Philosophy of Learning

In philosophy we study perception, knowledge, belief, memory, and reasoning—but not so much learning. In psychology learning is central, but in philosophy it is peripheral at best (perhaps part of philosophy of education). I propose to make some programmatic remarks about learning as a philosophical topic (I happen to have a strong interest in the subject). What is learning? It is the transition from ignorance to knowledge: at time t you don’t know something and then at time t+1 you do know it—the process between is the learning process. To define it we need to use the concept of knowledge: you can’t learn something false (“learn” is factive). To learn is to come to know. The OED gives us “acquire knowledge of or skill in (something) through study or experience or by being taught”. You can’t say “Today I learned that Paris is the capital of England”. The upshot of learning is knowledge not mere belief. So, any definition of learning will incorporate a definition of knowledge. What kind of knowledge? All kinds—any kind of knowledge can be learned: knowledge-how, knowledge-that, knowledge-what. You can learn how to play tennis, or learn that global warming is real, or learn what mango tastes like. Learning is as conceptually promiscuous as knowing. In all cases it signifies the transition from not knowing to knowing. There is no kind of learning that is not a case of knowing. You can learn consciously or unconsciously, but either way you end up knowing something you didn’t know before. Growth isn’t learning because getting bigger does not involve knowing anything. Learning always involves change, but not all change involves knowledge. All learning is epistemic (analytically so).

So far, so obvious. But we can raise unobvious questions: is it possible to learn a tautology? You can learn that Hesperus is Phosphorus, but can you learn that Hesperus is Hesperus? Intuitively, no; nor can you learn (discover) that bachelors are unmarried. Why? Because you already knew these things by knowing what the words mean; there is no transition to a new state of knowledge. If your kid came home from school and announced he had learned that bachelors are unmarried, you would want your money back. You have to learn a new proposition not an old proposition dressed in new words. Indeed, what can be learned is a criterion of identity for propositions, as Frege taught us. More challenging, what kinds of thing can learn? Can a rock learn? Not unless it can know things—and it cannot. What about genes? Here I am inclined to say yes, because genes can come to know that certain traits are conducive to survival. That is, they are naturally selected by the evolutionary process because they enable survival and reproduction. The evolutionary process is in effect a learning process: existing traits are the result of learning. The environment has taught the genes to produce certain traits and not others. The organism is a product of genetic learning. It is much the same with simple organisms: they learn things without being rational conscious agents, but by classical conditioning. Conditioning is a type of learning; so, the result must be a type of knowledge. Pigeons know which lever to peck as a result of trial-and-error learning. Genes are similar (as has often been noted) because they are the result of a selective process leading to a useful trait. So, I have no problem with the idea that genes are the result of a learning process, and hence know things. The concept of knowledge is broad enough to apply in their case—as it is to worms, rats, and baby brains. There is no reason to be snooty about knowledge and learning, as if you could only learn at school. You could in fact teach genes by artificial selection, hoping to improve the resulting organisms, thus causing them to learn. The genes of one species might be better at learning than another. Logically, the genes can learn and have done so over millions of years. This is why people speak of the genes as containing information, and of the genotype as a “book of the dead”. The analogies are strong enough to justify extending the concept of learning in this direction (and discounting the implicatures).

Here is a question: can there be Gettier cases of learning? Brief reflection suggests yes. If you see Jones parking a Ford, you will be justified in believing he owns a Ford, and this may well be true, even though the Ford he is parking isn’t the Ford he owns. You are right only by accident and hence don’t really know he owns a Ford. But by the same token, you didn’t learn that he owns a Ford; you came to believe it by accident. So, true justified belief is not sufficient for learning; learning requires full-blown knowledge, but Gettier cases show that knowledge cannot be defined as true justified belief. We can’t say that x learns that p if and only if xcomes to have a true justified belief that p. What you learn you must know, but knowledge isn’t definable as true justified belief.

