James W. Cornman, Keith Lehrer, George S. Pappas, Philosophical Problems and Arguments: An Introduction, 3d edition, 1982. [1st ed., 1968; 2d ed., 1974]
Chapter FOUR
The Mind-Body Problem


James W. Cornman

What is a person? For one thing a person is a complex being who can do many things. Unlike many other beings he can move himself; he can crawl, walk, and swim. These are clearly bodily activities. A person, then, surely seems to have a body, in which many processes and events take place, such as the beating of the heart, the functioning of the kidneys, and the complex functioning of the brain. Such bodily processes are essential for keeping a person alive and healthy. Indeed, we describe the state of a person's body by stating the condition of such vital bodily processes.

There are, however, many other things a person can do which do not seem to be bodily activities. A person, unlike many other beings who can move themselves, can think about things; decide on courses of action; hope for, desire, and dream about many different things. These seem to be mental activities, quite different from bodily activities and processes. They seem, therefore, to involve a mind rather than a body, a mind with states quite different from bodily states. We describe a person's mental state when we call him happy or sad, gay or depressed, in love or full of hate, nervous or calm, bold or afraid. A person, thus, seems to be not merely a complex body, but an entity with a mind distinct and quite different from his body.

We do not merely describe a person in this way; we also try to explain his bodily behavior and understand the workings of his mind. In doing this we usually become involved in claims about relationships between his mind and his body. We explain, for example, Mrs. Jones' uncharacteristic screaming at her children by referring to her splitting headache or to frightening thoughts she has. We claim that the reason Smith will not climb mountains is because he is deathly afraid of heights, or that Mr. Brown has stopped smoking because he has decided the risk of cancer is too great. We also explain the abnormal behavior of persons as being caused by guilt feelings, repressed desires, or neurotic fixations. On the other hand, we explain someone's pain by isolating its cause as some bodily injury; we use certain injections in his body to make him unconscious, and sometimes we perform brain operations in order to change the whole mental state, the whole personality of a person. It seems, then, that certain mental phenomena can affect the body and that certain bodily phenomena can affect the mind.

We have described a person as a complex entity with a mind and a body; an entity involving both bodily events and states, and mental events and states; an entity in which certain bodily events causally affect the mind and certain mental events affect the body. Furthermore, because it seems that the realm of the mental is quite distinct and different from the realm of the material, this description seems to lead to the theory known as dualistic interactionism. According to this theory, a person consists of two quite radically different parts, a mind and a body, each of which can causally act upon the other.

Dualistic interactionism is accepted by many people. Most of us, in our own cases at least, distinguish sharply between those mental phenomena of which we are aware, such as our own sensations, and our body with all its complex physical processes. Furthermore, a mind-body dualism seems to be essential to most religions. The body will disintegrate after death but, according to the doctrines of many religions, the soul, the immaterial part of us which is quite distinct and different from the body, will live on eternally. However, although it may be easy to explain the widespread acceptance of dualistic interactionism, such an explanation is not a philosophical concern. The primary philosophical problem is to find out whether dualistic interactionism or some other position is the most plausible view about the nature of a person. Obviously, there are many possible alternatives. There are several monistic views: reductive materialism, which claims that there are no minds but only bodies; idealism, which claims that there are no bodies, only minds; and a neutral theory, which claims that a person is neither mind nor body, but something quite different from either. There are also dualistic theories that deny all or part of the claim of causal interaction between minds and bodies. Epiphenomenalism denies that the mind can causally affect the body because the mind is merely a kind of by-product of certain complex physical processes. Parallelism claims that there is no causal interaction of any kind between minds and bodies. Each proceeds in its own way, parallel to but independent of the other.

Some of these views are plausible; others are not. All face problems but some problems are more damaging than others; and because no view is obviously correct, each requires a reasoned defense if we are to justify it. It will be our task to evaluate critically the leading alternative positions with the hope that we will be able to choose from among them one that can be shown to be more plausible than any other. We shall start with dualistic interactionism.


Before beginning the discussion proper we must first indicate how certain key expressions, containing the terms 'material' and 'mental,' shall be used. We have already discussed bodies (material objects) and minds (mental objects). We have also discussed events and states, both mental and material. Let us construe these two different kinds of objects, events, and states in the following ways:

Material object:
An object (such as a stone) that has size, shape, mass, and spatial and temporal position, and that can exist independently of any conscious being.
Mental object:
An object that is either a conscious being, that is, a being aware of things (such as a mind), or a being that cannot exist independently of some conscious being (such as a thought or sensation).
Material event:
Something (such as the movement of an arm) that occurs over a period of time and consists of only material objects.
Mental event:
Something (such as a dream) that occurs over a period of time and consists of only mental objects.
Material state:
A condition or situation (such as an infection) of some material object.
Mental state:
A condition or situation (such as a psychosis) of some mental object.

It is important to notice that, as characterized previously, the mental and the material are radically different. Whatever is mental depends essentially on consciousness or awareness, but what is material does not. Furthermore, it certainly seems that nothing mental has size, shape, mass, or spatial location; such qualities seem only to characterize the material. The only characteristic that the mental and the physical seem to have in common is that both can have temporal positions. We say "can have" here because although all material objects and all human minds have temporal positions, it may be that there are minds that do not exist in time, for example, the mind of God. It should also be noted that the word 'material,' rather than 'physical,'has been used throughout. This is because by 'physical' we shall mean 'part of the subject matter of the physical sciences,' and it may well be that not all physical objects are material objects. An object which is a person is, if the dualists are correct, neither a mental object nor a material object; rather, it is a composite of both kinds of objects. However, such a being falls within the subject matter of physics. We are not, then, interested so much in the physical as in the mental and the material, although the physical is relevant because part of the debate surrounding the mind-body problem concerns whether physics, which supposedly can explain the behavior of all material objects can explain all human behavior.


The classical exposition of dualistic interactionism is that given by Rene Descartes. According to Descartes we can clearly distinguish between three different kinds of substances: one the eternal substance God, and the other two, substances created by God. He says: "We may thus easily have two clear and distinct notions or ideas, the one of created substance which thinks, the other of corporeal substance, provided we carefully separate all the attributes of thought from those of extension."1

However, although there are these two radically different created substances, one which is extended and does not think (body) and one which thinks but is not extended (mind), Descartes claims that he, and therefore other people, are essentially thinking substances. Yet he finds that he is not only a mind, for as he says, "I have a body which is adversely affected when I feel pain, which has need for food or drink when I experience the feelings of hunger and thirst, and so on. . . ."2 But it is not that people are merely minds that happen to have bodies, according to Descartes. It would be better to call them embodied minds, for he claims to have found

that I am not only lodged in my body as a pilot in a ship, but that I am very closely united to it, and so to speak so intermingled with it that I seem to compose with it one whole. For if that were not the case, when my body hurt, I, who am merely a thinking thing, should perceive this wound by the understanding only, just as the sailor perceives by sight when something is damaged in his vessel. . . .3
These two kinds of substances which make up each person intermingle in such a way that they causally act upon each other. Although it might be that a mind interacts with each part of its body separately, Descartes' view is that the mind interacts only with the brain. This agrees with findings of science that various brain processes bring about certain bodily movements and that certain bodily events causally affect the brain. The usual view, then, is a mind-brain interaction theory. It is usually held, for example, that a material event that causally stimulates one of our five senses -- for example, light waves hitting the retina of the eye -- results in a chain of physical causation which leads to a certain brain process from which a certain sensation results. It is also held that because certain bodily behavior has been brought about by affecting the brain in certain ways, mental events act on the body by affecting the brain. Descartes thought he could pinpoint mind-brain interaction more precisely than this. He claimed that there is just one point of immediate "contact" or interaction between mind and body. Through this point of contact the causal effects of the mind are carried to all parts of the body and the causal effects of all parts of the body are transmitted to the mind. As Descartes says, "the part of the body in which the soul exercises its functions immediately is in nowise the heart, nor the whole of the brain, but merely the most inward of all its parts, to wit, a certain very small gland which is situated in the middle of its substance. . . ."4 Again he adds that

the small gland which is the main seat of the soul is so suspended between the cavities which contain the spirits that it can be moved by them in as many ways as there are sensible diversites in the object, but that it may also be moved in diverse ways by the soul, whose nature is such that it receives in itself as many diverse impressions, that is to say, that it possesses as many diverse perceptions, as there are diverse movements in this gland. Reciprocally, likewise, the machine of the body is so formed that from the simple fact that this gland is diversely moved by the soul, or by such other cause, whatever it is, it thrusts the spirits which surround it towards the pores of the brain, which conduct them by the nerves into muscles, by which means it causes them to move the limbs.5
This gland that Descartes thought to be the "seat" of the mind or soul is the pineal gland. It functions, according to him, as the intermediary that transmits the effects of the mind to the brain and the effects of the brain to the mind. He was wrong about this, however, because there is reason to think the gland is not affected by all brain processes that affect the mind nor by all mental phenomena that affect the body. Consequently, although we shall agree with Descartes in construing dualistic interactionism as a mind-brain theory, we shall disagree with him about the role of the pineal gland in this interaction. We shall also disagree with him on another point. As more recent dualists have claimed, it is, strictly speaking, wrong to speak of minds and brains interacting, because it is events that are causally related, not substances. Thus, although we shall sometimes talk of minds and brains interacting, and also of mental events and brain events interacting, these statements should always be construed to mean either that some brain event is causing a mental event or that some mental event is causing some brain event.

