We now skip forward in time to the twentieth and twenty-first centuries. Because of the failures of Cartesian-style dualism, especially the failure to get an adequate or even coherent account of the relationship between the mind and the body, it is widely assumed that substance dualism in any form is out of the question. This is not to say that no serious professionals are substance dualists. But in my experience most substance dualists I know are people who hold this view for some religious reasons, or as part of a religious faith. It is a consequence of substance dualism that when our body is destroyed our soul can continue to survive; and this makes the view appealing to adherents of religions that believe in an afterlife. But among most of the professionals in the field, substance dualism is not regarded as a serious possibility. A prominent exception is the defense of dualism offered by Karl Popper and J. C. Eccles.1 They claim that there are two quite distinct worlds, World 1 of physical objects and states and World 2 of states of consciousness. Each is a separate and distinct world that interacts with the other. Actually they go Descartes one better and also postulate World 3, a world of “culture in all its manifestations.”2
All forms of substance dualism inherit Descartes’ problem of how to give a coherent account of the causal relations between the soul and the body, but recent versions have an additional problem. It seems impossible to make substance dualism consistent with modern physics. Physics says that the amount of matter/energy in the universe is constant; but substance dualism seems to imply that there is another kind of energy, mental energy or spiritual energy, that is not fixed by physics. So if substance dualism is true then it seems that one of the most fundamental laws of physics, the law of conservation, must be false. Some substance dualists have attempted to cope with this problem by claiming that for each infusion of spiritual energy, there is a diminution of physical energy, thus preserving a constant amount of energy in the universe. Others have said that the mind rearranges the distribution of energy in the universe without adding to it or subtracting from it. Eccles says that the mind can affect the body by altering the probability of neuronal events without any energy input, and that quantum physics enables us to see how this can be done: “The hypothesis of mind-brain interaction is that mental events act by a quantal probability field to alter the probability of emission of vesicles from presynaptic vesicular grids.”3 There is something ad hoc about these maneuvers, in the sense that the authors are convinced in advance of the truth of dualism and are trying to find some way, any way, that will make dualism consistent with physics.
It is important to understand what an extreme doctrine substance dualism is. According to substance dualism our brains and bodies are not really conscious. Your body is just an unconscious machine like your car or your television set. Your body is alive in the way that plants are alive, but there is no consciousness to your body. Rather, your conscious soul is somehow attached to your body and remains attached to it until your body dies, at which time your soul departs. You are identical with your soul and only incidentally and temporarily inhabit this body.
The problem with this view is that, given what we know about how the world works, it is hard to take it seriously as a scientific hypothesis. We know that in humans consciousness cannot exist at all without certain sorts of physical processes going on in the brain. We might, in principle, be able to produce consciousness in some other physical substance, but right now we have no way of knowing how to do this. And the idea that consciousness might be produced apart from any physical substrate whatever, though conceivable, just seems out of the question as a scientific hypothesis.
It is not easy to make the idea that the mind is a separate substance consistent with the rest of what we know about the world. Here are three ways of trying to do it, each with a different conception of the mind.
First, divine intervention. Physical science is incomplete. Our souls are something in addition to the rest of the world. They are created by divine intervention and are not part of the physical world as described by science.
Second, quantum mechanics. The traditional mindbody problem arises only because of an obsolete Newtonian conception of the physical. On one interpretation of quantum measurement, consciousness is required to complete the collapse of the wave function and thus create quantum particles and events. So some form of consciousness is not created by the rest of nature, rather it is essential for the creation of nature in the first place. It is a primitive part of nature required to explain brain processes and everything else.4
Third, idealism. The universe is entirely mental. What we think of as the physical world is just one of the forms that the underlying mental reality takes.5
I mention these for the sake of completeness. I do not agree with any of them, and I don’t think I understand the second; but as none of them is an influential view in the philosophy of mind, and as I am trying to explain the philosophy of mind, I won’t discuss them further in this book.
There is a weaker version of dualism called “property dualism,” and that view is fairly widespread. The idea is this: Though there are not two kinds of substances in the world, there are two kinds of properties. Most properties, such as having an electrical charge, or having a certain mass, are physical properties; but some properties, such as feeling a pain or thinking about Kansas City, are mental properties. It is characteristic of human beings that though they are not composed of two different kinds of substances, their physical bodies, and in particular their brains, have not only physical properties, but mental properties as well.
Property dualism avoids postulating a separate mental substance, but it inherits some of the difficulties of substance dualism. What are the relationships between the mental and the physical supposed to be? How is it that physical events can ever cause mental properties? And there is a particular problem that property dualists seem to be beset with, and that is the problem of how the mental properties, granted that they exist, can ever function causally to produce anything. How can my conscious states, which on this view are not even parts of an extra substance, but merely nonphysical features of my brain, function to cause any physical events in the world? This difficulty, how mental states can ever function causally to produce physical effects, I described in chapter 1 as the problem of “epiphenomenalism.” According to epiphenomenalism, mental states do indeed exist but they are epiphenomena. They just go along for the ride; they do not actually have any causal effects. They are like the froth on the wave that comes up on the shore or the flashes of light that glisten off a lake—they are there all right, but they play no significant causal role in the physical world. Indeed, they are worse than the froth and the flash, because they could not play any causal role. The challenge is, How could they play any causal role in determining physical events when they are not themselves physical? If we assume, as it seems we must, that the physical universe is causally closed, in the sense that nothing outside it could have any effects inside; and if we assume, as it seems we must, that consciousness is not part of the physical universe, then it would seem to follow that consciousness can have no effect in the physical universe.
Property dualism does not force us to postulate the existence of a thing that is attached to the body but not really part of the body. But it still forces us to suppose that there are properties of the body, presumably properties of the brain, that are not ordinary physical properties like the rest of our biological makeup. And the problem with this is that we do not see how to fit an account of these properties into our overall conception of the universe and of how it works. We really do not get out of the postulation of mental entities by calling them properties. We are still postulating nonmaterial mental things. It does not matter whether we say that my conscious pain is a mental property of my brain or that it is an event in my brain. Either way, we are stuck with the traditional difficulties of dualism. One antidualist philosopher characterized these leftover mental phenomena as “nomological danglers” (“nomological” means lawlike). They are produced by the brain in a lawlike fashion, but then they do nothing. They just dangle there.6
Many, probably most, philosophers have abandoned dualism, but the situation is odd because to many dualists, the arguments I have just presented do not look at all decisive against all forms of dualism. I think a typical property dualist would say, “OK, the mind is not a separate substance but all the same it is just a brute fact of nature that creatures like us do have pains and tickles and itches, as well as thoughts and emotions and these are not in any ordinary sense physical. Nor are they reducible to anything physical.” And indeed some dualists bite the bullet and accept epiphenomenalism.
