary explanation of consciousness can contribute anything to the solution of the hard problem not already contained in more structurally-based forms of physical explanation. What I shall argue is that evolutionary explanation does not provide a solution to the hard problem: indeed, it would need to be supplemented by a more traditional physicalistic account to do so, and hence contributes nothing towards the solution of the hard problem, in spite of being a viable and useful form of explanation with more modest virtues.

Teleofunctions and Evolutionary Explanation

Would-be naturalizers of the mind have taken as their models a number of different paradigms from the natural sciences. Perhaps the most influential of these traces its roots to the Galilean method of resolution and composition, according to which explaining a phenomenon involves breaking it down into its component parts and then demonstrating how the behavior of the parts necessarily produces the behavior of the whole. Classical reductionism, type physicalism and local supervenience accounts are all inspired by this model. Recently, however, there have been two important kinds of moves away from the Galilean model in philosophy of mind. The first is the growing movement towards externalism, in which things going on outside of the organism can play a role in determining the nature of its mental states. The second is the re-emergence of approaches to the mind drawing upon the paradigm of the Darwinian revolution in biology in the work of writers like Millikan (1984), Papineau (1993), Dretske (1995) and Flanagan [], among others. What is distinctive about this form of explanation is that appeals to the function of a phenotypic feature of an organism play a crucial role in the explanation of that feature, and the presence of that functionally-characterized feature is explained historically through a process of natural selection.

I should stress here that the notion of "function" that is involved in such explanations is a teleological notion&emdash;in very rough, pre-theoretical terms, the function of a phenotypic feature is the selective advantage conferred upon the organism or upon the population bearing the gene for that feature. This use of the word 'function' should be carefully distinguished from the mathematical notion that is used in machine functionalist views of the mind.

There are differences in the details of how writers who champion evolutionary approaches to the mind try to explain mental features, but I think that these are by and large irrelevant to the line of investigation to be pursued here. I shall therefore give a schematic account of how evolutionary explanation proceeds, first in biology and then in psychology. Evolutionary explanation involves two mechanisms: variation (or mutation) and selection. A phenotypic feature first appears in a population through a process of mutation, which is generally understood by contemporary biology to be a random process. Most mutations are harmful, many are fatal. Some, however, confer advantages for their possessors in the biologically crucial task of passing on one's genes. These advantages may consist either in advantages conferred upon the individual organism that increase its chances of surviving long enough to breed (either by increasing the chances of longevity or making it more likely that it will breed sooner) or in advantages that increase the chances of survival of the gene in the offspring (by increasing the number of the offspring or increasing the chances of their viability, say by making them less attractive to predators, shortening the gestation period or increasing parental vigilance). The process of selection is one in which statistical forces operate to increase the chances that more adaptive phenotypic traits will endure and proliferate. As a result, the explanation of a phenotypic trait T will be of the following form:

T is present in population P because
(a) T was produced by way of spontaneous mutation in one of the ancestors of P, and

(b) T conferred upon its possessors in the ancestors of T selective advantage A (e.g., it enabled them to run faster, or detect bugs under tree bark, or produce more offspring).

It should be noted that not all phenotypic features are products of selection. Some are free riders carried on the same gene with traits that were selected for. Others may have become widespread for no reason connected with selective advantage. It is simply that evolutionary biology does not provide the right conceptual machinery to account for them.

In the case of psychology, then, evolutionary explanation will treat kinds of mental phenomena&emdash;whether faculties or kinds of mental state and process&emdash;as phenotypic traits of the organisms that possess them, and will attempt to explain them in terms of the selective advantage that accounts for their proliferation. In many cases, this will be a normal biological explanation, spanning over generations within a species; but this form of explanation can also be adapted to shorter-term adaptations within an individual organism, as learning and perception, for example, can be viewed as more rapid processes involving spontaneous variation, adaptation and selection. (Cf. Sayre, 1986, Millikan, 1984.) The function of a psychological phenomenon is then understood in terms of the selection history. To say that a certain cell in the frog's visual system is a "bug-detector" is not so much to say something about what that cell does in this particular frog (perhaps its bug-detector is damaged, or it is never exposed to flying insects), as to report the operation performed by cognate cells in its progenitors which made them more viable frogs: in this case, allowing them to efficiently detect flying insects in their visual field and eat them. Likewise, one might view the function of pain experiences in terms of the detection of tissue damage or immediate threats to bodily integrity, and perception in terms of the detection of (salient) objective features of an organism's environment.