Is innate knowledge learned? It is commonly said that it is not, because the animal had the knowledge all along; you don’t have to learn what you already know. So, innate knowledge might be described as unlearned knowledge. But this ignores a possibility alluded to above, namely that the genes learned it. The genes encode this knowledge, but it was instilled in them by the evolutionary process; so, they learned it from the world (“experience”) by natural selection. Thus, we can say that innate knowledge is acquired knowledge, though not acquired by the animal in question; it was acquired by the genes long ago and then passed on to the animal. Accordingly, all knowledge is acquired knowledge, i.e., learned. No knowledge is unlearned. Instinctual knowledge is learned as much as perceptual knowledge. The empiricist could correctly claim that all knowledge is learned knowledge, with the environment enlisted as teacher. The educated genes give us our innate knowledge (it wasn’t innate in them). All knowledge has to be learned from someone or something, given that evolution is the way instinctual knowledge comes about. Innate is not the opposite of learned, but a special case of it. Learning and knowledge go inextricably together.

Finally, we must say something humanistic about learning, because learning is highly prized by human beings (but not by worms and genes). People feel it is good to learn—skills, subjects, how to be a better person. The human animal is the learning animal par excellence. Why? Because we prize knowledge, and you can’t get far with that without learning. To know requires learning, and learning isn’t always easy. So, we sacrifice to learn. Knowledge is good for us, we feel, but you can’t possess that good without working for it—learning being a type of work (if only just looking around you). Learning is woven into our lives because knowledge is mentally nourishing. The quality of a man’s soul is calculated according to the learning that has gone into forming it. People say “I love to learn” and the verb is apt; we are nothing without learning. A blank slate is a worthless slate, sheer vacuity, rank nothingness. Learning is a moral imperative. We might even say it is the meaning of life. A worthy philosophical subject.[1]

[1] In this subject, as in others, it is methodologically unwise to focus exclusively on the human case; our mental concepts extend beyond us into the animal kingdom. We need to stay conceptually flexible if we want to limn the biological natural kinds. This is why anthropomorphic language comes so naturally in the case of worms and genes: they are at one end of a spectrum occupied by us (and by beings potentially superior to us—whatever that means). Thus, worms learn where the water is (and know it) and genes learn that fingers are useful (and know it). We are latecomers in the learning business not its paradigm.

Skeptical Solutions to the Mind Problem

Early thinkers had a body problem: what is body? What is the stuff we call “matter”? The pre-Socratics had a go at answering this question: matter is water, or variations on the four elements (earth, fire, water, and air). More advanced thinkers proposed invisible atoms, or spatial extension, or energy, or elemental mind. These are straight answers to the initial question—no funny business. None of them involves questioning the very existence of body, or explaining it away, or finding some sort of surrogate for it that is less mystifying. They are not skeptical about the existence of matter. They take it for granted that matter exists and try to say what it is—eminently sensible, perfectly rational. In parallel with this, people have asked what mind is: what is the thing we call “mind”? What constitutes it, what is it made of? We know it exists—we feel it at every moment—but what the devil is it? It doesn’t seem like body, so we can’t just apply our theories of matter to it; it requires separate treatment. But it strikes us as a tough nut to crack, a bit of an enigma, not the work of a couple of spare minutes. Let’s go back to thinking about matter—it’s less perplexing! We wanted a straight answer to our question: mind is real, but what exactly is it? We wanted a constitutive informative answer—certainly not one that questions the existence of mind. We don’t want to be told that mind doesn’t really exist and therefore has no nature. We want a straight solution to the problem of mind not a skeptical solution (that’s no solution).[1]