My guess is that dualism, in spite of being out of fashion, will not go away. Indeed in recent years dualism, at least property dualism, has been making something of a comeback, partly due to a renaissance of interest in consciousness. The insight that drives dualism is powerful. Here is the insight, at its most primitive: we all have real conscious experiences and we know that they are not the same sort of thing as the physical objects around us. This primitive insight can be given a more sophisticated formulation: the world consists almost entirely of physical particles and everything else is in some way an illusion (like colors and tastes) or a surface feature (like solidity and liquidity) that can be reduced to the behavior of the physical particles. At the level of molecular structure the table is not really solid. It is, as the physicist Eddington said, a cloud of molecules. It is just that from our point of view it seems solid. But at bottom the physical world consists entirely of microentities, the physical particles. However there is one exception. Consciousness is not just particles. In fact it is not particles at all. Whatever else it is, it is something “over and above” the particles. I believe this is the insight that drives contemporary property dualism.
David Chalmers7 puts the point by saying that it is not logically possible that the course of the physical universe should be different if the course of microphysical facts is the same. Once you have the microphysics then everything else follows. But that is not true for consciousness. You could imagine the whole physical course of the universe exactly the same, minus consciousness. It is logically possible that the course of the physical universe should be exactly as it is, but with no consciousness.
It is such apparent basic differences between the mental and the physical that drives dualism. I think dualism can be answered and refuted, but we do not yet have the tools to do it. I will do it in chapter 4.
The dualists said that there are two kinds of things or properties in the universe, and with the failure of dualism, it is natural to suppose that maybe there is only one kind of thing in the universe. Not surprisingly, this view is called “monism” and it comes in two flavors, mentalist monism and materialist monism. These are called “idealism” and “materialism,” respectively. Idealism says that the universe is entirely mental or spiritual; there exists nothing but “ideas” in the technical sense of the word, according to which any mental phenomenon at all is an idea. On some views—for example, Berkeley’s—in addition to ideas there are minds that contain the ideas. Idealism had a prodigious influence in philosophy, literally for centuries, but as far as I can tell it has been dead as a doornail among nearly all of the philosophers whose opinions I respect, for many decades, so I will not say much about it. Some of the most famous idealists were Berkeley, Hegel, Bradley, and Royce.
The single most influential family of views in the philosophy of mind throughout the twentieth century and leading into the twenty-first century is one version or another of materialism. Materialism is the view that the only reality that exists is material or physical reality, and consequently if mental states have a real existence, they must in some sense be reducible to, they must be nothing but, physical states of some kind. There is a sense in which materialism is the religion of our time, at least among most of the professional experts in the fields of philosophy, psychology, cognitive science, and other disciplines that study the mind. Like more traditional religions, it is accepted without question and it provides the framework within which other questions can be posed, addressed, and answered. The history of materialism is fascinating, because though the materialists are convinced, with a quasi-religious faith, that their view must be right, they never seem to be able to formulate a version of it that they are completely satisfied with and that can be generally accepted by other philosophers, even by other materialists. I think this is because they are constantly running up against the fact that the different versions of materialism seem to leave out some essential mental feature of the universe, which we know, independently of our philosophical commitments, to exist. The features they generally leave out are consciousness and intentionality. The problem is to give a completely satisfying materialist account of the mind that does not end up denying the obvious fact that we all intrinsically have conscious states and intentional states. In the next few pages I am going to sketch briefly the history of materialism in the twentieth century, up to the point where it finally reached its most sophisticated formulation in the computational theory of the mind, the theory that the brain is a computer and the mind is a computer program. This sketch is necessarily oversimplified. For reasons of space, I can only hit the high points, but I do want you to see those high points and how they relate to each other. There is a natural progression that leads from behaviorism to the computational theory of the mind and I want you to see that progression.
The earliest influential form of materialism in the twentieth century was called “behaviorism.” In its crudest version, behaviorism says the mind just is the behavior of the body. There is nothing over and above the behavior of the body that is constitutive of the mental. Behaviorism comes in two flavors, “methodological behaviorism” and “logical behaviorism.” I will consider each in turn.
Methodological behaviorism was a movement in psychology. It attempted to put psychology on a respectable scientific footing, along with other natural sciences, by insisting that psychology should study only objectively observable behavior. The “laws” that such a discipline was supposed to discover were laws that would correlate the input stimulus to the organism with the output response behavior; and for this reason, behaviorist psychology was sometimes called “stimulus-response” psychology. The behaviorists were so influential that for a time they even succeeded in changing the definition of psychology. Psychology was no longer the “science of the mind” but the “science of human behavior.” This view was called “methodological behaviorism” because it proposed a method in psychology rather than a substantive claim about the existence or nonexistence of the mind. The real objection to dualism, the methodological behaviorists claimed, was not that it postulates nonexistent entities, but rather that it is scientifically irrelevant. Scientific claims have to be objectively testable, and the only objectively testable claims about the human mind are claims about human behavior.
The big names in methodological behaviorism are John B. Watson (1878-1958) and B. F. Skinner (1904-1990). I think that, in fact, neither of them believed in the existence of any inner qualitative mental phenomena, but for the purposes of a scientific psychology, they only needed to insist on behaviorism as a method rather than as a specific ontological doctrine. It may be unfair to characterize Skinner as a methodological behaviorist, because in fact he objected to something he called “methodological behaviorism.” He thought of himself as a “radical behaviorist.” Nonetheless, his influences have been mostly methodological; so, I am going to follow the standard textbook account and characterize him as a methodological behaviorist. The only observable psychological phenomena are human behavior, so the right method for psychology has to be the study of human behavior and not the study of any mysterious inner, spiritual, mental entities. Methodological behaviorism was thus a research project in psychology and was surprisingly influential for decades.
