Knowing That I Think

 

 

What is it that I know when I know that I think? One view is that I know of myself that I have the attribute of thinking: I recognize that I (a self) instantiate a certain property. This requires that I must know of my existence in order to know that I think. The knowledge that I think therefore presupposes knowledge of the proposition that I exist, which precludes it from serving as a premise in the Cogito, on pain of circularity. It also raises the question of how I know that I exist, if not by knowing that I think: is this existential knowledge somehow primitive and independent of my knowledge of my mental states? Do I just perceive my existence directly? Yet (as Hume pointed out) I don’t seem to have any impression of myself as such. I seem to be referring to myself and predicating a property of that self, but my knowledge of the referent is puzzling. Do I really know the truth of a subject-predicate proposition in which the subject refers to myself and the predicate ascribes a property to that self?

            A famous response to the Cogito insists that I am only entitled to claim knowledge of the proposition that there are thoughts going on, from which it doesn’t follow that I exist as an entity over and above my thoughts. But we need to be more specific than that if we are to capture the import of the proposition in question: it is thinking in me that I know to exist, not thinking happening anywhere to anybody. But then we are back with the self and the problems that arise with that. You might think we could fix the problem by invoking the indexical expressions “here” and “now”: what I know is that there is thinking going on here and now. But again this doesn’t give us any distinction between my having thoughts and someone else having thoughts. How do we tie the reference down without invoking the self as thinking thing? I suggest we appeal to the demonstrative “this consciousness”: what I know is that there is thinking going on in this consciousness. When I say that I know that I think I am saying that I know there is thought occurring in this consciousness, where the demonstrative refers to a certain consciousness (mine). The consciousness in question may be quite brief—perhaps the duration of the thought—and it may also contain other mental states, such as perceptions and emotions. What is important is that it is not identical to the self: I have a state of consciousness but I am not that state of consciousness. I may persist for longer than it does, and I am not a state but a thing—the bearer of a conscious state. The concept expressed by “this consciousness” is far less committal than the concept expressed by “I”: it is simply that which currently exists as a particular state of consciousness. I can refer to this overall state, and when I attribute thinking to myself that is what I do. The proposition that I know is thus equivalent to the proposition that this state of consciousness includes thinking—which makes no reference to a self at all. There is thus something peculiar and misleading about the use of “I” in expressing what I know: the self is not really being referred to at all–the word “I” is not occurring as a “referring expression”. The sentence paraphrases out into a sentence about a state of consciousness.

            This analysis has a desirable result from the point of view of the Cogito: it allows it to escape triviality. For now we don’t have to accept that the premise presupposes the conclusion by referring to the very thing the Cogitoseeks to establish. The Cogito should read as follows: “There is thinking going on in this consciousness, therefore I exist”. Here the premise does not contain the conclusion; it is a substantive step from the presence of thinking in a particular (state of) consciousness to the conclusion that there is a thinking self. The self (Descartes’ thinking thing) is what has states of consciousness—or what is conscious; it is not consciousness itself. (The Cogito does not read, “I think, therefore my consciousness exists”.) In order to know the premise of the Cogito we have to know that two things exist—thoughts and the consciousness in which they occur. But neither of these logically presupposes the existence of the self. The self comes in via the principle that consciousness requires a bearer or a subject or an underlying substance (as it might be, the brain or an immaterial entity). Of course, that principle can be questioned (and historically has been), but at least the Cogito cannot be convicted of roundly begging the question, under this interpretation of it. And that seems intuitively right, because we don’t react to the Cogito by immediately complaining that the premise presupposes the conclusion; we naturally interpret the premise in such a way that the conclusion is a substantive extra step. All it is really saying is that a certain consciousness (this one) contains a certain attribute, with no reference made to a self as such. The Cogito may be a non sequitur, but it is not a tautology.