Existence of a Phenotype and Teleofunctional Essentialism

Before passing on to the topic of consciousness, I should pause to note that there are two different things about a phenotype that one might wish to explain in evolutionary terms. The first is its existence in a given organism or population: why do woodpeckers have long pointed bills? Because their ancestors who had the gene for long pointed bills were better able to feed themselves, hence survive and reproduce, than those cousins with shorter or duller bills. Why do animals have kidneys? Because those of their ancestors that developed renal systems were better able to eliminate harmful wastes within their bodies, hence better able to survive and breed than cousins that could not do so. The other thing that one might wish to explain, however, is the nature of a phenotype, by way of a form of biological or teleofunctional essentialism: what is that thing on the front of the woodpecker's face? It is a tool for extracting insects from beneath tree bark. What are those things hooked up to the bladder? They are devices for extracting impurities&emdash;because that is the action whose adaptive advantage accounts for their proliferation. (And hence your kidneys are still kidneys even if they do not in fact serve this purpose due to some form of kidney disease&emdash;what they are is determined, not by what they in fact do, but by what the phenotype of which they are tokens was selected for.) I wish to handle these topics separately in this paper. The question that is most obviously relevant to the hard problem of consciousness is that of whether evolutionary explanation can account for the existence of concsiousness; thus I shall address that topic first. I shall then examine whether evolutionary accounts of the nature of consciousness can make up for any shortfall in accounts of its existence that do not appeal to teleofunctional essentialism.

Dretske's Biological Explanation of Consciousness

Let us now consider an example of biological explanation in psychology. I shall use Fred Dretske's work as an example, largely because it is simpler and hence more easily presented than other biological theories such as that of Ruth Millikan. I do not believe that anything essential to my examination trades upon the differences between accounts. Dretske (1995) offers a representational account of both cognition and consciousness, and in fact the two notions turn out to be closely linked. To have a thought about a thing is to have a mental representation of it. Representation, in turn, is cashed out in terms of two notions: indication and function. A indicates B if A carries the information that B is present. But not all cases of indication are cases of representation. A represents B only if A has the function of indicating B. In the case of artefacts like writing and speech, the function is conventional in origin, and depends upon the actions of agents. But in the case of organisms, the origin of the function is natural, and is cashed out in biological terms, as outlined above. Both sensations and thoughts, for Dretske, are cases of natural representation, the difference between them deriving from a distinction between "systemic" and "acquired" indicator functions, respectively. (Dretske, 1995:6&emdash;19) (These are denoted representationss and representationsa.)

Consciousness, for Dretske, turns out to be closely related to natural representation. A conscious state is simply a state through which we are conscious or aware of something&emdash;Dretske uses the terms 'conscious' and 'aware' as synonyms (1995:98)&emdash;and "seeing, hearing, smelling, tasting, and feeling are specific forms&emdash;perceptual forms&emdash;of consciousness; consciousness is the genus; seeing, hearing, and smelling are species." (1995:99) In short, "conscious states are natural representations&emdash;representationss in the case of experiences and representationsa in the case of thought. Conscious creatures are creatures in whom such states occur." (1995:104) Here we have an account of the nature of both state consciousness (i.e., the sense in which a mental state is said to be a conscious state) and of creature consciousness (the sense in which a being is said to be conscious) that depends upon a teleofunctional notion of representation.

This characterization "yields a plausible and natural answer to questions about the function and purpose of consciousness." (1995:116) And this is an important question for Dretske:

If some mental states and processes are conscious, others not, one can ask, along with Rugg (1992, p. 275), whether conscious ones are more effective than unconscious ones. What is the point, the biological advantage, of having conscious states and processes? Those that are conscious must differ in some relevant way from those that are not. If this is not the case, then, as Davies and Humphrey (1993b, pp. 4-5) conclude, too bad for consciousness: "Psychological thoery need not be concerned with this topic." (Dretske 1995:116-117)

The answer, given Dretske's characterization of consciousness, is fairly straightforward. Animals need perception to do such things as find mates and food and avoid predators, and on Dretske's theory, consciousness goes hand in hand with perception. "Take away perception&emdash;as you do, according to the present theory, when you take away conscious states&emdash;and you are left with a vegetable." (1995:118) However, blindsighters and people with various kinds of agnosias can enjoy informational sensitivity while lacking the experience normally associated with perception. If the same results could be achieved without experience, why is perception accompanied by experience? Dretske's answer (admittedly only a sketch of a much fuller answer that would need to be supplied by detailed scientific research) is that persons and animals with these kinds of deficits do not in fact have all of the same abilities to negotiate their environments as do conspecifics without the deficits, and hence "it remains clear that people afflicted with these syndromes are always 'deeply disabled'." (1995:121) And thus