Logical behaviorism was primarily a movement in philosophy, and it made a much stronger claim than methodological behaviorism. The methodological behaviorists said that Cartesian dualism was scientifically irrelevant, but the logical behaviorists said that Descartes was wrong as a matter of logic.8 A statement about a person’s mental state, such as the statement that a person believes that it is going to rain or is feeling a pain in his elbow just means the same as, it can be translated into, a set of statements about that person’s actual and possible behavior. It need not be translatable into statements about presently existing behavior, for a person might have a pain or a belief that he was not then and there manifesting in behavior, but then the statement has to be translatable into a set of hypothetical statements about behavior, what the agent would do or would say under such and such circumstances.
According to a typical behaviorist analysis, to say that Jones believes it is going to rain just means the same as saying an indefinite number of statements such as the following: if the windows in Jones’s house are open, he will close them; if the garden tools are left outside, he will put them indoors; if he goes for a walk he will carry an umbrella or wear a raincoat or both; and so forth. The idea was that having a mental state was just being disposed to certain sorts of behavior; and the notion of a disposition was to be analyzed in terms of hypothetical statements, statements of the form “If p then q.” As applied to the problem of mental states, these statements would take the form, “If such-and-such conditions obtain, then such-and-such behavior will ensue.”
By the middle decades of the twentieth century, the difficulties of behaviorism had led to its general weakening and eventual rejection. It was going nowhere as a methodological project in psychology, and indeed was under quite effective attack, especially from the linguist Noam Chomsky. Chomsky claimed that the idea that when we study psychology we are studying behavior is as unintelligent as the idea that when we study physics we are studying meter readings. Of course we use behavior as evidence in psychology, just as we use meter readings as evidence in physics, but it is a mistake to confuse the evidence that we have about a subject matter for the subject matter itself. The subject matter of psychology is the human mind, and human behavior is evidence for the existence and features of the mind, but is not itself the mind.
The difficulties with the logical behaviorists were even more marked. No one had ever given a remotely plausible account of how you could translate statements about minds into statements about behavior. There were various technical difficulties about how to specify the antecedents of the hypotheticals, and especially about how to do it without circularity. I said earlier that the behaviorists would analyze Jones’s belief that it is going to rain into sets of statements about his rain-avoidance behavior. But the difficulty with that is that we can only begin to make such a reduction on the assumption that Jones desires to stay dry. So the assumption that Jones will carry an umbrella if he believes that it is going to rain is only plausible if we suppose that Jones does not want to be rained on. But then if we are analyzing belief in terms of desire, it looks like there is a kind of circularity in the reduction. We did not really reduce the belief to behavior; we reduced it to behavior plus desire, which still leaves us with a mental state that needs to be analyzed. Analogous remarks could be made about the reduction of desire. To say that Jones’s desire to stay dry consists in such things as his disposition to carry an umbrella will only seem remotely plausible if we suppose that Jones believes it is going to rain.
A second family of difficulties had to do with the causal relations between mental states and behavior. The logical behaviorists had argued that mental states consisted in nothing but behavior and dispositions to behavior, but this runs against our common sense intuition that there is a causal relation between our inner mental states and our outward behavior. My pain causes me to cry out and to take aspirin; my belief that it is going to rain and my desire to stay dry cause me to take an umbrella, etc., and it seems that this apparent truth is denied by the behaviorists. They cannot account for the causal relations between the inner experience and the external behavior, because they are in effect denying that there is any internal experience in addition to the external behavior.
The real difficulty with behaviorism, though, is that its sheer implausibility became more and more embarrassing. We do have thoughts and feelings and pains and tickles and itches, but it does not seem reasonable to suppose that these are identical with our behavior or even with our dispositions to behavior. The feeling of pain is one thing, pain behavior is something else. Behaviorism is so intuitively implausible that unsympathetic commentators often made fun of it. As early as the 1920s, I. A. Richards pointed out that to be a behaviorist you have to “feign anesthesia.”9 And university lecturers have a stock repertoire of bad jokes about behaviorism. A typical joke: a behaviorist couple just after making love, he says to her “It was great for you. How was it for me?”
The sheer implausibility of behaviorism had become an embarrassment by the 1960s and it was gradually replaced among materialist-minded philosophers by a doctrine called “physicalism,” sometimes called the “identity theory.” The physicalists said that Descartes was not wrong, as the logical behaviorists had claimed, as a matter of logic, but just as a matter of fact. It might have turned out that we had souls in addition to bodies, but the way that nature in fact turned out, what we think of as minds are just brains, and what we think of as mental states, such as the feeling of pain or having a tickle or an itch, are just states of the brain—and perhaps the rest of the central nervous system. This thesis was sometimes called the “identity thesis” because it asserted an identity between mental states and brain states. The identity theorists were anxious to insist on the contrast between their view and behaviorism. Behaviorism was supposed to be a logical thesis about the definition of mental concepts. The identity thesis was supposed to be a factual claim, not about the analysis of mental concepts, but rather about the mode of existence of mental states. The model for the behaviorists was one of definitional identities. Pains are dispositions to behavior in a way that triangles are three-sided plane figures. In each case it is a matter of definition. The identity theorists said no, the model is not definitions, but rather empirical discoveries of identities in science. We have discovered, as a matter of fact, that a bolt of lightning is identical with an electrical discharge; we have discovered, as a matter of fact, that water is identical with H2O, and we are now discovering, and the discovery is proceeding daily, that mental states are really identical with brain states.10
There were a number of objections to the identity theory. I find it useful to distinguish between the technical objections and the common-sense objections. The first technical objection was that the theory seemed to violate a principle of logic called “Leibnitz’s Law.”11 The law says that if any two things are identical, then they must have all their properties in common. So if you could show that mental states had properties that could not be attributed to brain states, and brain states had properties that could not be attributed to mental states, it looks like you would refute the identity theory. And it did not seem difficult to provide such examples. So I can say, for example, that the brain state that corresponds to my thought that it is raining is 3 cm inside my left ear; but, according to the objectors, it does not make any sense to say that my thought that it is raining is 3 cm inside my left ear. Furthermore, even for conscious states that have a location, such as pain, the pain may be in my toe, but the brain state that corresponds to that pain is not in my toe, but in my brain. So the properties of the brain state are not the same as the properties of the mental state. Therefore, physicalism is false.