We may state the theory of dualistic interactionism in a capsule way as follows: First, this theory holds, plausibly enough, that there are material things (objects, events, and states), as well as mental things (objects, events, and states). Second, this theory holds that mental things are completely different kinds of things from material entities. Mental entities, according to this theory, are totally nonmaterial or, as we could also say, immaterial. They thus would have no size, shape, weight, mass, or spatial location. Perhaps one could say that mental objects (as opposed to events or states) would be something like pure spirits. It is this second point that makes the theory dualistic. It is not enough to hold merely that there are mental and material things for, as we will see, nondualistic theories accept that claim. It is only when this claim is taken in conjunction with the further contention that mental things and material things are completely different kinds of things, with mental things having no material features except perhaps occurring or existing at a time, that we get the dualistic aspect of the theory. Third, dualistic interactionism holds that a person is not wholly a mental thing, nor wholly a material thing; instead, on this theory a person is a composite entity consisting of a mental object (a spiritlike, immaterial mind) joined together with a material body. Fourth and last, on this theory mental and material events causally interact, that is, they causally affect one another. It is this latter element that makes the theory an interactionist one.

Although accepted by many people, dualistic interactionism is by no means immune to powerful objections, objections that many philosophers have found so damaging that they have rejected the position. In general, there have been two kinds of objections, those based on the requirements of science and those based on philosophical grounds. We shall consider the three strongest objections of each kind.


First Philosophical Objection: Where Does Interaction Occur?

According to dualistic interactionism certain mental phenomena, such as fear, cause certain bodily behavior, and certain bodily events, such as spraining an ankle, causally bring about mental events, such as pain. Supposedly, this point of interaction between the mind and body is in the brain, because mental events directly affect brain processes. But, goes the objection, no mental event has a spatial location; no mental event occurs at some place. How then can it be said that the mental events that causally affect brain events are located in the brain? To be in the brain is to have a spatial location. Thus because they have no spatial location, mental events are not in anything. They are therefore not in the brain and thus do not interact with events that are in the brain.

This objection emphasizes something important about mental phenomena as the dualistic interactionist construes them -- they have no spatial locations. Where are your thoughts, your desires, your dreams, your sensations, and emotions? Surely not several inches behind your eyes, somewhere in your brain. No one examining your brain, no matter how thorough the examination, would ever find them there. They have no spatial location. It will not help to talk about some place as the seat of the mind, as Descartes did, because a seat is a place where something is spatially located, and the mind has no spatial location We can conclude, then, that there is no place at which mental events causally interact with brain events, because mental events do not occur anywhere. We can, consequently, reject the question, "Where do minds and brains interact?" as senseless. There is no such place.

But how does this affect the claim of the dualistic interactionist? The objection asserts that the interactionist is committed to claiming that the interaction takes place in the brain, because he claims that what mental events directly interact with are brain events that are in the brain. The core of the objection, then, is that if one thing interacts with another, the first must be located where the second one is. This seems quite reasonable regarding material events, especially when based on the doctrine that there is no action at a distance. But there is no reason to think that this is at all relevant to mind-brain causal action. Mental events, according to the dualistic interactionist, are neither close to nor at a distance from brain events, because they have no location. The interactionist, then, can rebut this objection by denying that she is committed to locating mental events in the brain. All she claims is that mental events interact with certain bodily events that are in the brain, but it does not follow from this that the mental events must also be in the brain.

The first objection is not fatal to interactionism, but it does bring to light the puzzling nature of the so-called interaction. It might bring someone to argue that there can be no mind-body interaction because brain events can interact only with something located at some place and mental events cannot be located at a place. This objection, however, begs the question at issue because it assumes that brain events can interact only with bodily events, and this is the very question at issue. We can, then, dismiss the first objection, although we should remember that there are some puzzling factors involved in such an interaction. It is these factors that give rise to the second objection.

Second Philosophical Objection: How Can Interaction Occur?

The main point emphasized by the first objection is that, for the dualist, mental events and bodily events are radically different. Consequently, it would seem that these two different kinds of events would have radically different kinds of causal abilities. Consider how material phenomena are causally affected. Material bodies and events are causally affected by something exerting physical force upon them in some way. To move or change a material body or to begin or change some bodily process it seems that some physical force must be exerted upon some material object. But because physical force is a product of mass and acceleration, whatever can exert physical force must have mass and must be capable of acceleration, that is, change of the rate of motion through space. But nothing mental has mass; nothing mental can accelerate, because nothing mental can travel from place to place. Therefore, states the objection, nothing mental can exert physical force; thus, nothing material can be causally affected by anything mental. Consider also how one body causally affects something else. As brought out earlier the causal efficacy of a body is the result of its physical force. But how can physical force be exerted upon that which has no mass, no size, no spatial r location? Therefore, if dualistic interactionism is a correct theory, there is neither action of mind on body nor action of body on mind.

This objection has been considered by C. D. Broad, one of the leading contemporary defenders of dualistic interactionism. He has summarized the objection as follows:

Now the common philosophical argument is that minds and mental states are so extremely unlike bodies and bodily states that it is inconceivable that the two should be causally connected. It is certainly true that, if minds and mental events are just what they seem to be to introspection and nothing more, and if bodies and bodily events are just what enlightened common sense thinks them to be and nothing more, the two are extremely unlike. And this fact is supposed to show that, however closely correlated certain pairs of events in mind and body respectively may be, they cannot by causally connected.6
Broad goes on to refute this argument as follows:
One would like to know just how unlike two events may be before it becomes impossible to admit the existence of a causal relation between them. No one hesitates to hold that draughts and colds in the head are causally connected, although the two are extremely unlike each other. If the un-likeness of draughts and colds in the head does not prevent one from admitting a causal connection between the two, why should the unlikeness of volitions and voluntary movements prevent one from holding that they are causally connected?7
Broad, then, is willing to admit that mental events, such as making decisions, are quite different from those things to which they are supposed to be causally relevant, namely, certain voluntary bodily movements. But because many causes are radically different from their effects, there is no reason to think mental events and brain events cannot causally interact merely because they are so different.

There are two replies that someone might make to Broad. First, he could point out that Broad is correct if the objection is construed to assert that mind-body interaction is logically impossible, but the objection should not be taken this way. It is meant to assert that the denial of mind-body interaction is completely justified. Surely it is possible that things which are extremely dissimilar causally interact, but when they are so dissimilar that the only characteristic they have in common is temporal position, then it seems most unreasonable to claim that they do interact. It is logically possible that a single ant will move the Washington Monument, but we are surely justified in saying, nevertheless, that it will not. And an ant is more like a monument than a mental event is like a bodily event.

The second reply to Broad is that the original argument, before he reformulated it, is not based merely upon the absence of like characteristics, but ultimately upon the absence of characteristics relevant to causal interaction. Although a draught and a cold are quite different, both are still material. A cold is a condition of certain parts of a person's body, and we can understand how a draught, which is a movement of air molecules, could have some sort of effect on something bodily. We can see certain effects of air on bodies time and time again. A flow of air occurs at a place, has a certain temperature, a certain moisture and pollen count, and a certain amount of physical force. Such characteristics are quite relevant to having causal affects upon material objects. The point of emphasizing the great dissimilarity between mental events and bodily events is not to justify the claim that dissimilarity rules out causation, but to emphasize that of all the usual characteristics relevant to causal interaction with material objects, the only one found in mental events, according to the dualistic interactionist, is temporal position, which by itself is surely not sufficient for causal action.