            The expression “this consciousness” is a very peculiar demonstrative, which only comes into use in attributions of self-knowledge; it has little to no public use. We use the word “I” when we want to talk about ourselves in the presence of others—referring to a certain person—but the words “this consciousness” refer to something invisible to others and not of their concern. Each person uses it to refer to his or her own private inner awareness as it evolves through time; it is the hook on which to hang self-attributions of mental states. When we think that we think we think something of the form, “There is thinking going on in this consciousness”—which we logically could do without ever using the word “I”. We refer thereby to an immediately given reality: our total phenomenological state at a given time. No doubt this entity is obscure and elusive (we are talking about consciousness after all), but it is what encompasses or grounds our more specific mental states. I would not wish to say that it is simply the set of all my current mental states (the “bundle theory” of consciousness), preferring to view it as a unitary mental reality—a kind of field of awareness. But we needn’t enter into the metaphysics of consciousness in order to recognize that the demonstrative “this consciousness” has a referent and that this referent plays a role in self-knowledge. Whenever I know something mental about myself I tacitly employ the demonstrative “this consciousness” in my thought. The logical form of “I think”, as it is used in the Cogito, is an existential quantification over thoughts combined with a locative designation of a specific field of consciousness: “There are thoughts located in this field of consciousness” (similarly for doubts, pains, etc). Thus I am not logically required to know of the existence of the self in order to know that I think, only of the existence of thoughts and consciousness. So we don’t need to postulate some kind of direct non-inferential knowledge of the self in order to explain my knowledge that I think.

 

                                    Is Language Necessarily Fregean?

 

 

 

Frege’s distinction between sense and reference gains much of its intuitive plausibility from certain often-rehearsed examples. Thus we have the cases of Hesperus and Phosphorus (the planet), Afla and Ateb (the mountain), Superman and Clark Kent (the person). Citing this type of example Frege invokes the notion of a “mode of presentation”, inviting us to consider the fact that a single object can be presented to us in many different ways; these ways are said to correspond to the sense of a name, so that two names of a single object can have different senses. The intuitive point is that we can have many visual perspectives on the same object as we view it from different angles and in different conditions. And it is not always obvious that it is the same object we are seeing from these different viewpoints. Thus senses are held to be constituted by (or supervenient on) perceptual aspects—in these cases visual aspects. This sounds all very straightforward and plausible, and so the theory of sense is launched on apparently solid foundations. Once it is extended beyond these initial examples to include sense and reference for predicates, connectives, and whole sentences the theory tends to lose its intuitive appeal; but in the case of names of perceptible material objects it looks to be on firm ground.

            This raises the question of whether the Fregean picture can be applied to any language. Are there types of language to which it does not apply? Is it only proper names for perceptible material objects that exemplify the theory? That would be a serious limitation. In particular, do all singular terms carry with them a mode of presentation that fixes their sense? The obvious problem concerns language about mental states—about sensations, thoughts, “I”, etc. Suppose I refer to a pain in my knee: does this pain offer me a variety of perceptual perspectives? Clearly, it does not—I do not see, hear, touch, smell, or taste my pains. Can we introduce two different names for the pain that are associated with two modes of presentation, analogous to visual perspectives on a material object? We certainly have no such names as things stand, and it is hard to see what the distinct modes of presentation might be. I don’t sense my pains from different perspectives. Russell would say that it is possible for me to have logically proper names for my sense data, where the meaning of the name is identified with the reference; but the idea that my sense data can be presented to me in different ways, according to my viewpoint, seems manifestly absurd. I don’t have perceptual relations to my inner states. Maybe we can try to cobble together some analogue of visual mode of presentation here—say by bringing in third-person reference to my inner states—but then we are trying to save a theory, not motivate it. The beauty of the usual examples is that they appear to give the theory a strong intuitive foundation (which then gets stretched to breaking point as it is generalized), but we cannot obtain this result by invoking mental language—here the idea of two modes of presentation of a single mental state looks contrived at best. I couldn’t perceptually encounter a single pain of mine in two different ways and then be surprised to find that I have encountered the same sensation twice. So mental language looks distinctly non-Fregean. Just as it is hard to extend the Fregean apparatus beyond the initial examples to include predicates and whole sentences, so it is hard to extend it to words for mental particulars (including the self). So it would seem that only a fragment of natural languages fits the Fregean model—names of people, places, and things. We might then speak of a Fregean theory of names, where this linguistic category is quite narrowly circumscribed, with no attempt to be more general. At least the theory is correct for the initial examples—or so it might be thought.