The identity theorists thought that they had an easy answer to these objections. The objections, they say, just rest on ignorance. When we come to know more about the brain, we will come to feel perfectly comfortable in attributing spatial location to mental states and attributing so-called mental properties to states of the brain. And, about the location of the pain in the toe, the identity theorists said that what we are interested in is not a putative object, the pain, but rather the total experience of having the pain. And that total experience extends all the way from the stimulation of the peripheral nerve endings in the toe to the brain itself. I think that the identity theorists were successful in answering this objection, but there were other objections that were more serious.
A common-sense objection to the identity theorists was that if the identity was indeed an empirical identity, something that could be discovered as a matter of fact, on the analogy with water and H2O, or lightning and electrical discharge, then it seems there would have to be two kinds of properties to nail down the two sides of the identity statement.12 Thus, just as the statement, “lightning is identical with an electrical discharge” has to identify one and the same thing in terms of its lightning properties and in terms of its electrical discharge properties; and “water is identical with H2O molecules” has to identify one and the same thing in terms of its water properties and in terms of its H2O properties, so the claim that, for example, “pain is identical with a certain type of brain state” has to identify one and the same thing in terms of its pain properties, and in terms of its brain-state properties. But if there are to be two independent sets of properties in the identity statement, then it looks like we have two different types of properties left over: mental properties and physical properties. It looks, in short, as if in order to make the identity thesis work, we have to fall back into property dualism. If all mental states are brain states, then there are two kinds of brain states, those that are mental and those that are not. What is the difference? The mental states have mental properties. The others have only physical properties. And that view sounds like property dualism.
This was a decisive problem for the identity theorists. The whole point of the theory was to vindicate materialism, to show that mental states were really identical with, were nothing but, were reducible to material states of the brain. But if it turns out that the brain states in question have irreducible mental properties then the project fails. It leaves us with an irreducible mental element. In doing research for this book I found at least one philosopher who thought of himself as an identity theorist who seemed willing to embrace this result at least as a possibility.13 Grover Maxwell calls his view the identity theory, but he says, “the way is entirely open for speculating that some brain events just are our joys, sorrow, pains, thoughts, etc., in all of their qualitative, and mentalistic richness” (p.235). This is quite similar to what I think is the correct view and will explain in chapter 4. But it was not a typical view among the identity theorists.
The standard identity theorists’ answer to this objection was less plausible than their answer to the Leibnitz Law objections.14 The answer they gave was that the phenomena in question could be specified without using any mental predicates. They could be specified in a topic-neutral vocabulary. Instead of saying, “There is a yellow-orange afterimage in me,” they prefer to say “There is something going on in me that is like what goes on when I see an orange.” Such a rephrasing of the identification of the mental states in a “topic-neutral” vocabulary was supposed to answer the objection because it enabled us to specify the mental element in a nonmental, neutral vocabulary: there is this thing going on in me and it can be specified in a way that is neutral between dualism and materialism, but it just turns out that the thing is a brain process. So we can specify the mental feature but in a way that is consistent with materialism.
I think this answer fails. The point that we can talk about mental phenomena without using a mental vocabulary does not change the fact that the mental phenomena continue to have mental properties. My yellow-orange after-image remains qualitative and subjective whether or not we choose to mention those features. If one wanted to refuse to talk about airplanes, one could just say, “some property belonging to United Airlines.” But that does not eliminate the existence of airplanes. To put the point succinctly, the fact that one can mention a phenomenon that is intrinsically qualitative and subjective in a vocabulary that does not reveal these features does not remove the features. In the end of course, the identity theorists wanted to deny that there were any such features, but that requires a separate argument. 15
One slightly more technical objection that really did concern the identity theorists and indeed eventually forced a modification in their views was the accusation of “neuronal chauvinism.”16 If the claim of the identity theorists was that every pain is identical with a certain kind of neuronal stimulation, and every belief is identical with a certain type of brain state, then it seems that a being that did not have neurons or that did not have the right kind of neurons could not have pains and beliefs. But why can’t animals that have brain structures different from ours have mental states? And indeed, why couldn’t we build a machine that did not have neurons at all, but also had mental states? This objection led to an important shift in the identity theory from what came to be called “type-type identity theory” to “token-token identity theory.” In order to explain this distinction I need to say a bit about the type-token distinction. If I write the word “dog” three times: “dog dog dog,” have I written one word or three? Well, I have written three instances, or tokens, of one type of word. So we need a distinction between types, which are abstract general entities, and tokens, which are concrete particular objects and events. A token of a type is a particular concrete exemplification of that abstract general type.
Using this distinction we can see how the identity theorists were motivated to move from a type-type identity theory to a token-token identity theory. The type-type identity theory says “Every type of mental state is identical with some type of physical state.” By their own lights this is a bit sloppy, because the identity in question is between actual concrete tokens and not abstract universal types. What they meant is: for every mental-state type there is some brain-state type such that every token of the mental type is a token of the brain type. The token identity theorists simply said: for every token of a certain type of mental state, there is some token of some type of physical state or other with which that mental state token is identical. They, in short, did not require, for example, that all token pains had to exemplify exactly the same type of brain state. They might be tokens of different types of brain states even though they were all tokens of the same mental type, pain. For that reason they were called “token-token” identity theorists as opposed to “type-type” identity theorists. Token-token identity seems much more plausible than type-type identity theory. Suppose I believe that Denver is the capital of Colorado and suppose you believe that Denver is the capital of Colorado. It seems unnecessary to suppose that in order to have the same belief we must be in exactly the same type of neurobiological state. My neurobiological state of believing that Denver is the capital of Colorado might be at a certain point in my brain, and yours might be at another point, without these being different beliefs.