What can Broad say to these replies to his claim? Both replies state, essentially, that there is good reason to conclude that minds and bodies do not causally interact either because, according to the dualistic interactionist, they have only one property in common or because whatever other properties each has, they are not causally relevant to the other. His best retort would be to rely on the reply of another contemporary interactionist, C. J. Ducasse. He says,

The causality relation does not presuppose at all that its cause-term and its effect-term both belong to the same ontological category, but only that both of them be events.

Moreover, the objection that we cannot understand how a psychical event could cause a physical one (or vice versa) has no basis other than blindness to the fact that the "how" of causation is capable at all of being either mysterious or understood only in cases of remote causation, never in cases of proximate causation. For the question as to the "how" of causation of a given event by a given other event never has any other sense than through what intermediary causal stens_ does one cause the other.8

There are two relevant claims here. The first is that the matter of determining what things are causally related is completely empirical; the only restriction is that it be events that are causally related. Thus before we have examined specific situations, we can impose no restrictions upon what kinds of events can causally interact. We must observe actual situations and do actual experiments to decide the issue. Thus, we must find out by observation and experiment whether minds and bodies interact rather than proclaim that they cannot or do not because they are so different. Nor should we proclaim what characteristics are relevant to causation. We must also find this out by observation and experimentation. The second relevant claim made by Ducasse is that when we come to proximate, or immediate causes we must accept them as brute facts. There is no way to explain them because we can explain how one event causes another only when the cause is remote rather than proximate, that is, only if the cause brings about the effect by means of some other intervening events. We can, for example, explain why heating gas causes the pressure on its container to increase by saying that increasing the temperature of gas causes the molecules of the gas to move more rapidly and thus hit the walls of the container with more force. But if an increase of temperature is an immediate or proximate cause of an increase of molecular speed, we cannot explain how this causal action works. Explanation comes to an end with proximate causes and we must merely accept that such causes have the effects they do. Consequently, although we can explain how a desire for drink causes us to reach for a glass of water by explaining that the desire causally affects the brain, which by means of the nerves causally affects the arm, we cannot explain how a desire affects the brain because this is a case of proximate causation. We should, then, as with all cases of proximate causation, accept it as a brute fact, a fact no more and no less mysterious than any brute fact.

Ducasse's reply seems to be satisfactory if it is indeed true that observation and experimentation provide grounds for claiming that minds and brains have a causal relationship that is proximate rather than remote. Ducasse is correct in warning us not to approach any situation with a preconceived view of what the relevant causal factors f~are. But unless he can provide some evidence based upon observation of minds and bodies that they do causally interact, and that this causal interaction is immediate or proximate, then we have a right to use the results of other observations to help us decide. Consequently, because in all other observed cases of causal interaction involving material events we find that both cause and effect involve objects with mass and spatial position, we have some evidence, meager though it may be, against the claim that immaterial minds and bodies interact. If there is no evidence for the claim to counteract this contrary evidence, then, in spite of the claims of Broad and Ducasse, we should follow the evidence and conclude that immaterial minds and bodies do not interact.

The crucial question now before us is not whether the mental and the material causally interact; rather, the issue concerns whether this causal relation is at least sometimes proximate. We can approach this topic by again citing an example of Broad's. He says,

It is perfectly plain that, in the case of volition and voluntary movement, there is a connexion between the cause and the effect which is not present in the other cases of causation, and which does make it plausible to hold that in this one case the nature of the effect can be foreseen by merely reflecting on the nature of the cause. The peculiarity of a volition as a causal-factor is that it involves as an essential part of it the idea of the effect. To say that a person has a volition to move his arm involves saying that he has an idea of his arm (and not of his leg or his liver) and an idea of the position in which he wants his arm to be. It is simply silly in view of this fact to say that there is no closer connexion between the desire to move my arm and the movement of my arm than there is between this desire and the movement of my leg or my liver. We cannot detect any analogous connexion between cause and effect in causal transactions which we view wholly from the outside, such as the movement of a billiard-ball by a cue. It is therefore by no means unreasonable to suggest that, in the one case of our own voluntary movements, we can see without waiting for the result that such and such a volition is a necessary condition of such and such a bodily movement.9
In this passage Broad is claiming not only that in the case of voluntary bodily movements we have evidence that minds and bodies are causally connected, but also that in such a case we have perhaps the best evidence available that there are causal connections. We can, I think, agree with Broad on his first, more modest, claim that there are times when we decide to move one of our arms and its subsequent movement clearly seems to have resulted from our decision. That we seem to experience causal connections between some decisions and some bodily movements is surely some evidence that there are causal connections between minds and bodies. However, Broad thinks that we can support a stronger claim, namely, that we have adequate (indeed, even fully sufficient) evidence for such a causal connection because of one unique feature of decisions. A decision unlike any other causal factor involves the idea of the effect. This unique connection between cause and effect provides, according to Broad, the grounds for this stronger claim. However, at this point Broad seems to be arguing contrary to the lesson we learned from Ducasse in his defense of mind-body interaction. That is, the only factors we should count as causally relevant are those we find to be so through experience. The idea a person has of, for example, his arm's movement should be declared causally relevant to the movement of the arm only if there is evidence that mental phenomena such as ideas are causally relevant to bodily phenomena. To proclaim that they are relevant is no better justified than to proclaim that they are not. Thus Broad cannot justify his defense of interactionism by reliance upon a claim that having an idea of an event under certain conditions (for example, when deciding) is relevant to the causation of the event.

We should certainly reject Broad's stronger claim. But it might will be wondered whether his weaker claim is of any real use to us in the present context. The weaker claim is just that we seem to experience causal connections in our own cases between our own decisions and our own subsequent behavior, and this is some evidence that there are causal connections between the mental and the material. However, as we noted on p. 151, whether there are such causal relations is not the issue. Instead, the question concerns whether such causal connections, between the mental and the material, are immediate or proximate and Broad's example does not really bear on this question at all.

If we reflect on Broad's example, we notice that a mental event (a decision or volition) is causally connected to subsequent bodily movement. The relationship between these two events, however, is not that of proximate causation because there would be many neural and muscular events intermediate between them. However, consider the causal chain of events which, we may here suppose, starts with the mental event. This causal chain is a chain of events. It is reasonable to think that the event immediately following the mental event (decision) is itself a material event. This is because it is reasonable to believe that this next event after the mental event is a brain or neural event. This latter event produces another material event, and this one another, and so on until the bodily behavior occurs. The important point is that the relation between the mental event and the material event which comes next in the causal chain is a relation of immediate, proximate causation. Hence, it does not really matter whether a mental event, such as a decision or the occurrence of a thought, is itself the immediate cause of some bodily behavior. As long as we grant that there is some causal relation between the mental event and the subsequent behavior, as it seems clear there is, then we can argue as on p. 151, that there is a relation of proximate causation between a mental event and a material event. This relation will obtain between the mental event and the next event in the causal chain leading to the behavior. Of course, we can no more explain this case of proximate causation than we can any other case of proximate causation, whether between two material events or between a mental and a material event.

The argument here is simple and direct. If we agree that mental events are causally related to such things as bodily behavior, as of course we will and should, then it is plausible to hold that causal relations between mental events in the causal chain leading to the behavior and some other material event in the same chain are proximate or immediate. Here, then, is some evidence to counteract the evidence against dualistic interaction. It is not, strictly speaking, evidence in favor of dualistic interaction. It is, instead, evidence in favor of the claim that relevant causal relations, between the mental and the material, are proximate, and thus a demand for an explanation of how such interaction can occur is otiose. This means that we have not yet found grounds sufficient for rejecting dualistic interactionism, but neither have we found grounds sufficient for accepting it. Moreover, there is still a lingering mystery about how dualistic interaction can occur, despite the fact that we are in no position to explain how this occurs. Finding that such relations are causally proximate removes the force of the objection, but does not quite dispel the feeling of mystery. Consequently, the acceptability of dualistic interactionsim depends not only on the seriousness of remaining objections to it, but also on whether we can find reasonable positive evidence in its favor.

Third Philosophical Objection: The Problem of Other Minds

The third objection is based upon what is called the problem of other minds. Each of us thinks that he knows that there are other persons, beings with minds as well as bodies, beings who perform mental as well as physical acts and who are in both mental and physical states. But if, as dualistic interactionism claims, the mind is completely distinct and different from the body, there is no way to justify the belief that there are other beings with minds; hence there is no way of knowing whether there are other persons. All I perceive whien I see or hear another entity is bodily behavior -- movements and sounds. But bodily behavior is surely not mental. Thus I never perceive another being's mind. Furthermore, there is no way I can check to discover whether, as in my own case, some of this bodily behavior is accompanied by anything mental. It may be, but I have no way to find out. Other beings whom I believe to be persons may be only automata. Dualistic interactionism, by construing minds as radically different from bodies, has forced us to a conclusion contrary to what we all believe. Surely, states this objection, a theory that can avoid this consequence is to be preferred to dualistic interactionism.