Over the years various solutions have been ventured: dualism, materialism, behaviorism, functionalism, representationalism. These all accept the existence of the mind (but see below) and offer to say what it is. But they have not been met with general acceptance; all sorts of objections have been raised. Some are too reductive, others not reductive enough, others not clear enough. Qualia-consciousness theories seem too narrow and raise the same question about themselves (what is consciousness?); materialist theories are just hard to believe on their face (compare “All is water”); immaterialist theories are plain baffling and ignore the role of the brain; behaviorist theories are simply looking in the wrong place; computational-representational theories are too speculative and suspiciously circular. In any case, no one thinks any of these are obviously correct or evidentially well-established. Frankly, they are all clearly inadequate, though not without containing some elements of truth. We haven’t been able to say convincingly what mind is. But if we can’t say what it is (we don’t know what it is), why should we say it is? Why not just go eliminative? That will get rid of the headache—and who says folk psychology is sacrosanct? Thus, we have a skeptical attitude prompted by what looks like a skeptical paradox: the mind simply has no definition, no nature, no correct theory. It’s like the gods or phlogiston or the elan vital—non-existent. So, let’s not talk about it anymore—let’s treat it with the disrespect it deserves. We already have matter, so why do we need mind? Matter is enough for any universe to be going on with. It might even be thought that mind, as normally conceived, is intrinsically impossible: a logically incoherent mental theater, a den of mysterious intentionality, a museum myth (you know the anti-mind rhetoric). Let’s admit it, mind is a baffling business. That would be one response to the difficulty of the mind problem. Junk it, bin it, commit it to the flames. Succumb to ontological skepticism.

But it is not the only possible response: we might try for a skeptical solution, i.e., a proposal that accepts that the mind itself is unreal but offers to preserve our talk of mind. The discourse can be saved while its putative subject matter is “eschewed”. Mental talk has a use but it has no reputable ontology (like religious language). It serves a pragmatic purpose but without any existential commitments. It has assertion conditions but no truth conditions—no corresponding facts. It serves the community’s interests but has no real-world counterpart (again, you know the rhetorical patter). This line is not difficult to conjure: mental discourse is instrumentally useful—a device for making predictions, summing up data, having something to say to your spouse after a long day. Maybe we can’t live without it, or retain our sanity. Still, it is pure mythology (again, like religious discourse). As an extra flourish, we might add that mental language serves an expressive function—we use it to get things off our chest, or to impress a potential mate, or to wind down at night. We do like to chatter about psychological matters, don’t we. This approach resembles the case of ethics: no ethical facts, but it doesn’t all go up in smoke, because ethical talk is emotive or expressive or psychologically healthy (blah-blah-blah). This is a skeptical solution to a felt ontological problem. There is a similar approach in philosophy of mathematics: numbers are useful fictions—non-existent but linguistically practical.

So far, I haven’t said anything particularly new or upsetting to orthodoxy, but that is about to change—radically. For I wish to suggest that all the standard theories of mind are really skeptical solutions in disguise. None of them is really a straight solution, not when you dig a bit deeper. This is fundamentally because they simply cannot be taken seriously as straight solutions; they are all tacit denials of the reality of mind as we know it. They are all ways of replacing the mind with something else and then claiming that mental discourse is really about that. Immaterialist dualism claims that talk of the mind is talk of a curious invisible gaseous substance tenuously linked to the body and brain. This is obviously pure myth, scarcely more credible than postulating immortal gods to explain natural phenomena. The brain obviously has an awful lot to do with the mind and it is not immaterial (whatever that means). The intellectual function of the immaterial substance idea is to provide a prop for mental discourse to hang onto—a sort of impalpable peg. The thought is that we can talk as if this is true and comfort ourselves that we are not talking about nothing. Okay, that may be a worthy goal—it would be hard to give up all psychological talk—but it is hardly sound ontology. This is a skeptical solution masquerading as a straight solution. It clearly beggars belief as the literal truth about the nature of mind, as people subconsciously realize (one thing it does is allow religion to peddle its myth of immortality). It is an imaginative crutch adopted for a pragmatic purpose. If a philosopher or scientist had actually proposed it as a skeptical solution, it would be regarded as such—a myth to allow us to carry on using mental language in the face of the skeptical challenge and its accompanying paradox. No one thinks that the mind looks like a cloud, or has a cloudy composition. This is just what we resort to when we have given up hope of a straight solution. It is a useful fiction.