Unfortunately, the identity theorists were often rather feeble at giving examples. One of their favorite examples was to say that pains are identical with C-fiber stimulations. The idea was that according to the type identity theorists, every pain is identical with some C-fiber stimulation and according to the token identity theorists, this particular pain might be identical with this particular C-fiber stimulation, but some other pain might be identical with some other state of a brain or some other state of a machine. Unfortunately, all of this is rather bad neurophysiology. A C-fiber is a type of axon; and it is true that certain types of pain signals, not all, are carried by C-fibers to the brain. But it would be ridiculous, neurophysiologically, to think there is nothing to pains except having your C-fibers stimulated. The C-fiber is just part of a complex pain mechanism in the brain and nervous system. Be that as it may, this was the sort of example that the identity theorists gave, and a good deal of the debate centered on whether or not we would get such type identities or whether token identities were all that we could hope for. In the long run the token identity theorists have been more influential than the type identity theorists.
But now they are faced with an interesting question. What is it that all of these tokens have in common that makes them tokens of the same mental-state type? If you and I both believe that Denver is the capital of Colorado, then what is it exactly that we share if there is nothing there but our brain states and we have different types of brain states? Notice that the two answers that would traditionally be given to this, the dualist answer and the type-type answer, will not do for the token physicalist. The token physicalists cannot say that what they have in common are the same irreducibly mental properties, because their whole idea was to eliminate, or get rid of, such irreducible mental properties. Nor can they say that they are the same type of brain state, because the whole move from type identity theory to token identity theory was to avoid having to say that every token of a particular mental-state type is identical with a token of a certain brain-state type.
At this point the materialists made a move that was crucial for subsequent philosophizing about the mind. They said: what token brain states have that makes them mental states is a certain type of function in the overall behavior of the organism. Not surprisingly, this doctrine was called “functionalism,” and when spelled out it developed into views like the following:17 to say that Jones believes that it is raining is to say that he has a certain event, state, or process going on in him that is caused by certain sorts of external stimuli—for example, he perceives that it is raining; and this phenomenon, in conjunction with certain other factors, such as his desire to stay dry, will cause a certain sort of behavior on his part, the behavior of carrying an umbrella. Mental states, in short, are defined as states that have certain sorts of functions, and the notion of function is explained in terms of causal relations to external stimuli, to other mental states, and to external behavior. We could write this out as follows: Jones’s perception that it is raining causes in him the belief that it is raining. The belief that it is raining and the desire to stay dry cause the behavior of carrying the umbrella. What, then, is a belief? A belief is anything that stands in these sorts of causal relations. At that point the identity theorists introduced a beautiful technical device to capture precisely this feature of their theory. The technical device is called a “Ramsey sentence,” after its inventor, the British philosopher, Frank Ramsey. In the previous conjunction of sentences we simply knock out the expression “the belief that it is raining” and put in “x.” Then we preface the whole sentence with an existential quantifier, which says “there is an x such that.” So it now comes out as follows: “there is an x such that the perception that it is raining causes x, and x together with the desire to stay dry causes the behavior of carrying an umbrella.” So, on this account, what is a belief really? It is anything, any x, that stands in these (and many of other such) causal relations. Mental states such as beliefs are not defined by any intrinsic features, rather they are defined by their causal relations, and these causal relations constitute their function. Beliefs, for example, are caused by perceptions and together with desires they cause actions. Such causal relations are all that there is to having a belief.
And what about the leftover reference to desires and perceptions? They too will be analyzed functionally. Just as there is an x that is the belief, and is defined by its causal relations, so there is a y that is the desire, and a z that is a perception, and they too are defined by their causal relations.
So several of the objections to behaviorism were met by the functionalist account. One objection was the apparent circularity in behaviorism of having to use desires to explain beliefs, and beliefs to explain desires. This objection is answered by the functionalist in one fell swoop, if we analyze beliefs and desires simultaneously in terms of their causal relations. Furthermore, we have immediately answered the objection that behaviorism left out the causal relations between mental states and external behavior, because we have defined mental states partly in terms of their capacity to cause external behavior. Furthermore, another appeal of the functionalist account of mental states is that it seemed to assimilate the mental realm to a very familiar realm of human functional entities. Thus if we ask, What is a carburetor? What is a thermostat? What is a clock?—all of these questions are answered causally, by describing the causal functions performed by carburetors, thermostats, and clocks. None of these things are defined by their physical structure. A clock, for example, can be made out of gears and wheels, it can be made out of an hourglass with sand in it, it can be made out of quartz oscillators, it can be made out of any number of physical materials, but the defining feature of a clock is that it is any physical mechanism that enables us to tell the time. Analogous remarks could be made about carburetors and thermostats. Mental states are like carburetors, thermostats, and clocks. They are defined not by their physical structure and not by any Cartesian mental essence; rather, they are defined by their causal relations. A belief is just any entity that, standing in certain relationships to input stimuli and to other mental states, will cause external behavior.
The underlying impulse of functionalism was to answer the question, Why do we attribute mental states to people at all? And the answer was, we say they have such things as beliefs and desires because we want to explain their behavior. Functionalism seems to have captured all of these intuitions.
The functionalists naturally wanted to know what was the nature of the inner mental brain states that enabled them to cause behavior. How did the mental states differ from other sorts of brain states? One answer was to say that this is not really a suitable question for philosophy at all; it should be left to psychologists and neurobiologists. We can treat the brain as just a “black box,” which produces behavior in response to stimuli, and we need not, as philosophers, worry about the mechanism inside the black box. This view was sometimes called “black-box functionalism.”
But black-box functionalism is intellectually unsatisfying in that it does not answer our natural intellectual curiosity. We want to know, really, how does the system work?
At this point there occurred one of the most exciting developments in the entire history of the philosophy of mind in the twentieth century. To many of us who participated in the developments (though not to me), it seemed like not merely an exciting development, but at long last a solution to problems that had beset philosophers for more than 2,000 years. The idea was based on a convergence of work in philosophy, cognitive psychology, linguistics, computer science, and artificial intelligence. It seemed that we knew the answer to the question that faced us: the way the system works is that the brain is a digital computer and what we call the “mind” is a digital computer program or set of programs. We had made the greatest breakthrough in the history of philosophy of mind: mental states are computational states of the brain. The brain is a computer and the mind is a program or set of programs. A principle that formed the foundation for any number of textbooks was this: the mind is to the brain as the program is to the hardware.18
This view is sometimes called “computer functionalism,” though I have also baptized it “Strong Artificial Intelligence” to distinguish it from Weak Artificial Intelligence, which aims to study the mind by doing computer simulations as opposed to purporting to create a mind. On the Strong AI view, the appropriately programmed digital computer does not just simulate having a mind; it literally has a mind.