The core of this argument can be restated as follows: If the mind-body dualist is correct, then no statements about bodily bahavior entail any statements about minds. Therefore no deductive argument based on what I perceive can be used to justify any of my beliefs that there are other minds, because no premises about what I perceive entail conclusions about other minds. Furthermore, if the dualist is correct then the only case in which I know that mental activity accompanies bodily activity is my own. But no inductive argument based upon such scant evidence is sufficient to justify mv belief that there are other minds. I can justify this belief in only three ways: by deductive inference, bv inductive inference, and noninferentially, by perception. Therefore if the dualist is correct I cannot justify my belief that there are other minds.

We can agree with this third objection that all else being equal any theory which contradicts what we think is true should be discarded in favor of a theory that accords with our beliefs. Thus we must remember this objection when we begin to compare the various mind-body alternatives. However, there are two things about this objection that we should note before we move on. The first is that not everyone who considers the problem of other minds thinks that it is insoluble for a mind-body dualist. Although a discussion of this point belongs more properly in Chapter Two, we can indicate here one attempt to handle this problem. If we agree with the person who is skeptical about our knowledge of other minds that our canons of evidence allow a belief to be justified only by perception, deduction, or induction based on a variety of observations, then we must also agree that if the dualist is correct there is no knowledge of other minds. A. J. Aver, in an attempt to resolve this problem, and also the problem of our knowledge of the past, says:

If it is required of an inductive argument that the generalization to which it leads should be based on a wide variety of experienced instances, both candidates fail the test. One has only a limited experience of the connection of "inner" states with their outer manifestations; and one has no experience at all of the connection of a present with a past event. But these are not ordinary limitations; what is suspect about them is that they are logically necessary. As we have several times remarked, it is by insisting on an impossible standard of perfection that the skeptic makes himself secure.10
Ayer's point here is that the skeptic is demanding that we use canons of evidence so restrictive that it is logically impossible to meet their requirements in these cases. Why should we use those canons the skeptic requires? Why not those that can account for our usual claims to knowledge? Although we do not think that Ayer's quick treatment effectively refutes the skeptic, he has at least suggested the beginnings of a way that may save the dualist from the skeptic.

The second point to note is that this is a telling objection to dualism only if the skeptic is wrong. Perhaps the correct conclusion is that we really do not have knowledge of other minds, perhaps Ayer and others who try to refute the skeptic are the ones who are wrong. Although it is true that all else being equal, we should accept the nonskeptical position, it may be that, as is usually the case, all else is not equal. Possibly we should sacrifice the knowledge claim rather than some other. In other words, the knowledge claim is only one among many other factors that we must weigh in our evaluation of the various mind-body positions. It has no privileged status.


We have examined three philosophical objections to dualistic interactionism, two against interaction and one against dualism. We have found that none of them inflicts irrerjarable damage, although together they cast some doubt upon the position. Let us now turn to three objections based on certain scientific claims.

First Scientific Objection: Interaction Violates Conservation of Energy Principle

The first scientific objection is based upon the principle of the conservation of energy, which states that the amount of energy in a closed physical system remains constant. According to this objection, however, if there is causal interaction between mental events and bodily events, then the principle is violated. When some bodily event causes a mental event, then the physical energy involved in the bodily event is expended in such a way that it is not transferred to anything else; energy is lost. When some mental event causes a bodily event, then the energy gained or lost by the resultant bodily event has not been transferred from or to anything physical so that the total amount of energy is changed. According to this objection, because both minds acting on bodies and bodies acting on minds would violate the principle of the conservation of energy, we have good reasons for concluding that there is no such interaction.

The following example will illustrate this objection. Surely King Canute was absurd in thinking that he could stop the tide merely by willing that it stop; similarly, anyone who tried to start or stop a billiard ball by an act of will would be frustrated. Starting a billiard ball by an act of will would cause the ball to gain kinetic energy, which (because it was not transferred from anything else) would constitute an overall gain in energy. Stopping a billiard ball by an act of will would cause the ball to lose kinetic energy, which (because it was not turned into heat or potential energy, nor transferred to anything else) would constitute an overall loss of energy. According to the first scientific objection, because the only relevant difference between starting or stopping a billiard ball in motion and starting or stopping a brain process is the amount of energy involved, if doing the one violates the principle of the conservation of energy, and is thus physically impossible, then so is doing the other. A converse example illustrates the converse problem. If a rolling billiard ball suddenly stops only because it brought about a mental event, then because the kinetic energy of the ball was neither turned into heat nor into potential energy nor transferred to anything else, physical energy is lost and the principle violated again. This is surely physically impossible. Therefore, according to this objection, because the only relevant difference between a rolling ball causing a mental event and a brain process causing a mental event is the amount of energy lost, if one violates the principle then so does the other.

Both Broad and Ducasse have replied to this objection. Ducasse states his reasons for rejecting it as follows:

(A) One reason is that the conservation which that principle asserts is not something known to be true without exception, but is, as M. T. Keeton has pointed out, only a defining-postulate of the notion of a wholly closed physical world, so that the question whether psycho-physical or physico-psychical causation ever occurs is (but in different words) the question whether the physical world is wholly closed. And the question is not answered by dignifying as a "principle" the assumption that the physical world is wholly closed.

(B) Anyway, as C D. Broad has pointed out, it might be the case that whenever a given amount of energy vanishes from, or emerges in, the physical world at one place, then an equal amount of energy respectively emerges in, or vanishes from, that world at another place.

(C) And thirdly, if "energy" is meant to designate something experimentally measurable, then "energy" is defined in terms of causality, not "causality" in terms of transfer of energy. That is, it is not known that all causation or, in particular, causation as between psychical and physical events, involves transfer of energy.11

We can, I think, quickly show that the first two reasons have little force, but the third is considerably more powerful. It is surely true that in a certain sense the conservation principle is not an empirical scientific law, because it is not a generalization derived from careful observations and experiments. It is, then, unlike Boyle's law, Hooke's law, and others, which is why it is more properly called a scientific principle. Nevertheless no one has ever found reason to reject it, and because it is an essential ingredient in many scientific theories which have great explanatory and predictive power, these theories and thereby the principle are surely justified. Consequently, if as previously claimed, the theory of dualistic interactionism involves a violation of a principle that has been justified, the theory is doubtful and there is reason to reject it. Ducasse's second reason can also be rejected because, like the first reason, it does no more than show that it is possible that something is true, which by itself provides no grounds for claiming that it is true. Just as there is no reason to think that the conservation principle does not hold for the physical universe, so there is no reason to think that by chance or even design the amount of energy of the physical world is kept constant by counterbalancing losses and additions. This may be what happens but it is highly unlikely that the many, many gains and losses of energy that supposedly result from millions of mind-body interactions all balance out evenly. Such an improbable hypothesis cannot carry much weight. Consequently, we can rely only on the third of Ducasse's reasons to save interactionism from the first scientific objection.

Ducasse's third reason is based on an important truth, namely, nothing in the definition of 'causation' entails that all cases of causation involve a transfer of physical energy. It is therefore at least logically possible that some mental events cause bodily events and that some bodily events cause mental events without in any way affecting the amount of energy involved in the bodily events. But can we accept either of these logical possibilities, or is there some reason sufficient for rejecting them? We must consider each one separately because each faces special problems. Is there any reason to reject the claim that a bodily event can cause a mental event without expending energy which is lost to the physical world? If, as envisioned in the billiard-ball analogy, bodily causes must always behave like a rolling ball which loses energy, then we must reject this claim. But physical energy need not be required to bring about a mental event, because mental phenomena involve no physical energy. Thus no energy is transferred from bodily causes to mental effects and thus there is, accordingly, no reason to think that bodily causes of mental events should behave like a ball stopping. Such bodily causes could maintain their total amount of energy or perhaps transfer it to some other bodily event, thereby being the cause of a bodily event and a mental event at the same time. If someone objects that such a dual causation is most mysterious, we can answer by reminding him about the brute, unexplainable nature of immediate causations. We must take them as we find them. We can, then, accept as a plausible position the hypothesis that bodily causation of mental events involves no loss of energy. Thus there is no reason to think that physicopsychical causation involves a violation of the of the conservation principle.