Belief in dualism plays the role of instinct in Hume’s discussion of induction: it provides an explanation of our practices without admitting they have an objective rational basis. We are prone to imagine such things and that is why we keep talking the way we do about the mind, while accepting that we can’t say what it is, or even that it exists. Materialism also provides a basis for mental talk without being believed in as the literal truth about the mind. No one really believes it, not in their heart of hearts (except perhaps a gullible philosopher); it just provides a useful substitute for the mind as we ordinary conceive it. We know what the brain is (we think), so we can’t be made subject to a skeptical paradox about it; well, that’s what grounds our mental talk—it is as ifwe are talking about the brain. We don’t really think the mind is the brain and nothing but the brain; but if we pretend it is, then we can carry on talking as if the mind exists. The brain plays the role of the community in Kripke’s reconstruction of Wittgenstein: it is the alternative reality that can be wheeled in to ground our existing practices, given that nothing factual about the individual can ground them. Again, if a skeptical philosopher had suggested the brain as a skeptical solution, we would see the point, given that it is quite hopeless as a straight solution—the mind being clearly not the brain (as physically described). Behaviorism and functionalism are much the same: clearly not identical to the mind—located in completely the wrong place—but a viable-looking substitute for the mind proper. Behavior prompts mental assertions, so it can provide assertion conditions, but there is no account to be had of truth conditions—for there are none, what with the mind not existing and all. No one really believes that the mind we know so well from the inside is behavior, but it can play a role in explaining our mental talk; we thus eliminate the mind while hanging onto behavior—on pain of having nothing to talk about. Again, behavior plays the role of the community in Kripke’s story—it substitutes for the fact we have been unable to find. We have examined all the available facts and none measures up, so we have opted for a skeptical account of our practice with mental language. In the case of the computational-representational theory of the nature of mind, I will only say that this is a speculative and mainly metaphorical account of the mind, limited at best to only a fragment of the mental. It is best viewed as a skeptical solution not a straight solution—a useful fiction. It is as if there is a language of thought, but of course there isn’t really any such thing. In sum, all the theories offered as straight solutions turn out to be skeptical solutions in disguise. They don’t really set out to tell us what the mind literally is—they are too incredible for that—but they do offer an account of mental discourse that allows us to preserve our mental talk in a kind of diluted form, analogous to expressive accounts of ethics, fictionalist accounts of numbers, and instrumentalist accounts of physics. According to this view, then, most of the philosophy of mind consists of skeptical solutions to what-is problems—not the official story, not by a long chalk.

I have so far suppressed mention of mysterianism, but it is actually key to understanding what is going on here. For it provides an option distinct from straight solutions and skeptical solutions, namely that there is a straight solution but we can’t find it, deeply so. That blocks the move from “we can find no straight solution” to “there is no straight solution”, thereby pulling the rug out from under the skeptical solutionist. If the mind is deeply mysterious, yet has an underlying real essence, there will be no necessity to contrive a fancy (and fanciful) skeptical solution, with all of its implausibility and bullet-biting. So, the dialectic of straight and skeptical solutions does not exhaust the field. We can agree that no straight solution can be found but still insist that the mind has an intrinsic nature that preempts the need for a skeptical solution. The standard theories then lapse without needing to deny the mind or construct a skeptical surrogate for it. Skeptical solutions arise from assuming an anti-mysterian position at the outset. The point I have wanted to make in this paper is that orthodox philosophy of mind has labored under the delusion that it is giving straight solutions to (quite legitimate) problems; but the standard theories are really skeptical solutions of a familiar kind. Put very simply, philosophers have tacitly denied the existence of minds, as we ordinarily understand them, and attempted to find a way to hang onto our linguistic practices; they are like Kripke’s semantic skeptic but extended to the mind as whole.[2]

[1] For background see my “Existence, Consciousness, and Skeptical Solutions”, “Family Resemblance and Skeptical Solutions”, and “Philosophy of Popularity”.

[2] Returning to the question about body, it is fair to report that this problem has also tended to generate skeptical solutions with respect to physics, as straight solutions have come to seem more and more unsatisfactory. Hence, instrumentalism, verificationism, operationalism, structuralism, and mathematicalism. Mysterianism can block these moves; it might indeed be the prevailing state of affairs. It’s either skeptical solutions or mysterian (straight) solutions. Known straight solutions are hard to come by. This is the shape that philosophy is apt to take. We can add epistemic solutions to straight solutions and skeptical solutions.