With the advent of the computer model of the mind, it seemed that at long last we had the solution to the problems that had bothered Descartes, and indeed to problems that go back 2,500 years to the early Greek philosophers. In particular, it seemed we had a perfect solution to the traditional mind-body problem. The relation of mind and body seemed mysterious, but the relation of program to computer hardware, the relation of the software to its physical implementation, is not the least bit mysterious. It is a relation that is understood in every Computer Science department in the world, and this understanding is routinely employed on a daily basis to program computers.
So far I have criticized materialist views as they came up. But now I am going to set out the computer theory of the mind and save criticisms of it and other versions of functionalism till the next chapter. Before explaining in detail how the computer theory of the mind is supposed to solve our problems, I want to introduce several crucial notions. These notions are important not only for their relevance to contemporary philosophy but, indeed, for intellectual life in general. The notions I hope to explain briefly are those of an algorithm, a Turing machine, Church’s thesis, Turing’s theorem, the Turing test, levels of description, multiple realizability, and recursive decomposition. These notions lie at the heart of what was until recently, and in some quarters still is, the single most influential view of the nature of the mind in cognitive science and related disciplines. Furthermore, several of these ideas are so important that it is essential to your general education, quite apart from philosophy, that you should be fully familiar with these concepts.
Algorithms. An algorithm is a method for solving a problem by going through a precise series of steps. The steps must be finite in number, and if carried out correctly, they guarantee a solution to the problem. For this reason algorithms are also called “effective procedures.” Good examples are the methods used to solve problems in arithmetic, such as addition and subtraction. If you follow the steps exactly, you will get the correct solution.
Turing Machines. A Turing machine is a device that carries out calculations using only two types of symbols. These are usually thought of as zeros and ones, but any symbols will do. The idea of the Turing machine was invented by Alan Turing, the great British logician and mathematician. The striking feature of the Turing machine is its simplicity: it has an endless tape on which the symbols are written. It has a head that reads symbols on the tape. The Turing machine head will move to the left or to the right, it can erase a zero, it can print a one, it can erase a one, it can print a zero. It does all of these things in accordance with a program, which consists of a set of rules. The rules always have the same form; under condition C, perform act A: C 
A. For example, a rule might be of the form, if you are scanning a zero, replace it with a one and move one square to the left.
A Turing machine is not a machine in the ordinary sense. You cannot go into a store and buy a Turing machine. It is an abstract mathematical concept. For example, the Turing machine has an infinite tape and thus an infinite amount of storage capacity. No real machine has that. Real machines break down, get rusty, or have beer poured on them. Turing machines have none of these defects because they are purely abstract. However, though the concept of a Turing machine is the concept of something formal and abstract, for practical purposes the kind of computer you buy in a store is a Turing machine. Ordinary commercial computers implement algorithms by manipulating two sorts of symbols. Contemporary electronics is so sophisticated that the modern computer can carry out these symbolic operations at the rate of millions per second.
Church’s Thesis. Due originally to Alonzo Church (arrived at independently by Turing, so sometimes called the Church-Turing Thesis), this thesis states that any problem that has an algorithmic solution can be solved on a Turing machine. Or another way to say the same thing is that any algorithm at all can be carried out on a Turing machine. The idea of a machine that just uses binary symbols, zeros and ones, is sufficient to carry out any algorithm whatever. This is a very important thesis because it says in mathematical terms that any problem that is computable can be computed on a Turing machine. Any computable function is Turing computable.
Turing machines can come in many different kinds, states, and varieties. In my car there are specialized computers for detecting the rate of fuel consumption, for example. But in addition to the idea of these specialpurpose computers, or Turing machines, there is the idea of a general-purpose computer, something that can implement any program at all. And Turing, in an important mathematical result known as Turing’s Theorem, proved that there is a Universal Turing machine that can simulate the behavior of any other Turing machine. More precisely, Turing proved that there is a Universal Turing machine, UTM, such that for any Turing machine carrying out a specific program, TP, UTM can carry out TP.
What made these ideas so exciting was the following thought: Suppose the human brain is a Universal Turing machine? I cannot exaggerate the excitement that this idea generated, because it gave us at long last not just a solution to the philosophical problems that beset us, but it gave us a research program. We can study the mind, we can find out how the mind really works, by discovering which programs are implemented in the brain. An immensely appealing feature of this research program is that we do not actually have to know how the brain works as a physical system in order to do a complete and strict science of the mind. The specifics of the brain are really irrelevant to the mind, because any other physical system would do as well, provided only that it was stable enough and rich enough to carry the programs. On this view, the neurobiological details of brain operation are irrelevant to the mind. We just happen, by a kind of evolutionary accident, to be implemented in neurons, but any sufficiently complex hardware system would do as well as what we have in our skulls. To get a really adequate scientific account of the mind we need only to discover the Turing machine programs that we are all using when we engage in cognition.
The Turing Test. However, we need a test. We need a test that will tell us when a machine is genuinely behaving intelligently, and when it is not. This test was also invented by Alan Turing, and is called the Turing test. There are different versions of it, but the basic idea is this: we can side-step all the great debates about the other minds problem, about whether or not there really is any thinking going on in the machine, whether the machine is really intelligent, by simply asking ourselves, Can the machine perform in such a way that an expert cannot distinguish its performance from a human performance? If the machine responds to questions put to it in Chinese as well as a native Chinese speaker, so that other native Chinese speakers could not tell the difference between the machine and a native Chinese speaker, then we would have to say that the machine understood Chinese. The Turing test, as you will have noticed, expresses a kind of behaviorism. It says that the behavioral test is conclusive for the presence of mental states.
Levels of Description. Any complex system can be described in different ways. Thus, for example, a car engine can be characterized in terms of its molecular structure, in terms of its gross physical shape, in terms of its component parts, etc. It is tempting to describe this variability of descriptive possibilities in terms of the metaphor of “levels,” and this terminology has become generally accepted. We think of the microlevel of molecules as a lower level of description than the level of gross physical structure or physical components, which are higher levels of description. Most of the interest of this distinction is that it applies in a dramatic fashion to computers. At a lower level of description your computer and mine might be quite different. Yours may have a different type of processor than mine, for example. But at a higher level of description they may be implementing exactly the same algorithm. They may be carrying out the same program.