            But is it? Note the heavy emphasis on visual modes of presentation in describing the motivating examples: for example, the way Venus looks in the morning and the evening. Here we have a clear conception of two appearances of a single entity—the way a planet can appear differently in the visual field. There can be variations of brightness, apparent size, position, relation to other celestial bodies, color of sky, etc. We also have a clear conception of tracking a single object visually through time and space. Vision supplies us with well-delineated perceptual aspects, both distinct from each other and yet clearly of the same object. It is ideally suited to getting Frege’s theory of sense and reference off the ground. But what about blind people: can’t they use and understand ordinary proper names? Evidently they can, and yet they associate no visual mode of presentation with those names. If Frege tried to motivate his theory by reminding blind people of how they see objects, he wouldn’t get very far. So is his theory limited to proper names as used by sighted people? That would surely be a grave limitation.

            It might be suggested that blind speakers can exploit their sense of hearing to arrive at suitable modes of presentation. Thus they might detect planets by the use of an instrument that converts information contained in light into audible clicks: they apply the names “Hesperus” and “Phosphorus” on the basis of this instrument (there could be blind astronomers). They might then discover that what produces one pattern of clicks is the very object that produces another pattern. The sense-determining mode of presentation is identical to the pattern of clicks, it may be said. Accordingly, what these speakers mean by “Hesperus” and “Phosphorus” is not what sighted people mean, since they associate different sensory modes of presentation with the names in question.

It is certainly possible to construct Fregean modes of presentation this way, but it is striking how unintuitive the resulting theory is—patterns of clicks caused by an object don’t seem much like visual aspects of the object. The clicks are evidence for the planet in its two different appearances, but they hardly constitute aspects of the planet. We see objects, but we hear only the sounds that objects make, not objects themselves. Doesn’t it seem peculiar to claim that the clicks constitute what the blind speakers mean by their names? “What do you mean by ‘Hesperus’?” we ask; “I mean the following sequence of clicks…” they answer. Is this what the blind speakers expresswhen they utter the name? It is the same with names for people: a blind speaker might hear the same person talking on different occasions and not realize it is the same person, but is it plausible to suggest that (say) “Superman” and “Clark Kent” mean these different vocal sounds? Are these different modes of presentation of the same man, suitable for conversion into senses? And what if a speaker is both blind and deaf—are we then to suppose that his modes of presentation are tactual or olfactory? And what if he has no senses at all? Must his names then be meaningless, devoid of sense?

            Someone wedded to the Fregean scheme might insist that this is all really perfectly intuitive—the use of visual aspects was not essential to getting the theory off the ground. It is equally plausible when we shift to other senses. Then consider a further sensory deprivation: the speakers suffer from blindsight and analogues of blindsight for the other senses. Can’t they still have names for material objects, even though they experience no conscious modes of visual (or other) presentation for those objects? They can detect the presence of objects by use of their eyes, but there is no conscious visual seeming, just a sensory blank. How then could the sense of the names be derived from their visual (or other) experience? It might be replied that they harbor unconscious visual modes of presentation, and that these constitute the senses of the names. But how can a speaker’s consciousunderstanding of a name be constituted by an unconscious state of his mind or nervous system? The unconscious information delivered by the eyes in cases of blindsight cannot be what the speaker consciously means by her words; it surely plays no part in forming the speaker’s linguistic understanding. It is neither grasped nor expressed, existing merely at an unconscious level. There is absolutely nothing intuitive about the claim that the meaning of names for speakers with blindsight is made up of unconscious information buried deep in their brains. And the same is true for “deafhearing” and the like: unconscious auditory data, pertaining to clicks or voices, is not going to cut it as a candidate for modes of presentation. And yet meaning goes on in such cases. Perceptual aspects of things are simply not necessary to the use and understanding of names, even when those names denote ordinary material objects. They seem, in fact, quite irrelevant. It is perfectly true that we, as we are constituted, experience perceptual aspects of things, notably visual aspects (if we are sighted); but this is not a requirement of meaning—it is not essential to language as such, not even to naming.