Can we accept or should we reject the claim that mental causation of bodily events does not affect the amount of energy involved in the bodily event? If, once again, we accept the billiard-ball analogy, then we must reject the claim. To start something moving is to give it kinetic energy, and if that is how mental causes affect the body, then mental causation of bodily events violates the conservation principle. Broad, in reply to this objection, counters the billiard-ball analogy with one of his own. He says,

Take the case of a weight swinging at the end of a string hung from a fixed point. The total energy of the weight is the same at all positions in its course. It is thus a conservative system. But at every moment the direction and velocity of the weight's motion are different, and the proportion between its kinetic and its potential energy is constantly changing. These changes are caused by the pull of the string, which acts in a different direction at each different moment. The string makes no difference to the total energy of the weight; but it makes all the difference in the world to the particular way in which the energy is distributed between the potential and the kinetic forms. . . .

Here, then, we have a clear case even in the physical realm where a system is conservative but is continually acted on by something which affects its movement and the distribution of its total energy. Why should not the mind act on the body in this way?12

Broad's analogy brings out the point that there are two quite different ways in which one thing can causally affect the movement of another. Either it can cause it to change its speed, as in the billiard-ball example, or it can causally affect the direction in which the object moves, as in the pendulum example. The first kind of cause changes the total amount of energy involved; the second need not. If Broad's analogy is apt, then various mental events can be said to affect brain processes, not by starting or stopping them, but rather by affecting the course they take. Thus if we assume for purposes of discussion that in each brain only one process occurs at one time and that it is started and stopped by other bodily events, then such a brain process is like a string pendulum which is started and stopped by the expenditure of physical energy. But after something hits the weight and begins its movement, where it goes depends upon the length of the string attached to it. Thus attaching strings of different length to the weight changes the course of the weight but in no way affects the overall amount of energy of the weight. According to this analogy we are to take the causal role of different mental events to be like the causal role of different lengths of string. There would, consequently, be different results in the brain, which would in turn have different bodily results, so that the body would be affected in many different ways given the same input of energy. But while the string analogy solves one problem it raises another. To change the direction of motion without a physical cause is no less a violation of scientific principles than to violate the conservation principle. If mental causes must act like strings to change the direction of motion, then Broad's reply to the first objection is not enough. We must, therefore examine in more detail the way in which, according the Broad, mental events affect brain processes. To do this we must turn to his examination of the second scientific objection.

Second Scientific Objection: No Place for Mental Causes in the Explanation of Human Behavior

Broad states this objection, which he calls the "argument from the structure of the nervous system," as follows:

It is admitted that the mind has nothing to do with causation of purely reflex actions. But the nervous structure and the nervous processes involved in deliberate action do not differ in kind from those involved in reflex action; they differ only in degree of complexity. The variability which characterizes deliberate action is fully explained by the variety of alternative paths and the variable resistances of the synapses. So it is unreasonable to suppose that the mind has any more to do with causing deliberate actions that it has to do with causing reflex actions.13
This argument is based on physiological facts. All human bodily behavior is brought about by neural processes; which behavior occurs depends causally on which nerve fibers are affected and how they are affected. These neural responses in turn depend upon the level of resistances of the various synapses connecting the neurons, or nerve cells, because the path that a nerve impulse will take depends upon the relative resistances of certain synapses. It is also true that the kind of neural processes involved in reflex actions (that is, actions clearly having no mental causes) is no different torn that involved in other kinds of human behavior. It is, consequently, reasonable to suppose that there is no place for immaterial mental causes in any kind of human behavior.

The relevance of this point is direct. If dualistic interactionism is correct, then of course there is a place for immaterial mental causes in the case of human behavior. Thus, it would appear that if the mental causes were removed from any specific causal chain which leads to behavior, then there would be a gap in the chain. That is, at some point there would be a broken place in the chain between two material events -- this would be the gap resulting from the removal of the mental causes -- and the material events on either side of the gap would be unconnected. But, according to the argument under consideration, there is no reason whatever to think there are such gaps. The relevant causal chains are completely filled with neural and other material events. Thus, removal of mental causes would leave no gaps, as it should if dualistic interactionism is correct. Hence, it is concluded that dualistic interactionism is not correct.

There are two possible interpretations of this objection. The first depends upon interpreting 'no place for mental causes' to mean that there is no place within the causal chain for mental causes, and the second depends upon interpreting it to mean that there is no place in the explanation for mental causes. Broad claims that the interactionist is committed to a gap in the explanations of certain human actions if we do not consider mental causes, but he is not committed to a gap within causal chains if mental causes are omitted. Thus, Broad thinks that the interactionist need not worry about the first interpretation just so long as he is careful to specify correctly the way mental causes affect the body. He goes on to suggest what this way is.

[The facts considered in the second objection] suggest that what the mind does to the body in voluntary action, if it does anything, is to lower the resistance of certain synapses and to raise that of others. The result is that the nervous current follows such a course as to produce the particular movement which the mind judges to be appropriate at the time.14
In this passage Broad shows how it can be that there is no gap within the neural causal chain that constitutes a nerve process which must be filled by a mental event. Mental events, according to Broad, would not be parts of such causal chains as M is in Figure 1. They would rather work upon the chains by affecting the distribution of resistance. among certain synapses as in Figure 2.

Thus interactionism is not committed to what there is reason to think is false, namely, that there is a gap between some neural events and others, a gap no neural event fills. In such a manner Broad tells us how the first interpretation can be avoided, and also more carefully specifies the nature of the immediate causal action of mind on body.

We are still left, however, with the second interpretation. In this case the interactionist seems to be committed to a gap, a gap in the explanation of certain human behavior if no mental events are included in the explanation, because, if he is right, it would seem we cannot fully explain why certain synapses have the resistance they do without reference to mental causes. If, therefore, the interactionist is right, then mental causes are essential to explanations of human behavior. This brings us to a crucial point. Are there reasons for accepting or for rejecting the claim that facts about mental phenomena are essential to any full explanation of human action? Broad tries to provide reasons for accepting the need for such mental factors by arguing that

In deliberate action, the response is varied appropriately to meet the special circumstances which are supposed to exist at the time, or are expected to arise later; whilst reflex action is not varied in this way, but is blind and almost mechanical. The complexity of the nervous system explains the possibility of variation; it does not in the least explain why the alternative which actually takes place should as a rule be appropriate and not merely haphazard. And so again it seems as if some factor were in operation in deliberate action which is not present in reflex action; and it is reasonable to suppose that this factor is the volition in the mind.15
Broad bases his argument on the following two premises: (1) With so many responses possible because of the great complexity of the nervous system, we must explain why so often only the appropriate responses occur; and (2) it is reasonable to explain this appropriateness by the effect of mental causes upon the appropriate nerve synapses.

Although we think that this may be the best argument for accepting Broad's claim, there are several reasons for rejecting it. Many human actions are not merely reflex actions and do not seem to involve mental causes., but are usually appropriate. During our waking hours much of what we do, such as our habitual responses and absent-minded behavior, seems to be done without thought or decision or volition or any other mental cause. Yet these actions are not reflex actions. Very often they are appropriate to the situation, and this appropriateness requires an explanation. But it certainly seems no matter how we explain the appropriateness of these actions, we shall not include a mental factor. And if we can explain these actions without reference to mental causes, there is no reason to think that a mental factor is needed to help explain fully any other human actions, even those we would call deliberative actions.

Broad would probably reply at this point that habitual actions are those in which the synapses have acquired regular resistances as the result of continuous causal action by the mind, but there is no reason to think such habitual acquiring of synapse-response requires previous mental causation. The best illustration that mental factors are not needed either for "learning" or for the one appropriate response out of many possible responses is afforded by the complex computers we find operating today. Not only do some of these machines have a huge number of possible responses available from which they usually "choose" the appropriate one, but they also are capable of improving in their responses. They can "learn" in playing chess and thus improve their game. All this requires explanation, but surely no part of the explanation of machine behavior requires a mental causal factor. Consequently, there seems to be no reason to think that mental causal factors are needed to explain certain human actions. There seems to be no place in the explanation of human actions that requires mental factors. And, because it seems that dualistic interactionism requires that there be such a gap, we have found a strong objection to dualistic interactionism, an objection that may tip the scale of evidence in favor of some other position.