Philosophy of Popularity

Are there any concepts for which a skeptical theory (solution) is obviously correct? Consider popularity: what is it to be popular? Suppose someone thinks that popularity is defined by a specific observable quality of a person such as good looks (suitably defined). They might even think that popularity is defined by having blonde hair and blue eyes. Such a person is likely to be surprised when they come across someone who is popular but does not have good looks or blonde hair and blue eyes, or someone who has those characteristics but is not popular. It is obvious that they don’t have the concept of popularity, which is the property of being well liked by many people. They have mistaken an obviously social-relational concept for an individual intrinsic-essence concept. Who does that? The correct analysis of the concept of popularity is as stated, not as our eccentric individual believes. This is not a skeptical solution because it is not skeptical of widespread assumptions. There is really no instance of an obviously correct skeptical solution. If everyone bizarrely believed that popularity is defined by good looks or blonde hair and blue eyes, it might be reasonable to speak of a skeptical answer to the question by giving the correct definition—skeptical of a widespread belief, that is. But there are no serious cases of that: the idea of a genuinely “skeptically” defined concept that everyone thinks is a “straight” defined concept is unheard of. No one actually believes that popularity is definable by the kind of criterion I stipulated in the case of the eccentric individual. That would be bizarre to the point of impossible.

Why is this? Because we generally know what our words mean. So, someone who believes that we mistake skeptically defined concepts for the straight kind is supposing that we can make this kind of mistake about what their words mean. If “game” is really defined in the family resemblance manner, how could anyone think that the concept of a game is not so defined? How could you be under the illusion that games have a common essence if in fact they don’t as a matter of the meaning of “game”? No one even half rational could think that popularity is defined by some observable quality independent of being liked by many people. The proper conclusion, then, is that the whole idea of skeptically defined concepts that are mistaken for straight concepts is radically misguided. No one could think that “game” has an essentialist definition if it really has a family resemblance definition, on pain of not understanding the word “game”. This is a philosopher’s myth. How could it be the case that “game” has a family resemblance definition and yet for thousands of years no one recognized that fact? It would be like everyone throughout history mistakenly thinking that “popular” means “good looking” or “has blonde hair and blue eyes” when it actually means “liked by many people”. The idea is preposterous. Thus, the contention that family resemblance has any philosophical utility is an error. There could never be the kind of revision Wittgenstein envisages. The idea that “meaning” or “rule-following” really means a kind of skeptical definition, along the lines sketched by Kripke, verges on self-contradiction; if it meant that, we would know it, so we wouldn’t need any persuading. Imagine a philosophy book on popularity that earnestly argues that popularity is not definable by good looks but by being well liked; it would not sell a million.[1]

[1] Wittgenstein’s idea that “nothing is hidden” about meaning is in flat contradiction with his contention that the meaning of “game” is given by family resemblance: for it is clearly hidden from us that this is so. We think, according to him, that “game” is an essence concept but that it is no such thing—how could that be so if meaning is transparent? There can only be skeptical solutions if meaning is hidden.

Existence, Consciousness, and Skeptical Solutions

There are straight solutions and skeptical solutions, skeptical paradoxes and skeptical arguments. They need not all take the same form; they just need to be cases in which something we uncritically assumed comes into doubt and is rescued by lowering theoretical expectations. X isn’t really the kind of thing you thought it was but this other kind of thing; the skeptical argument is rebuffed (not refuted) by recasting the nature of the problematic phenomenon. I am going to discuss the problems of existence and consciousness in this light; I claim there is an illuminating analogy between them, not so far removed from Hume’s position on causation and induction.[1] We will be proceeding at a high level of generality.