            Here are a couple of supplementary points. Your visual modes of presentation of a given object, say a person you are talking to, will vary over time, as you move, your interlocutor moves, the light varies, and so on. These will probably number in the thousands even within a short period of time, given the sensitivity of vision. Are we to suppose that the name you use for your interlocutor varies in its sense as the mode of visual presentation varies? That would mean that the name has thousands of different senses as the conversation continues. But doesn’t it have the same sense over the entire period? Is it really as fluid and ever changing as your visual experience? Secondly, animals and preverbal children see the world under visual aspects—they experience visual modes of presentation—and yet they do not understand language. Clearly vision is not ipso facto a linguistic matter—no sense is grasped just by seeing an object from a particular angle. So how can senses be identified with perceptual aspects? At the least something would need to be added—but then that will be what constitutes meaning, not the perceptual aspect itself. Sensory modes of presentation are at too low a cognitive level to constitute linguistic sense.

            At this point it might be conceded that the motivating examples are flawed, mistaking the parochial for the general, but there is still the question of solving Frege’s puzzle. If we reject modes of presentation, aren’t we stuck with reference alone, and hence will be unable to distinguish “a = a” and “a = b”? At least we can understand how identity statements might be informative if we invoke the apparatus of perceptual aspects, because it is clear that we can discover that two visual aspects belong to the same object. But it doesn’t follow that this kind of identity knowledge is what is expressed by identity statements; it might merely be collateral knowledge, not part of the semantic content of the statement in question. Here we need to state an alternative theory, so that we can see that Fregean modes of presentation are not necessary for solving Frege’s puzzle. Consider this theory: “a = b” means “the denotation of ‘a’ = the denotation of ‘b’”. Here we get a difference of meaning because each definite description contains a reference to a distinct name. This is not a Fregean theory (in fact Frege explicitly rejects such a theory) because, despite the use of definite descriptions, it is not a theory that invokes modes of presentation of objects—Venus does not present itself to us as the denotation of “Hesperus”. This is a meta-linguistic theory, not a mode of perceptual presentation theory. I won’t defend the theory here, merely noting that it provides an alternative to Frege’s official story, as well as giving us an invariant meaning for names used by different speakers (sighted and blind, people with blindsight, and across varying perceptual encounters). Modes of perceptual presentation are thus not part of the meaning of sentences containing names, even names of people we see every day; and the proposition expressed by an identity statement is not identical to a proposition concerning perceptual aspects coinciding in a single object. I can indeed learn that two perceptual appearances are appearances of the same planet, but that is not what I learn when I learn that Hesperus is Phosphorus—instead I learn that the planet called “Hesperus” is identical to the planet called “Phosphorus”. This is the self-same proposition that is learned by the sighted, the blind, people with blindsight, and anyone else who understands the name; it is not that these varying individuals all learn something different, depending upon their contingent perceptual peculiarities.   

            Frege came up with a novel theory of meaning, centered on the distinction between sense and reference, which has had a profound influence. When students are first presented with the theory they are schooled to reflect upon Hesperus and Phosphorus and the like; they are enticed to consider the different modes of presentation we can have of those objects. This all seems very intuitive and commonsensical; they are therefore easily persuaded of the fundamental correctness of Frege’s theory. To be sure, doubts set in when Frege attempts to generalize the theory, and few have followed him the whole way. Still, it is felt that he was definitely onto something with ordinary proper names: sense, reference, mode of presentation, cognitive value—they all make perfect sense. But if what I have argued in this paper is right, that is all an illusion, borne of tendentious examples. Frege’s theory is defective from the ground up—even for the cases thought most favorable to it. Basically, it is too narrow, because too geared to visual perception.  [1] Not only is language not necessarily Fregean, as witness mental language; no fragment of actual language is Fregean either, not even “Hesperus” and “Phosphorus”.