Broad has not avoided the second scientific objection. How has he fared with the first? We have seen that he can give an explanation of how mental events can affect neural processes without violating the conservation principle, but we have not investigated how he might answer the change-of-direction objection. The answer is that, strictly speaking, mental events do not cause changes in the direction of neural currents, because what they immediately bring about are changes in the distribution of resistances rather than changes in the direction nerve currents take. It is the relative levels of resistances that cause the current to take a certain direction. This is surely understandable. If you ask, however, how mental events can affect the relative distribution of resistances, the answer is that once again we have a case of immediate, and therefore unexplainable, causation. This, we find, is the best reply a dualistic interactionist can give to the objection. It is surely not completely satisfactory and does not fully dispel the mystery of how mental events can affect the body. Yet it does at least neutralize the damage of this objection to a great extent. Nevertheless, it may well be that some other theory will be able to handle this problem more easily.

Third Scientific Objection: What Evolves from Material Phenomena Is Material

The third objection based upon scientific premises derives from the theory of evolution. It begins by noting that according to the accepted doctrine of evolution human beings have evolved over a long period of time from other less complex forms of life; that is, human beings' ancestry can be traced back through their primitive ancestors, to apes, to certain forms of sea life, and finally to single-cell living beings which themselves resulted from certain material forces operating at certain places in the universe. Human beings, then, are complex beings who have evolved from primitive forms of life. And these primitive forms of life in turn resulted from physical and chemical reactions among nonliving things. Human beings ultimately, then, have evolved from nonliving, completely material things. One possible theory is that everything in this universe can trace its ancestry back to simple hydrogen atoms which under different conditions of temperature and pressure resulted in heavier and more complex atoms and molecules, some of which became the basis for life on this planet. Human beings, then, have evolved from simple material particles by means of a continuing process of increasing complexity. Consequently, according to this objection, human beings are no different in kind from any other material objects. He may be more complex than most material objects and be made up of unique kinds of molecules, but he evolved from the same basic particles as did the trees, flowers, flies, amoebas, and viruses. We should conclude, therefore, that just as these other things are merely material objects and do not have minds, so also human beings are merely material objects with no mind

This objection to dualism has two forms. The first states that because humans have evolved from primitive particles which were material only and had no minds, humans themselves have no minds. The second form of the objection states that because humans have evolved from the same primitive particles as all material objects that do not have minds, humans themselves have no minds. Neither form is cogent. The first is based on the premise that only material objects can evolve from material objects. That is, material processes can causally bring about only other material processes. But as we have already seen when we examined the second philosophical objection, there is no reason to think that certain material processes cannot causally bring about mental events as well as material events. As Ducasse pointed out, we must always examine a particular event to see just what causal results it produces. There is no reason to think that the causal results of certain material events are never mental events. Thus it may well be that somewhere along the path of evolution something material evolved which had mental events among its causal results. Consequently, the fact that humans evolved from matter casts no doubt on a mind-body dualism.

The second form of this objection can be rebutted in a similar way. It is not at all strange that everything that evolved from primitive matter except sentient beings is itself merely matter. Sentient beings are quite different from other material objects. It is true that if all we knew about humans were that they evolved from the same things that all objects without minds evolved from, then we would have some reason to think that humans are merely material. But all of us have other knowledge of humans, especially about their abilities and, at least in our own cases, about events and states that seem quite different from material phenomena. Thus the second form of the third scientific objection, like the first form, provides little reason to reject a mind-body dualism. Humans with immaterial minds as well as bodies might well have evolved in their own unique way from matter. The theory that describes the path of evolution does not cast any doubt on mind evolving from matter.

We have examined six objections to dualistic interactionism, four against interaction, and two against dualism. Neither of the objections to dualism, the problem of other minds or the objection from evolution, is strong enough to overcome what seems to be true, that is, people have both minds and bodies, and that bodies and minds are quite different. Two of the objections against interactionism -- the objection that questions where interactions take place and the objection from the principle of the conservation of energy -- have been either dismissed or neutralized to some degree. The other two objections to interaction, however -- the philosophical objection from the lack of factors in both mental and material phenomena relevant to the one causally affecting the other, and the scientific objection from the lack of a gap in the physiological explanation of human behavior -- cast some doubt upon dualistic interactionism. We are not, then, justified in accepting this position until we have examined alternative positions to find out whether any are less doubtful than dualistic interactionism. And because two of the strongest objections are directed at the causal interaction between minds and bodies, one obvious candidate for a less doubtful theory is a dualism that avoids interaction. This, in essence, is the position of parallelism.


Parallelism is one form of mind-body dualism. Like interactionism it claims that a person has both an immaterial mind and a body, that he consists of mental and bodily events and processes, and that mental and material phenomena are radically different. It differs from interactionism, however, in that it denies that there is any causal interaction between minds and bodies. Mental events proceed over a temporal period, some causing others, but none causally affecting any material events. Similarly, material events occur at different places and times, some causing others, but none causally affecting any mental events. The two different kinds of events proceed completely independently of each other. In the case of an individual person it is granted that certain bodily events, such as breaking an arm, regularly precede certain mental events, such as having pain, and that certain mental events, such as deciding, regularly precede certain bodily behavior, such as moving the pawn instead of the bishop. But it is claimed that in such cases there is no causal interaction at all. Having an arm broken does not cause pain, and deciding to move a pawn does not cause someone to move it. Such events merely parallel each other, in the sense that certain mental events are accompanied by certain bodily events and certain bodily events are accompanied by certain mental events. Parallelism, therefore, escapes the two objections which we found cast doubt on dualistic interactionism. Can we therefore conclude from this that we should choose parallelism over interactionism? Not yet at least, because there is an objection to parallelism which does not face interactionism. If it is damaging, then we may have to reject parallelism as inferior to interactionism.

An Objection to Parallelism: Cannot Explain Observed Regularities

If parallelism is correct and mental events and material events proceed completely independently of each other, then there is no reason why there are regular relationships between certain of them. There is no reason why what follows the breaking of an arm should not be pain one time and joy another time. We can understand why the breaking of an arm should be followed by pain if bone breaks cause pain, but such regularity where there is no causal relationship calls for some kind of explanation. It seems unlikely that such regularities of parallel mental and bodily occurrences would happen merely by chance. Consequently, such regularities must be explained, but how can parallelism explain them? It cannot rely on the usual kind of causal explanation, the kind that interactionism uses, and no other kind of explanation seems available. This objection, then, is that parallelism, unlike interactionism, cannot adequately explain what requires explanation and, consequently, it should be rejected in favor of some other theory, such as interactionism, which can provide the relevant explanations. Parallelists have answered this objection in the past in two different ways. It will be our task to see if either answer is adequate. Historically, the two different kinds of explanations of mind-body regularities offered by parallelists have been based either on the theory of occasionalism or on the preestablished harmony theory. Let us consider each.

One Reply: Occasionalism

Occasionalism, propounded by the Catholic philosopher Malebranche, is the theory that on the occasion that certain bodily events occur, God, who can do all things possible, causes certain mental events, and on the occasion that certain mental events occur God causes certain bodily events. Thus although there is no causal action between minds and bodies, we can explain the regularity among certain mental and physical events by stating that God, who has a most orderly and powerful mind, constantly causes the same kind of mental event each time a certain kind of bodily event occurs, and the same kind of bodily event each time a certain kind of mental event occurs.

A Second Reply: The Preestablished Harmony Theory

The preestablished harmony theory, as proposed by Leibniz, claims that the procession of bodily events and the procession of mental events both proceed according to a preestablished plan, presumably God's. Thus, which material event follows a certain material event is predetermined, and which mental event follows a certain mental event is predetermined. In addition, there is a predetermined harmony between these two independent series of events. That is, the two independent series are so arranged that certain events in the material series are always accompanied by certain events in the mental series, and vice versa. This situation has been likened to two clocks, one of which has a face and hands but no bells to toll the hours, and one of which has bells but no face or hands. If someone were to observe that each time the hands on the one clock were in one position the other clock struck once, and when the hands were in a different position the second clock struck twice, and so on, he might conclude that there is some causal connection between the two clocks, that is, one causes something to happen in the other. But if he examined the situation more carefully he would realize that there is no causal connection between the two clocks at all. It is just that some being regulated each one and then set them running in such a way that whenever the hands of one were in a certain position the other happened to strike its bell a certain number of times. These two clocks run parallel to each other and exhibit a joint regularity or harmony which results not from the causal effects of one clock on the other at certain times and not from the continual intervention of some outside causal force, but rather from the causal effect of some being who at some previous time set each clock independently so that each would run in a certain way. Now, says Leibniz:

put the soul and body in the place of these two timepieces. Then their agreement or sympathy will also come about in one of these three ways. The way of influence [interactionism] is that of the common philosophy. But since it is impossible to conceive of material particles or of species or immaterial qualities which can pass from one of these substances into the other, this view must be rejected. The way of assistance [occasionalism] is that of the system of occasional causes. But I hold that God should help only in the way in which he concurs in all other natural things. Thus there remains only my hypothesis, that is, the way of preestablished harmony, according to which God has made each of the two substances from the beginning in such a way that, though each follows only its own laws which it has received with its being, each agrees throughout with the other, entirely as if they were mutually influenced or as if God were always putting forth his hand, beyond his general concurrence.16
The two parallelist positions have one thing in common: They both postulate the existence of some unobservable entity -- which they call God -- in order to explain certain observed mind-body regularities. Such an entity is called a theoretical entity because it is an unobservable entity postulated as part of a theory designed to explain certain observed phenomena. Leibniz justifies his particular postulation in two steps. First, he claims that it is necessary to postulate something or other because mind-body regularities cannot be explained as the result of mind-body causal interaction, and where postulation is necessary for explanation it surely is justified. Second, he justifies his own particular postulation as preferable to that of Malebranche's on the grounds that Malebranche's hypothesis requires more action by the postulated entity than is necessary. Surely we should postulate nothing more than is necessary to explain what is observed. And because mind-body regularities can be explained by postulating God but not postulating his continual intervention at each instance of mind-body regularity, Leibniz is justified in rejecting Malebranche's theory as inferior to his own.