Multiple Realizability. The notion of different levels of description already implicitly contains another notion that is crucial to the computational theory of mind, and that is the idea of multiple realizability. The point is that a higherlevel feature, such as being the Word program or being a carburetor, may be physically realized in different systems, thus one and the same higher-level feature can be said to be multiply realizable in different lower-level hardwares. Multiple realizability seems to be a natural feature of token identity theories. The different tokens of different types at the lower level may be different forms of realization of some common higher level mental feature. Just as the same computer program may be implemented in different sorts of hardware and thus is multiply realizable; so the same mental state, such as the belief that it is going to rain, might be implemented in different sorts of hardware, and thus also be multiply realizable.
This diagram illustrates the distinction between levels of description and the multiple realizability of the higher level in lower levels:
One and the same system represented by the line AB can be realized in different lower level systems, represented by lines BC, BD, BE, BF, and BG
Recursive Decomposition. Yet another important idea, already implicit in what I have said, is that big complex problems can be broken down into little simple problems, which can be broken down into even simpler problems, until we reach the level of ultimate simplicity. Doing multiplication with several digits, for example multiplying 28 x 71, may seem to us a complex operation, but the beauty of the idea of a Turing machine is that at bottom, all such problems break down into simple maneuvers with zeros and ones. You print a one, you erase a zero, you move one square to the left or one square to the right. That is all the machine needs to know how to do in order to carry out not only arithmetic but the most incredibly complex algorithms for other sorts of tasks. The complex tasks can be broken down (decomposed) into simple tasks by repeated application (recursively) of the same procedures until all that is left are simple binary operations with two symbols, the zeroes and ones. In the early heady days, some people even said that the fact that neurons were either firing or not firing was an indication that the brain was a binary system, just like any other digital computer. Again, the idea of recursive decomposition seemed to give us an important clue to understanding human intelligence. Complex intelligent human tasks are recursively decomposable into simple tasks, and that is how we are so intelligent.
The collection of ideas that I have just explained contains the tools necessary to articulate the single most influential and powerful theory of the mind in the last decades of the twentieth century. The brain is a digital computer, in all probability a Universal Turing machine. As such it carries out algorithms by implementing programs, and what we call the mind is a program or a set of such programs. To understand human cognitive capacities it is only necessary to discover the programs that human beings are actually implementing when they activate such cognitive capacities as perception, memory, etc. Because the mental level of description is a program level, we do not need to understand the details of how the brain works in order to understand human cognition. Indeed, because the level of description is at a higher level than neuronal structures, we are not forced to any type-type identity theory of the mind. Rather, mental states are multiply realizable in different sorts of physical structures, which just happen to be implemented in brains but could equally well have been implemented in an indefinite range of computer hardwares. Any hardware implementation will do for the human mind provided only that it is stable enough and rich enough to carry the programs. Because we are Turing machines we will be able to understand cognition by reducing complex operations into the ultimately simplest operations, the manipulation of zeros and ones. Furthermore, we have a test that will enable us to tell when we have actually duplicated human cognition, the Turing test. The Turing test gives us a conclusive proof of the presence of cognitive capacities. To find out whether or not we have actually invented an intelligent machine we need only apply the Turing test. And we now have a research project; indeed, it is the research project of cognitive science.
We try to discover the programs that are implemented in the brain by designing programs for our commercial machines that will pass the Turing test, and then we ask the psychologists to perform experiments on humans to see if they are following the same program as the program on our computer. For example, in one famous experiment involving the memory of numbers, the reaction times of the subjects seemed to vary in the same way as the processing time of a computer. This seemed to a lot of cognitive scientists good evidence that the humans were using the algorithmic procedures of the computer.
Such was the appeal of the computational theory of the mind in the early days of cognitive science. If I have not made it sound appealing to you, then I have not done a good job of exposition, because to many it was immensely exciting at the time. It spawned a thousand research projects and it garnered a nearly equal number of research grants. But, alas, it is hopelessly mistaken. I thought so at the time, and nothing since the early days has changed my opinion. In the next chapter, I will explain why it is mistaken. For now, I want you to appreciate its appeal.
With some hesitation (because it oversimplifies) I present a chart that shows the relations between the theories we have so far discussed.
One of the interesting features of materialism is that just about every conceivable materialist position has been taken by some philosopher or other. And to complete the story of modern materialism, I want to mention two other versions: eliminative materialism, the idea that mental states do not exist at all, and anomalous monism, Donald Davidson’s idea, which is a version of the token-identity theory.
The eliminative materialists argued as follows.19 Why do we say that people have beliefs and desires and other sorts of mental states? We say these things because we wish to be able to explain their behavior. Our postulation, therefore, of beliefs and desires, etc., is the postulation of a kind of theoretical entity, much as the postulations of electrons or electromagnetic force in physics are postulations of theoretical entities. It is characteristic of such postulations that if the theory proves to be false, that is sufficient to establish that the entity does not exist. The now-obsolete theory of phlogiston, that the burning of an object consisted in the release of a substance called “phlogiston,” has been refuted and with the refutation of the theory, we no longer believe in the existence of phlogiston. What then is the theory that postulates beliefs, desires, etc.? Well, it is common sense, or grandmother psychology, and, in the literature it is usually called “folk psychology.” But now, so the story goes, folk psychology is almost certain to be shown to be an inadequate, indeed, a false theory. Why? Well, for one thing, folk theories have always been refuted by scientific progress. Furthermore, folk psychology is going nowhere as a research program. Our folk theories of rationality, for example, are not much of an improvement on Aristotle’s. But if the theory that postulates beliefs, desires, etc., is a false theory, then these entities do not exist. So eliminative materialism simply is a version of materialism that eliminates mental states altogether. They are shown to be illusions in the way that sunsets and phlogiston are illusions.
Another and related argument against the entities of folk psychology was one that appealed to the absence of type-type reductions of the folk-psychological notions to neurobiological phenomena. A mature neuroscience is very unlikely to make much use of notions such as belief and desire, because these notions do not match the categories of neurobiology. With the absence of a type-type reduction of beliefs and desires, it seems reasonable to suppose that such entities do not exist.