 

  [1] The vision-centered character of Frege’s thinking about sense and reference is evident in his well-known analogy with the telescope. He compares the subjective idea to the retinal image, the reference to the distantly seen moon, and the sense to the optical image on the lens. A telescope is obviously a visual instrument. Thus the entire scheme is shaped by the existence and nature of vision. But there is no necessary link between vision and meaning, as blind speakers make abundantly clear. If we abandon the visual examples and analogies, simply equating the sense of a name with a definite description that is not tied to any sensory system, we sacrifice a great deal of the intuitive force of theory: no longer can we speak of modes of presentation, but also we are simply claiming synonymy between names and descriptions. We don’t have a theory of what the differences of sense between descriptions consist in, as we do when we invoke modes of perceptual presentation. At the very least a theory with the structure of Frege’s theory needs to be expounded and motivated without reliance on perceptual perspectives. Sense can’t be identified with “mode of presentation” but only “mode of description”. He is then merely proposing a description theory of names, not a general theory of meaning rooted in our basic awareness of the world. A sense might be construed as a mode of conception but not as a mode of presentation—and we can’t elucidate the former by means of the latter. We can certainly set up a structure of the kind Frege recommended, but we can’t motivate it in the way he did; we can’t make it look self-evident by appeal to perceptual examples. 

 

 

Intentionality and the Inner

 

 

The mind is directed outward onto something other than itself. This intentional object might be a particular or a universal, mental or physical, existent or non-existent. It is not identical to the mental state that takes it as object. Moreover, the outer object is essential to the mental state that is directed towards it, constitutive of that state. The object can generally be apprehended and known by others—say, a color or shape or object in space. It is not in itself private or hidden. The mind is characterized by outer objects: “externalism” is true. But the mind is also said to be “inner”: not a public thing, but a private and unobservable thing. An experience of color, say, is something inner—invisible, unlike color itself. These two properties of the mind seem to be in tension, as if one of them has to go: for how can what is constituted by something outer also be essentially inner? How can one and the same thing be both private and public—both “in here” and “out there”? How can the mind be internal and external at the same time?

            But there is no irresoluble contradiction here; though there is rhetorical dissonance, which perhaps accounts for the lack of emphasis on the inner nature of mind from philosophers impressed with Brentano’s thesis. We just need to distinguish object from act: the act is inner and the object is outer. An inner mental act has an outer non-mental object. Another person can be in the dark about one’s inner states even though they have public objects, since it may not be determinable which object the mental state concerns. I can fail to know what color you are experiencing even though I am acquainted with all the colors: I don’t know which color is the intentional object of your inner experience–it might be red or it might be green (as in an inverted spectrum case). Perhaps I can’t know your inner state, though I am perfectly familiar with the objects that define it, viz. colors. The mind is both inner and outer, with no contradiction between these two properties. Externalism does not imply third-person knowledge, still less that the mental is not inner.

            Indeed, it would seem that being inner is a necessary condition for also being outer, since the mind is necessarily inner—so it cannot be directed outwards without itself being something inner. Intentionality requires innerness. There can be no such thing as an outer mental state that has an outer object. A perceptual experience could not be an outer thing as a matter of metaphysical necessity; so an experience of color cannot be public—though colors are. Expressions of the mind can be public (behavior) but not states of mind themselves. It is not clear exactly why this is so: why should (mental) intentionality be possible only for states that are inner? There seems to be no explanatory or conceptual link here: why does directedness to the outer depend on what is directed being inner? Why do we apprehend the public world from a place of privacy? But it appears to be so, since there are no counterexamples. The converse is not so obvious: not all inner states have intentional objects—such as bodily sensations like pain (though this has been disputed). So being inner does not entail having an outer object. But having an outer object does entail being inner: no intentionality without innerness.  [1]

            In so far as acceptance of intentionality leads to reluctance to acknowledge the innerness of mind, the reluctance is misplaced: the mind can be both inherently inner and also necessarily directed outwards. Note that this notion of innerness is not the same as the notion of the internal–what lies spatially within the skin or head. That notion is neither necessary nor sufficient for innerness. The mind cannot be both internal and external in thissense of “internal”: but it can be both inner and outer in the sense intended above. That indeed is definitive of the mind: the mind is that which is inner and yet directed outwards—inner act, outer object.  [2]

 

Colin McGinn

           

 

 

 

  [1] I am speaking of mental intentionality here, not linguistic intentionality (reference): the latter is public, unlike the former. I hold (with many others) that linguistic intentionality is dependent on mental intentionality, i.e. outer signs depend on inner mental states. Thus my claim is that all original intentionality is inner.