Objection to Both Theories: They Postulate a Deus ex Machina

The principle used to reject occasionalism is that if an explanation can be given without postulating something, then the postulation should not be made. How does this principle apply to Leibniz' own version of parallelism? If, as he claims, it is impossible that minds and bodies interact, then some postulation is necessary for explanation and therefore is justified. But although we have seen that mind-body interaction may be quite mysterious and even unlikely, we have found no reason to think it impossible. Thus it is not necessary to postulate the causal action of an unobservable entity to explain mind-body regularities, and Leibniz' reasoning against interactionism and for a postulated cause fails.

Can we now reject Leibniz' theory -- and with it occasionalism and, therefore, parallelism -- or is there some other way to justify postulating a theoretical entity which might apply in this case? There is one. If it can be shown that by a particular postulation we can not only explain the phenomena requiring explanation, but can also correctly predict facts that otherwise would have gone undiscovered, then we can justify accepting the postulation on the ground of its fruitfulness in increasing knowledge. Such predictive power is important in another way because it enables the hypothesis postulating the theoretical entity to be tested by observation and experimentation, and thereby confirmed or disconfirmed. Such testability is essential for a hypothesis to be scientific. But when a hypothesis lacks testability and predictive power and is not needed to explain anything, then it clearly should he rejected. It would be merely an ad hoc hypothesis, and any entity it postulates to explain something would be what Leibniz calls a deus ex machina, that is, a theoretical entity the sole use of which is to enable its theory to explain what the theory otherwise could not explain.

Is Leibniz' hypothesis of preestablished harmony ad hoc and therefore can his claim that occasionalism requires a deus ex machina be turned against his own theory? Leibniz' hypothesis about God as the cause of mind-body regularities would have predictive power only if we could read God's mind and discover which of the kinds of mind-body regularities not yet observed he will bring about in the future. But such mind reading is beyond our ability. Consequently, the hypothesis has no predictive power and therefore is not testable by observation and experimentation. It is indeed an ad hoc hypothesis, and its postulated entity is deus ex machina. It should be rejected in favor of interactionsim, although this theory faces problems of its own. This is especially true where, in spite of the difficulties, it does seem to be the case that mental events and bodily events do causally interact. We should not reject a theory that is in accordance with the way things seem for a second competing theory having in its favor only that it can avoid certain difficulties that face the first theory. Consequently, we can reject parallelsim, whether based on the preestablished harmony theory or on occasionalism, as a candidate to replace dualistic interactionism as the most plausible mind-body theory. Parallelism as an attempt to avoid the difficulties of interactionism goes too far in claiming that mind and body are completely independent.


If we review the objections to dualistic interactionism we can recall that one of the most forceful, objections to mind-body causal interaction is the one based on the lack of a gap in the physiological explanation of behavior. We found that although this objection casts some doubt on the existence of psychophysical causation, that is, the causation of a material event by a mental event, it has no force when applied to physicopsychic causation. Thus we have found no reason to doubt that certain material events can cause mental events, and we have rebutted the objection from evolution on these grounds. Consequently, this objection to mind-body interaction can be avoided without going to the extreme of parallelism. All that we need to deny is that mental events causally affect bodily events. This leads us to epiphenomenalism, a view stated by Thomas Huxley, who claims

All states of consciousness in us, as in [brutes], are immediately caused by molecular changes of the brain-substance. It seems to me that in men, as in brutes, there is no proof that any state of consciousness is the cause of change in the motion of the matter of the organism. If these positions are well based, it follows that our mental conditions are simply the symbols in consciousness of the changes which take place automatically in the organism; and that, to take an extreme illustration, the feeling we call volition is not the cause of a voluntary act, but the symbol of that state of the brain which is the immediate cause of that act. We are conscious automata. . . .17
We can see from this quotation that epiphenomenalism, like interactionism and parallelism, is a mind-body dualism. Humans (and according to Huxley, even some brute animals) are conscious beings. That is, certain mental events occur to humans. In addition, of course, humans have bodies. Where epiphenomenalism differs from the other two dualistic theories is in its view of the relationship between the mind and body. According to the epiphenomenalist a mental event is merely an epiphenomenon, or in other words a by-product of certain material processes. When these material processes occur, they both cause other material processes and produce by-products, which themselves have no effect on anything else at all. Santayana has likened the relationship between bodily events and mental events to the relationship between a mountain stream running over and around rocks and into pools, and the babbling sound produced by the flowing water. The babbling sound is caused as a by-product of the water flowing around the rocks. It does not affect the course of the water, which speeds on its way affected only by the rocks and other objects in its path. Neither does the babbling by-product at any one moment affect the sound which results at any later moment. Each moment's sound is caused by the action of rocks and water, only to die out without a single effect of its own. Similarly, each mental event is the causal by-product of some material event in the uninterrupted series of material events. Each mental event is produced, occurs, and ends without causally affecting anything at all.

Epiphenomenalism is attractive for several reasons. One reason, which is probably what attracted Huxley, is that it fits in nicely with the theory of evolution. As more and more complicated physical processes evolve, it is not hard to conceive of consciousness evolving as a by-product which does not causally affect the basic evolving material processes. Second, because it claims that only material events are causally efficacious, epiphenomenalism avoids the problem of a gap in the physiological explanation of human behavior which faces interactionism. Third, epiphenomenalism is also attractive to many people who greatly value scientific controllability. If epiphenomenalism is correct, we do not need to know anything about mental events to be able to explain, predict, and control human behavior, because mental events would play no role in causally determining behavior. As a consequence of this, no hidden mental factors are necessary for accurate predictions. A fourth reason is that, unlike parallelism, epiphenomenalism requires no deus er machina to explain mind-body regularities because it claims that each mental event is the causal by-product of a certain material event. Epiphenomenalism, therefore, avoids the most crucial objections to its two rival dualistic theories. It does, however, share one objection with both of these theories (that is, the objection to dualistic theories which derives from the problem of other minds) and one with interactionism alone (that is, the objection derived from the apparent lack of factors in material phenomena relevant to causing mental events). However, because we found neither of these objections to be very damaging, it may well be that we should accept epiphenomenalism unless it faces crucial objections that we have not yet examined. Let us, hence, turn to an examination of the three strongest objections that have been raised against epiphenomenalism.

First Objection to Epiphenomenalism: Rejects the Effects of People's Minds on the Course of Events

If epiphenomenalism is true, then no mental phenomena have any causal effect upon the history of humanity. Thus none of people's hopes, desires, dreams, joys, or sorrows have in any way affected the course of human events. Nor is it correct to talk of psychosomatic illnesses, or to claim that any psychological disturbances affect human behavior. We should not explain someone's behavior by referring to his neurosis, or psychosis,. Indeed, according to this objection, if epiphenomenalism is true, the entire course of human history would have been exactly the same if human beings felt no joys or sorrows, had no hopes or fears, and sought no goals. But surely this is an absurd conclusion. Human hopes, fears, aspirations, and the like are intimately connected to the course of human events. Epiphenomenalism should, therefore, be rejected.