What kind of property is existence? When a thing exists what kind of fact is that? We know what it is for something to be red or square (or we think we do) but what is it to exist? The property seems elusive, peculiar, queer. Perhaps our most naïve idea about it is that existence is the property of being spatial: “X exists” means“X is in space” (not just “in the mind”). To be in space is to be a three-dimensional solid thing you could stub your toe on (see Dr. Johnson on Berkeley). It is a familiar concrete fact—a chunk of physical stuff in space. But this runs into trouble with the mind: it exists but it isn’t in space—or need not be in order to exist (we can’t rule out disembodied minds just by the definition of “exists”). And what about numbers, time, and space itself? Back to the drawing board. Now we start to get worried—what else could it be? It can’t just be being an object of thought, because we can think about non-existent things (e.g., the golden mountain). Nor can it consist in a disposition to be perceived: this is neither necessary nor sufficient (for familiar reasons). Nor can it be a sense-datum: I have impressions of red and square but not of existence per se. Nor can existence be identified with being a part of God’s mind. We then start to wax metaphysical: existence is a primitive sui generis indefinable property that we just have to accept without explanation. It is what it is and not some other thing (like the taste of pineapple, but not sensory). Or we might opt for mysterianism: existence is a real property with a complex real essence but its nature is unknowable by us. These are all straight solutions: they assume that existence is a genuine first-order property of things and then offer answers to the question of what that property is. But it turns out to be more difficult to pinpoint this property than we anticipated; and here is where the skeptic will pounce. There is, he says, no such property as existence (compare the property of being a ghost): nothing really exists (because “exists” is meaningless). The idea of existence is hollow, chimerical, nonsensical. We are thus in the presence of a skeptical paradox: common sense is revealed as common delusion. Existence is a myth. This is hard to accept: surely some things exist and some do not! It is carrying skepticism to absurd lengths—and yet no account of the property of existence has been provided. The alleged property vanishes into thin air.

Here is where a skeptical solution might bring some balm to the proceedings. Maybe it isn’t a regular first-order property at all, yet our talk of existence can be saved, if at a reduced volume, so to speak. We were wrong in our initial assumption of its ontological category: existence is real enough (or just enough) but it isn’t a property at all. Rather, it is a different kind of creature altogether: it is the circumstance of a first-order property having instances—for example, for red things to exist is for the propositional function “x is red” to have at least one instance (to be true for at least one substitution instance of that function). This is a skeptical solution because it questions the commonsense assumption that the predicate “exists” is logically predicative. True, there is no such property, but “exists” still has a meaning (a role in the language game), given by the paraphrase in terms of propositional functions and instantiation. This is a kind of anti-realist answer to the problem of existence: existence is not a real property—it is a humdrum fact about predicates having instances. We know that British mail boxes are red—well, that’s all that existence comes down to (“x is red” is true when “British mail box” is substituted for “x”). It isn’t some weird and wonderful property of objects, but a fact about truth and substitution into predicate expressions. Problem solved, or dissolved. We accept that existence isn’t a genuine property, but we can keep on using the word “exists”. In short, the concept existence is a second-order concept applicable to propositional functions (bits of language).

Much the same can be said about Russell’s treatment of definite descriptions. We start out with the question of the meaning of these expressions and look for what they refer to. Naively, we suppose the meaning is the reference, but then we notice that not all definite descriptions have reference–and yet they have meaning. What to do? We could try saying they refer to the idea corresponding to them, or possibly to themselves, but this runs into trouble as soon as it leaves the gates. So, we cast around for something more esoteric, more metaphysically charged, and we come up with non-existent but subsistent objects (we pull a Meinong). True, this gives us the ontological willies, but what’s a poor boy to do? No ordinary object will do the job. The skeptic jumps into the ring at this point and declares a knockout: definite descriptions are meaningless! Meaning is reference and some descriptions have no reference—thus, they are meaningless. Meinongian objects are just a refusal to accept this. Granted, this is a paradox, but that’s life folks! Now the wily B. Russell steps in to save the day: he proposes a skeptical solution, namely that descriptions are not referring expressions at all, but are short for a quantified conjunction of propositions. The mistake was to suppose that they mean by referring; in reality, they mean by quantifying and predicating. The paradox has been defanged, but it was right up to a point—if descriptions were referring expressions, they would be meaningless when empty. We solve (dissolve) the problem by offering a revised conception of what they are—quantifying expressions (with identity thrown in plus conjunction). We have to change our view about what definite descriptions are, but this is a small price to pay if we want to avoid Meinong and meaninglessness. Skeptical solutions are better than no solution at all. Or so we are told.