  [2] This is why externalism about mental content does not solve the problem of other minds: experiences don’t cease to be inner events just because they are of something external.

 

Inscrutable Mental Causation

 

 

It is a matter of common observation that the sun causes plants to grow. Our ancestors will have noticed this long ago and wondered why it is. The answer is by no means obvious. Maybe it’s because the sun is a god and he commands plants to grow; maybe the sun god sends out waves of love that are absorbed by plants; maybe it has something to do with the shape of the sun or its motion or its size or its color; or maybe the sun is a distant fiery object that emits radiation and this interacts chemically with tiny receptors on the surface of plants. We now know that it is the last of these: it is in virtue of the process of photosynthesis that plants grow by absorbing sunlight. This process is well understood down to the fine details: light energy gets converted to chemical energy which is stored in carbohydrate molecules synthesized from carbon dioxide and water and involving electron stripping and oxygen release. But it is conceivable that our ancestors could have got stuck at the macro level of the sun causing plant growth, never achieving the kind of micro understanding now taught in biochemistry classes. If so, they would never have understood the nature of the causation involved; they would know there is a causal connection but not what it depends upon. For that they would need to provide an analysis of light and an analysis of plants that revealed their intrinsic nature. This is a highly non-trivial enhancement of the cognitive abilities present in simply observing a causal connection. A creature might be blocked from achieving that kind of causal understanding by innate intellectual limitations.

            Now think about mental causation. This is a vexing and obscure subject. We observe certain broad macro causal relations: desires cause decisions, perceptions cause knowledge, reflection causes changes of heart. We also observe instances of such general causal relations: my desire to have a drink caused my decision to go into a bar, my seeing my cat at noon caused me to believe that my cat was home at noon, my reflection about a friend’s actions caused me to drop that friend. Not everyone agrees that there are real causal relations within the mind—partly because they cannot see what kind of causation might be involved—but it is certainly a natural way to talk about the mind and its states. We can then ask how this causation works: in virtue of what do mental states cause other mental states (or actions)? Immediately we are brought up short: we don’t have a story comparable to the story about sunlight and plant growth. We don’t even have a rough outline of such a story. We can’t say what it is about desire that allows it to cause decision, or what it is about seeing that brings about knowledge, or how reflection causes changes of mind. We don’t have a story about the properties of the cause that lead to its effects, comparable to the photon theory of light and the concept of energy transformation. We don’t have a causal theory to go with our causal observations—we have no analogue of electron stripping and oxygen release. We have no understanding of the mechanisms of mental causation.

            It might be suggested that there is such machinery—processes in the brain: so we do have a story about mental causation. It may not be as well worked out as the theory of photosynthesis, but it plays the same logical role: it tells us in virtue of what the macro level causal relations obtain. Brain science tells us the nature of mental causation as biochemistry tells us the nature of botanical causation. But this line of thought faces formidable difficulties, well known and extensively debated. The problem, in short, is that it leads either to epiphenomenalism or to materialism. It leads to epiphenomenalism because it threatens to make mental properties causally redundant, and it leads to materialism because materialism is the only way to avoid epiphenomenalism. I won’t rehearse this familiar story: intuitively, the problem is that causation in the brain is at the wrong level to account for mental causation—it is too extrinsic, too foreign. This is why materialism starts to seem unavoidable: it is the only way to ensure that mental causation is possible. A materialist theory of mental causation leads to materialism about the nature of the mental. If neurons play the role of energy-converting cells in photosynthesis, then neurons must be what the mind is made of. But that kind of reductionism has its own difficulties, also well known. Materialism about light and plants is not a problem, but materialism about the mind is a high price to pay for the causal efficacy of the mental.

            Why not just locate the causal machinery in the mental states themselves? It is in virtue of being a desire that desires cause decisions; it is in virtue of being a desire for a drink that this desire causes me to decide to go to a suitable bar. The causally operative properties are just mental properties themselves, conceived as existing separately from underlying brain states. This is the analogue of asserting that it is in virtue of being light that plants grow in sunlight—with nothing further to be said by way of causal elucidation. That is, mental causation is primitive. But this position is hard to accept because mental states are not basic properties of the universe; they are less primitive than biochemical properties of plants. Is there really nothing more to causation in the mind than our ordinary catalogue of mental properties? Also, how exactly does this irreducibly mental causation work? How does simply being a desire bring it about that a decision is made? We need some sort of explanation, some account of the causal connection. Simply being light doesn’t explain how light produces plant growth, so why should simply being a desire explain how desires have the effects they do?