there seems to be no real empirical problem about the function, or at least a function, of sense experience. The function of sense experience, the reason animals are conscious of objcts andtheir properties is to enable them to do all those things that those who do not have it cannot do. This is a great deal indeed. If we assume...that there are many things people with experience can do that people without experience cannot do, then that is a perfectly good answer to questions about what the function of experience is. That is why we, and a great many other animals, are conscious of things. Maybe something else besides experience would enable us to do the same things, but this would not show that experience didn't have a function. All it would show is that there wasmore than one way to skin a cat&emdash;more than one way to get the job done. It would not show that the mechanism that did the job didn't have the function of doing it. (1995:121-122)

I include this extended quote because of the way it nicely spells out the extent of Dretske's commitment to biological explanation, and how it is supposed to work.

Does Biological Explanation Explain Consciousness?

Now I wish to address two questions about this account of Dretske's. First, is it any kind of explanation at all? And second, if so, does it solve the hard problem of consciousness? The first question rears its head because there is a long tradition (among proponents of mechanistic explanation) of casting doubt upon explanations that turn upon teleological notions like function. However, it should be apparant that biological explanation is good at explaining some things, even if it does not explain the same things that mechanistic explanation explains. Biological explanation can explain why a phenotypic trait is present in an individual or a species, provided that there are viable stories to be had about (a) the emergence of that trait in at least one individual through a process of variation, and (b) the survival and proliferation of that trait in a population through the conferral of selective advantage (adaptedness) upon its bearers relative to other members of the population. In point of fact, biological explanation seldom actually accounts for the process of mutation that leads to the initial appearance of the trait. This is so for two reasons: first, these processes are believed to be random, hence anomic, and hence not subject to special explanations. (A Lamarckian theory, by constrast, would require more in the way of explanations of mutation, as it regards these as non-random.) Second, the kinds of explanation that would be needed here&emdash;a biochemistry and/or biophysics of DNA-change and an embryological explanation of how particular DNA sequences produce particular phenotypic features&emdash;are largely beyond the scope of current science. (Particularly in the case of the latter.) However, for purposes of explanation of species change, it is generally regarded as a harmless idealization to treat the mechanisms underlying mutation and embryology unspecified.

The exceptions, of course, are cases where there is reason to regard the production of a particular phenotype, or a particular change in phenotype, by these methods as problematic. One would be suspicious, to say the least, of a biological explanation that depended on the idea that any mutation could produce within an animal an organ that served as a perpetual motion machine, because one has reason to doubt that there can be perpetual motion machines. And likewise catastrophist theories of evolution, which countenance the possibility of mutations from, say, reptiles to birds or mammals in a single mutation, have come under suspicion because it is hard to see how there could be a mechanism that would produce such changes all at once, or do so in identical ways in a sufficient number of offspring to sustain a breeding population.

Thus, what evolutionary exlanation really explains is the proliferation of a phenotype, given ihe plausibility of its initial appearance. The initial appearance is treated as something that can plausibly be attributed to random processes of mutation, ultimately to be explained by biochemistry or biophysics and embryology. Given these assumptions, selection tells a useful and genuinely explanatory story about the function and proliferation of the phenotype&emdash;and arguably a story that cannot be told in mechanistic terms.

What, however, does this contribute towards the solution of the hard problem of consciousness? The answer, I think, is very little. For what such a theory can give an account of is why consciousness would flourish, given that it has appeared in the first place. And this seems quite reasonable&emdash;creatures that are conscious are likely to have great adaptive advantages over those that are not, and particular forms of consciousness are likely to confer particular kinds of adaptive advantage. All of that seems correct insofar as it goes. But what this does not do is explain how consciousness comes upon the scene at all: It does not tell us (a) how the mutation that first conferred consciousness came about, or (b) how some feature of DNA gives rise to consciousness in beings who possess it. (Even if this explanation is divided into an explanation of how DNA gives rise to physiological structure and physiological structure to the capacity for consciousness, the problem is not lessened.) In short, there is nothing about the specifically evolutionary or selectional side of the story that sheds any light upon the existence of consciousness&emdash;and this is precisely where the hard problems lie.