All well and good, we are on familiar ground, which I have re-described using the apparatus of skeptical paradoxes and skeptical solutions. But how does consciousness fit in? Surely, I am not going to claim that consciousness is existence! Indeed not (but the panpsychist and neutral monist come close). We have analogy not subsumption. It goes like this: we want to find out what consciousness is—what kind of feature or fact it is. So, we search for a straight solution: is it a sensation or a thought, a feeling or a quale, or intentionality, or what it’s like, or a disposition, or an informational brain state, or an immaterial perturbation? Suppose we come up with no clear answer—no straight solution to our problem. Each of these suggestions runs into trouble. What should we conclude? According to the eliminative skeptic, we should conclude that consciousness does not exist (“I am conscious, therefore I exist” is BS). For we can find nothing with which to identify it—no fact or property or state or event or process. There is nothing it is.  That is a somewhat alarming conclusion (though eliminative behaviorists will welcome it). Are we defenseless against it? No, because we can give up the assumption that consciousness is an intrinsic feature of a mental state: it isn’t like being a pain or a sensation of red or a feeling of panic; it isn’t intrinsic at all. It is second-order, relational, extrinsic to the mental state—as existence is second-order, relational, and extrinsic to the object. Existence is not an intrinsic property of things that have it but consists in a relation between a propositional function and its instances; it is not in the object. Similarly, consciousness is not in the mental state, not constitutive of it, not essential to it. Consciousness arises when, and only when, the subject has a belief about his mental state: for you to consciously think that it’s raining is to believe that you think it’s raining. It is a higher-order property of a mental state (if a property at all). Thus, there is such a thing as consciousness, but it is not what we supposed—a feature of a mental state. There is no such thing as that, so the skeptic is right up to a point; but talk of consciousness can be saved by re-interpreting it as short for the existence of a higher-order thought or belief. This is a skeptical solution because it abandons the naïve assumption that consciousness is a first-order intrinsic property of a conscious state. There is nothing funny going on—no remarkable property that conscious beings have and non-conscious beings lack. We don’t have to postulate wacky witchy facts to accommodate consciousness, as we don’t have to postulate wacky witchy facts to explain existence. Both consist in mundane facts concerning higher-order thoughts and instantiated propositional functions, respectively. Nor do we have to postulate a Meinongian ontology of consciousness, as if consciousness consisted in a special class of immaterial Cartesian objects (golden clouds perhaps) tenuously connected to the real world. We could even be physicalists about existence and consciousness if we could see our way clear to accepting that higher-order beliefs are physical (brain states) and propositional functions are also physical (bits of spoken language).

I have not tried to evaluate these skeptical solutions; I have merely articulated a pattern common to both. Now I want to urge that both share common defects, not in order to refute them, but to underline the commonality. In the case of existence, we have the point that the concept of an instance already embeds the concept of existence, since these have to be existent instances.[2] Also, we have the questions of what a propositional function is and what the relation of instantiation is: is there a straight solution to these questions or must they too receive a skeptical solution? Similarly, in the case of consciousness, we have the question of circularity: doesn’t the concept of a higher-order consciousness-conferring thought presuppose that that thought is conscious? If not, how does it confer consciousness on the first-order thought? Also, what is the relation of aboutness that is being invoked when we say that one thought is about another—can this problem be given a straight solution or must it be a given a skeptical solution? Isn’t this all an elaborate charade to avoid facing difficult problems? In any case, existence and consciousness present similar dialectical patterns. They also illustrate the reach of the argumentative structure initiated by Hume (and extended by Kripke).[3]

[1] There is now a large literature on this, going back to Kripke’s Wittgenstein on Rules and Private Language. I presuppose this literature and undertake to expand the key concepts.

[2] See my Logical Properties, chapter 2.

[3] Note that skeptical solutions don’t have to include the idea of replacing truth conditions by assertion conditions or bringing in community practices; they can take other forms too.