            We now have three possible theories of mental causation: first, it doesn’t exist at all; second, it reduces to physical causation in the brain; third, it is primitive and inexplicable. For completeness I will add the “theory” that mental causation is magical and requires divine intervention. None of these theories is attractive; in fact, they are all pretty horrible. They seem more like symptoms of panic than sober theorizing; none of them could claim the mantle of obvious truth. They are what you come up with when you can’t think of anything better. Are we then clean out of options? Must we accept defeat? Consider again our causally curious ancestors puzzling over the sun and plant growth: I remarked that they might never have been able to hit on the theory of photosynthesis; they might, indeed, never have progressed beyond the sun god theory. The problem was not elementary and required sophisticated physical concepts—atomic physics no less. Might we be in a similar position with respect to mental causation? Is it that we can’t construct or discover the properties and principles that actually lie behind mental causation? There is something about desires that enables them to cause decisions, but we don’t know what it is, not even sketchily. It isn’t brain correlates and it isn’t just the mental properties themselves (as we conceive them); it’s something else that eludes our understanding. It might not always do so—we might with diligence produce a good causal story: but we also might not. At any rate, nothing we now know about the mind provides us with anything comparable to the theory of photosynthesis. We don’t even grasp the psychological analogue of physical analysis, i.e. investigating things in their fine structure to tease out how they work. We don’t have an analogue for the mind of the macro-micro distinction: we don’t have a notion of micro causation to go with our notion of macro causation. That is, we can’t analyze the mind into simpler parts that operate together to produce observable results—a kind of atomic psychology. Of course, the mind is composed of many elements (faculties, states, modules, concepts), but we have nothing analogous to the atomic understanding of matter, or indeed basic chemistry. This is why we are at a loss to provide an account of mental causation: we don’t know all there is to know about the mind; therefore we have a poor understanding of mental causation. We don’t know how one mental state leads to another. We are like people that observe the sun’s effects on plants but don’t have a clue how to explain it. There is machinery in there somewhere but it is inscrutable to us.

            This position implies that there is more to the mind than what we know about, including the neural basis of mind. There is a whole level of description of familiar mental phenomena that eludes us, at present or permanently. This is because we lack any plausible account of mental causation—we don’t know in virtue of what properties and principles this kind of causation works. Mental causation is an enigma to us. The two suggestions we have considered—brain causation and primitive mental causation—are both unsatisfactory, but nothing else suggests itself. The nature of mental causation is not visible from the perspective of the brain or from the perspective of introspection. We must be missing something, something vital. No doubt what we are missing is intimately connected to properties of the brain and to mental properties themselves, because these are not irrelevant to mental causation; but neither identifies the level at which mental causation is to be explained. That is, neither corresponds to the level occupied by the theory of photosynthesis. The level of description provided by a computational theory of mental processes is the kind of level we need, but I don’t believe that computation can play the same kind of explanatory role as photons, energy transfer, and carbon synthesis. It is not in virtue of computations that one mental state causes another, though computations may well be involved in many mental processes. Computations are too abstract to be causes; they are not energy mechanisms.  [1] We would need to connect them to neural processes, but then we are back with the brain theory of mental causation. Rather, there are unknown causal factors at work in mental causation.

            These are no doubt difficult issues, in which obscurities combine and deepen, and ignorance seems the most likely hypothesis. The mind-body problem and the problem of causation: put them both together and the going is bound to get tough. I suspect we know almost nothing about the real nature of mental causation, despite knowing many causal truths about the mind. We notice it in ourselves, but we have no deeper understanding of what is involved.

 

  [1] How does one computation lead to another? Is it by neural realization or is it a primitive property of computations? This is the same dilemma as before—not surprisingly since computations are modeled on actual mental processes such as conscious calculation.