Moreover, note that kinds of explanations that would need to be supplied by biochemistry, biophysics and embryology are physical and structural explanations, and precisely the kind of explanations that Chalmers et. al. have called into question with respect to the hard question. As a result, this is not one of the cases in which it is safe to treat the emergence of phenotype through spontaneous variation as a harmless idealization: there is reason to doubt that physicalistic properties determine consciousness, and therefore there is reason to doubt that the mechanisms underlying biological mutation could produce consciousness in the first place. Evolutionary explanation could explain the presence of consciousness in us given the assumption that it appeared in our ancestors through random mutation. But if no DNA structure could determine (the capacity for) conscious experience, then the selective story never gets off the ground.

Let me draw an analogy to make the point absolutely clear. Suppose someone conjectured that some species of animal was powered by a perpetual motion machine. One can certainly see how such a power source would be to an animal's advantage: it would not need to take in energy through nutrition to replentish itself, and hence would not be subject to certain hardships of privation that would imperil those around it. In short, this is a phenotypic feature that would be highly likely to proliferate, and a selective story would be easy to tell about it. The problem is that no biological mutation can produce a perpetual motion machine, and hence the selective advantage it would confer (however large) cannot be explained in this way. Likewise, if (as Chalmers et. al. have argued) no structural, or more broadly, physical properties of organisms can determine consciousness, the evolutionary story does not get off the ground, since the selective story never has anything to say about how a phenotype first appears, or how it is derived from genotype. (Of course, one might have selectional stories to tell about non-physical traits that could be passed on as well, but this would not be a form of strictly biological&emdash;or naturalistic&emdash;explanation.)

The moral of the story should be clear: an evolutionary story about consciousness can only explain consciousness if there is a story about mutation and embryology about how the physical properties of genotype can give rise to the phenotype in at least one individual. The selectional story contributes nothing to this explanation, but only explains the survival and proliferation of phenotypic features that have already appeared on the scene. In short, a naturalistic evolutionary story about consciousness presupposes a physicalistic story about the emergence of the phenotype somewhere in the history of the species. If physicalistic theories cannot address the hard problem, evolutionary theories will provide the naturalist no solace.

Two Objections

However, one might intervene here with two objections. First, this critique has only considered evolutionary explanation of the existence of consciousness in isolation for evolutionary and teleofunctional accounts of its nature. Perhaps these might allow us to circumvent the problems developed above. Second, there is an important disanalogy between the explanation of consciousness and the explanation of a purported perpetual motion organ&emdash;namely, that we know that the former exists at least as surely as we know that the latter does not exist. As a result, perhaps we are entitled to treat the assumption of the existence of consciousness as a harmless abstraction after all. I shall address these objections in order.

The first objection might go as follows: there are differences between two types of evolutionary explanation. One type of explanation explains features that are not themselves defined in teleofunctional terms by selectional history. The second type explains features that are themselves teleofunctionally-defined by reference to their selectional history. The second type of explanation looks in some ways like a definition or a tautology, as the very same features that make a feature an F explain the proliferation of F's. Now if one is a teleofunctional essentialist about consciousness&emdash;as Dretske and Millikan both seem to be&emdash;the very nature of consciousness is to be understood as that of a feature whose essential properties consist in the function it was selected to perform&emdash;e.g., enabling the animal to see, hear, smell, etc. objects in its environment. In short, consciousness simply consists in whatever faculty is that confers these abilities, and its essential property is that of conferring them.

Now how does this affect the hard problem of consciousness? It really depends on what is included in the biological function of consciousness. Is it part of the function of consciousness that it do the things that it does in a way that involves the phenomenological properties that are the subject-matter of the hard problem, or does it treat these as non-essential concombinants? Let us consider the first case first: the function of consciousness is understood in terms both of what it allows the organism to do (see, hear, etc.) and how it does it (namely, in a way involving a phenomenology). On this use of the word 'consciousness', it picks out a feature that essentially has a phenomenology, even though there might be other mechanisms that give the same informational sensitivity without it. (In Ned Block's terminology (1995), it incorporates both access consciousness and phenomenal consciousness.) If this is what the teleological essentialist means, the criticisms I advanced earlier are untouched&emdash;for the problem remains of how mutations in DNA could produce a state having these properties.