Anomalous monism is a view put forward by Donald Davidson20 for which he advances the following argument:
Step 1: There are causal relations between mental phenomena and physical phenomena.
Step 2: Wherever there are events related as cause and effect they must fall under strict, deterministic causal laws.
Step 3: But there are no such strict deterministic causal laws relating the mental and the physical. In Davidson’s terms, there are no psycho-physical laws.
Therefore,
Step 4: Conclusion. All so-called mental events are physical events.
They have to be physical events to instantiate physical laws, and when we describe them as mental, we are just picking out a category of physical events that satisfy a certain mental vocabulary. They are mental under one description, but the same events are also physical under another description. The result, then, is a kind of materialism, a materialism that says that the subject matter of the psychological sciences will never be describable by the kinds of universal laws that we get in physics, not because they are a mysterious kind of spiritual or mental entity, but rather because the descriptions we use to pick them out, the mental descriptions, do not relate in a lawlike fashion to physical phenomena picked out under physical descriptions. The only argument that Davidson gives for this point is that mental phenomena, like beliefs and desires, are subject to constraints of rationality, and rationality has “no echo in physics.”
I have tried to be as fair as I can in laying out the standard versions of materialism over the past century. If I have not made them seem the least bit appealing then I have failed in my task of expounding other people’s views. I have to confess, however, that I think all of these theories are hopelessly inadequate. In subsequent chapters I am going to discuss their inadequacies. For the purposes of the immediate discussion I will assume that behaviorism is not a plausible form of materialism and that we need to examine the different forms of physicalism, especially functionalism.
Most of the discussion in the next chapter will be about the historical tradition of functionalism culminating in Strong Artificial Intelligence. I will not say anything about anomalous monism, because it turns out that it falls under the general heading of token identity theories. I will now be brief about, but I hope not unfair to, eliminative materialism. I mentioned three arguments for eliminative materialism. The first argument says that the entities of folk psychology are postulated as part of a theoretical structure. But in general, that is not true. My actual conscious thought processes of making up my mind to try to get something because I desire it are all directly experienced by me.
The second argument is that the propositions of folk psychology are in all likelihood going to prove to be false. But the problem is, if you look at the authors who hold this view, they are extremely implausible in their specification of the propositions of folk psychology. Sometimes they attribute to us beliefs that we obviously do not hold. For example, one author attributes to us the belief that if we believe that p and if we believe that if p then q we will believe that q.21 This is an incredible claim. It would imply, for example, that anyone who believes each member of a complicated set of propositions, a, b, c, etc., that occur in the premises of a proof, where the other premises occur in conditionals of the form, if a then d, if b then e, if c then f, etc., would automatically believe all of the logical consequences. If this were true, such complex logical and mathematical proofs could never surprise us, because we believed the conclusion all along! The absurdity derives from confusing our logical commitment to the truth of a proposition with actually believing the proposition before becoming aware of our commitment. Complex logical and mathematical proofs show what our belief in the premises commits us to believing in the conclusion. They do not show that we really believed the conclusion all along.
And indeed, eliminative materialists are extremely hesitant to state the propositions of folk psychology. I believe there is a reason for this. Many of the propositions of so-called folk psychology are not in fact empirical propositions. They are in a sense, constitutive principles, they are analytic principles of our mental contents. So, for example, here is a proposition of folk psychology: beliefs can typically be either true or false. Now, the problem with treating that as if it were a hypothesis that might turn out to be false is that it is part of the definition of belief, it is a constitutive principle. It is like saying that touchdowns in American football count six points. The difficulty with the eliminative materialists is that they treat the propositions of so-called folk psychology as if they were empirical hypotheses, but in many cases they are not. If you read in the newspaper that investigators at MIT using the latest computer technology have discovered that touchdowns do not in fact count six points, but count only 5.99999 points, then you know that they have made a stupid mistake. The proposition that touchdowns count six points is part of the definition of a touchdown, as it is currently defined by the rules of American football. You cannot discover that it is false the way that you can discover that ordinary empirical propositions are false. Some of Churchland’s examples are like this. He says that it is a proposition of folk psychology that someone who fears that p does not want p to happen. But if you add an “other things being equal” clause, that is part of the definition of fear. If I am afraid of something then, other things equal, I do not want the thing I am afraid of to happen. So you can’t show that folk-psychological entities do not exist by first showing that our beliefs about them are false because many of the basic propositions of folk psychology are similarly definitional, or analytical, or constitutive principles of the entities of folk psychology. This is why the enemies of folk psychology are so inadequate in their efforts to formulate refutations of it. This does not prove that the entities of folk psychology exist, but that one argument to show they do not exist does not get off the ground.
The last argument against folk psychology is even worse. The idea is that because we cannot do a smooth type-type reduction of beliefs, desires, etc., to neurobiology, that therefore somehow or other these entities do not exist. But compare a similar proposition: we cannot do a smooth type-type reduction of sports utility vehicles, tennis rackets, or split-level ranch houses to the entities of atomic physics. We cannot do a type-type reduction for reasons implicit in this chapter: tennis rackets, etc., are multiply realizable in physics. Indeed, atomic physics really has no use for the notion of a sports utility vehicle, a split-level ranch house, or a tennis racket. But does anyone in his right mind think that it follows from this that these entities do not exist? As a general formal argument, the fact that we do not get type-type reductions of some entity into more basic sciences does not show that the irreducible entities do not exist. Quite the contrary.
There is an interesting irony in all of this discussion. Reductionists and eliminativists tend to think their positions are quite different. Reductionists think mental entities exist but can be reduced to physical events. Eliminativists think mental entities do not exist at all. But these amount to very much the same conclusion. Reductionists say there is nothing there but brain processes materialistically described. Eliminativists say there is nothing there but brain processes materialistically described. The apparent difference is a difference in vocabulary. The earlier materialists wanted to show that mental states did not exist as such by showing that they could undergo a type-type reduction to the entities of neurobiology. The later eliminative materialists wanted to show that the entities of common-sense psychology do not exist at all by showing that they cannot undergo a type-type reduction to the entities of neurobiology. Neither argument is any good, but what they suggest is that these people are determined to try to show that our ordinary common-sense notions of the mental do not name anything in the real world, and they are willing to advance any argument that they can think of for this conclusion.