If, on the other hand, the definition of 'consciousness' is narrowed to include only the conferral of adaptive advantage and excludes the phenomenological concommitants, the problem shapes up differently. In this case, the selectional story does not call for a prior physicalistic story that explains the emergence of the phenomenology, because the phenomenology does not enter the selectional and teleofunctional story. This, however, does not so much solve the hard problem as ignore it. In this case, the teleofunctional essentialist is simply using the word 'consciousness' in a different way from the way it is used by those interested primarily in the phenomenology. But this does not mean that these features are not real, even if they are not subject to biological explanation. (Compare: not all phenotypic features of animals are products of selection, but they are not less real as a result.) The teleofunctionalist may give up on the project of giving a teleofunctional account of phenomenology&emdash;and may even be right to do so&emdash;but the problem of explaining such features, and hence the hard problem of consciousness does not go away as a result. It may well be that phenomenal consciousness is not a property that confers selective advantage on its bearers. Or, alternatively, it may be that it does confer this advantage but its presence cannot be accounted for by biochemistry, biophysics and embryology. But even if Davies and Humphrey are right that this means that "psychological theory need not be concerned with this topic" (1993:4-5), this does not mean that the hard problem goes away. It merely means that it is not solved by psychological theory, and its solution is not needed for scientific psychology to proceed apace. (Cf. Horst, 1996, Chapter 11.)

Finally, let us return to the analogy drawn earlier between biological explanations of consciousness and of perpetual motion organs. The analogy consisted in the fact that, however straightforward the selectional story for such features would be (because both traits would confer clear advantages upon their possessors), the biological explanation would be imperiled by the implausibility of explaining the appearance of the trait in the first place in physicalistic terms (in the form of biochemical, biophysical and embryological stories). But of course there is an important disanalogy between the two cases as well: we know (or at least have very strong reason to believe) that no physical device can be a perpetual motion machine, and we know at least equally well that there are such things as conscious states in human beings, particularly in one's own case. There is a difference between assuming that a thing that definitely does exist is a product of a mutational process and assuming the same thing about a thing that definitely does not exist!

This is, of course, correct. But it does not damage my argument in the slightest. No one who thinks that there is a hard problem of consciousness believes that consciousness does not exist. Nor do they dispute that, if one could explain consciousness in physicalistic terms, one would thereby have an explanation of it that could be further exploited in an evolutionary theory. The problem lies in the combination of facts that: (1) there seems to be a problem with providing a physicalistic explanation of phenomenology, (2) the selectional side of the biological story presupposes the explainability of the emergence of a trait in biochemical, biophysical and embryological terms, and (3) this explanation would necessarily be a physicalistic explanation. In short, biological explanation of the hard problem of consciousness is no more plausible than the physicalistic explanation of the initial appearance of a trait, because the selectional story contributes exactly nothing to the solution of this particular problem, but only tells why such traits would proliferate once they appeared.


I should repeat that I have not attempted here to argue that the hard problem of consciousness cannot be given a physicalistic solution. What I have argued, rather, is that, if it cannot be given a physicalistic solution, it cannot be given a solution in teleofunctional biological terms either. The reader who is persuaded by the arguments of writers like Kripke, Nagel, Searle, Jackson, Chalmers and Horst may thus read this as a refutation of the possibility of naturalizing consciousness in Darwinian terms. The reader who is not thus persuaded may see it as a reduction of questions about two kinds of naturalism to a question about a single kind.





Block, Ned. 1995. "On a confusion about the function of consciousness." Behavioral and Brain Sciences. []

Chalmers, David. 1995. "Facing up to the problem of consciousness." Journal of Consciousness Studies, Vol. 2 No. 3. Special Issue, Part 1.

Chalmers, David. 1996. "The Conscious Mind." New York: Oxford University Press.

Davies, Martin and Humphries, G. W. 1993. Consciousness. Oxford: Blackwell.

Dretske, Fred. 1995. Naturalizing the Mind. Cambridge, Mass.: MIT Press.

Flanagan, Owen. []

Horst, Steven. 1996. Symbols, Computation and Intentionality: A Critique of the Computational Theory of Mind. Berkely and Los Angeles: The University of California Press.

Jackson, Frank. 1982. "Epiphenomenal Qualia." Philosophical Quarterly 32: 127&emdash;136.

Kripke, Saul. 1971. "Naming and Necessity." In Semantics of Natural Language, ed. D. Davidson and G. Harman. Dordrecht: D. Reidel.

Millikan, Ruth. 1984. Language, Thought and Other Biological Categories. Cambridge, Mass.: MIT Press.

Nagel, Thomas. 1974. "What is it like to be a bat?" Philosophical Review 83 (1974): pages 435&emdash;450.

Papineau, David. 1993. Philosophical Naturalism. Blackwell.

Sayre, Kenneth. 1986. "Intentionality and Information Processing: An Alternative Model for Cognitive Science." Behavioral and Brain Sciences 9(1): 121-138.

Searle, John. 1992. The Rediscovery of the Mind. Cambridge, Mass.: MIT Press.