There are actually two different attacks on epiphenomenalism embodied in this one objection, one with some force and one quite mistaken. The first claims that it surely seems that the mental side of human beings has played a causal role in the lives of human beings. This, as we have seen, lies behind much of the initial plausibility of interactionism, and also counts against parallelism. We should, therefore, also count it against epiphenomenalism, although we should also remember that there may be overriding reasons for accepting epiphenomenalism. The second attack goes beyond the first and claims that if epiphenomenalism is true then the mental side of human beings is irrelevant to the course of human events. Although such a charge might be leveled at a parallelist who rejected both preestablished harmony and occasionalism, it is quite off the mark when applied to epiphenomenalism. The mistake in this claim is that from the fact that A does not cause B, it is inferred that A is in no way relevant to whether or not B occurs. But this is a fallacious inference because if B is the cause of A, then B occurs only if A occurs. Therefore if A were not to occur, then B would not occur and the whole course of things could be changed. For example, assume that a certain brain process causes someone to pull the trigger of a gun and also has the causal by-product of a desire to kill someone. Thus, if the assassin of President Kennedy had not had that desire, then neither the brain process which caused it, nor the pulling of the trigger which also resulted from the brain process would have occurred. In such ways the mental side of human nature is related to what happens even if epiphenomenalism is true. Thus we can reject the second prong of the first objection to epiphenomenalism while remembering the first.

Second Objection to Epiphenomenalism: Theory Makes Its Own Justification Impossible

The second objection has been stated by J. B. Pratt, who says,

To say that a thought is even in a minute degree a co-cause of the following thought would be to wreck [epiphenomenalism]. In the process known as reasoning, therefore, it is a mistake to suppose that consciousness of logical relations has anything whatever to do with the result. . . . We may happen to think logically; but if we do, this is not because logic had anything to do with our conclusion, but because the brain molecules shake down, so to speak, in a lucky fashion. It is plain, therefore, that no conclusion that we men can reach can ever claim to be based on logic. It is forever impossible to demonstrate that any thesis is logically necessary.18
From this Pratt further concludes that the epiphenomenalist is in a hopeless position because he wants to assert that he can prove his own theory, but his own theory implies that proofs are impossible.

This is quite a popular objection which is usually raised against the determinist rather than the epiphenomenalist. However, because epiphenomenalism is committed to the claim that all mental phenomena have causes, the objection applies equally well or, more accurately, equally badly, to it, for this is a totally misguided objection to both views. Let us grant for the purposes of the discussion that every event, whether material or mental, is causally determined. Thus, each time I come to some conclusion I have been caused to do so by certain preceding events. Does it follow from this, first, that my conclusion has not been proved and, second, that I have not proved it? First, a conclusion is proved deductively, for example, when it is shown to follow deductively from true premises. It does not matter how it is shown or by whom or under what conditions. A computing machine can be used to derive certain conclusions, but this does not show that the conclusion has not been proved. A justification of proof of a claim depends on logical relations among statements, not on psychological and causal relations among thoughts or molecules. Thus. because epiphenomenalism makes claims about causal rather than logical relations, it does not imply that conclusions cannot be proved.

Second, epiphenomenalism does not imply that humans cannot prove conclusions even if we assume that to prove a conclusion is to proceed through certain steps of one's own free will, because epiphenomenalism does not deny either that humans are able to proceed through such steps or that they have free will. Epiphenomenalism does imply that I am caused to proceed through the steps of a proof, but this does not imply that I do not do it of my own free will. It is true that if causal determinism and free will are incompatible, and if I am caused to do something, then I do not do it freely. However, although epiphenomenalism implies mental determinism, it does not imply that this is incompatible with free will. Furthermore, as we have seen previously in Chapter 3, there is reason to deny the thesis of incompatibility. We can, therefore, reject Pratt's objection to epiphenomenalism. The theory does not imply that no theory can be proved, so the epiphenomenalist can consistently claim that his theory is provable and that he can prove it. Whether he has proved it, however, is yet to be decided.

Third Objection to Epiphenomenalism: Requires Nomological Danglers

Herbert Feigl has stated the third objection to epiphenomenalism. He attempts to evaluate competing mind-body theories by giving them a comparative ranking much as we are doing. Feigl first ranks epiphenomenalism over interactionism, but then he rejects it for still another theory. He justifies his rejection of interactionism by claiming that it is inconsistent with a basic goal of science. According to Feigl, science should strive to reach the point wiaere all behavior, human and nonhuman, can be explained and predicted by the physical sciences and the relevant publicly observable behavior. Consequently, he tninks that epiphenomenalism is preferable to interactionism, which requires private, that is, unobservable causes, and thus is incompatible with this goal of science.

Feigl rejects epiphenomenalism because he thinks it requires us to interpret certain scientific laws in a very peculiar way. He says,

It accepts two fundamentally different sorts of laws -- the usual causal laws and laws of psychophysiological correspondence. The physical (causal) laws connect the events in the physical world in the manner of a complex network, while the correspondence laws involve relations of physical events with purely mental "danglers" These correspondence laws are peculiar in that they may be said to postulate "effects" (mental states as dependent variables) which by themselves do not function, or at least do not seem to be needed, as "causes" (independent variables) for any observable behavior.19
Feigl's objection to epiphenomenalism is that it requires that there be two quite different kinds of causal laws. Usually causal laws are laws expressing causal connections between events each of which is part of the continuing series of causes and effects that causally determines what occurs at each moment. Thus the usual causal laws relate events which, although caused, are themselves causal factors determining what occurs after them. However, if epiphenomenalism were true, then psychophysical laws -- that is, laws relating mental and physical events -- would be quite different. They would be laws expressing a causal relationship between physical events in causal chains and mental events that are neither part of a causal chain nor causally affect some chain. These mental events would be what Feigl has called "nomological danglers," -- that is, factors which, although integral components of certain laws, dangle uselessly because they are unnecessary for the explanation and prediction of human behavior. Feigl thinks that any theory that requires laws involving nomological danglers is inferior to a theory that requires only the usual kind of law. Consequently, although he thinks that epiphenomenalism is preferable to interactionism, which requires nonobservable causes, he also thinks that a theory that also does not require nonobservable effects would, in turn, be preferred to epiphenomenalism. As we shall see when we examine the double-language theory, Feigl thinks that he has found such a theory.

There are two things we can say about Feigl's argument. First, he rejects interactionism for reasons similar to those expressed in the second scientific objection. We saw that Broad does admit that interactionism leaves a gap in the physiological explanation of human behavior, but we have not yet decided how damaging this problem is. Second, although we can agree that if there were no other grounds available for choosing between two theories -- if all else were equal -- then we should accept the one that does not require nomological danglers, Feigl's objection by itself does not seem particularly forceful. Although it shows that psychophysical laws would be unique if epiphenomenalism were true, it does not show that epiphenomenalism requires anything different of any scientific procedures of observation and experimentation. It has consequences only for how we interpret the laws based on what is observed. Such an objection then is surely not fatal and not even terribly damaging.


We have already rejected one dualistic theory, namely parallelism. Of the two remaining dualistic theories, dualistic interactionism and epiphenomenalism, little more remains to be said beyond explicit comparison of the two. When we weigh the various objections to each, we find that both share two objections.

  1. The objection from the problem of other minds.
  2. The objection from the apparent lack of characteristics relevant to causal interaction.

We have also found that epiphenomenalism faces two objections which interactionism avoids:

  1. The objection that epiphenomenalism denies what seems to be true, that mental events have causal efficacy.
  2. The objection that epiphenomenalism requires nomological danglers.

And we have found two objections to interactionism avoided by epiphenomenalism:

  1. The objection that interactionism requires something contrary to empirical evidence, namely, that there is a gap in a purely physiological explanation of human behavior.
  2. The objection that interactionism requires that mental events causally affect the body in a way that either is inexplicably mysterious or violates a scientific principle.

How are we to evaluate the relative strength of these objections, and consequently how are we to decide between the two theories? It surely appears that objection (5) is the most serious because it accuses interactionism of requiring something that conflicts with empirical evidence. It would seem, then, that the seriousness of (5) outweighs that of (4), and perhaps we should discount the seeming efficacy of mental events and choose epiphenomenalism over interactionism. However, before making this decision let us look once again at objection (5), because it has become crucial.

We have been accepting that what Broad says about the place of mental events in the explanation of human behavior correctly states what is required by interactionism. We have been assuming that one of the most likely ways that mental events affect the body is by varying the resistance of certain nerve synapses in the brain and thereby changing the paths of certain nerve impulses. It seems obvious, then, that we must include something like the effect of the mental events on the resistance of synapses if we are to explain certain human behavior. Thus objection (5) seems cogent. Nevertheless it may be that interactionism can avoid it. It is true that if mental events do causally affect the brain, then a complete explanation must include mental causes. But it is not clear that an explanation adequate for all the needs of the physiologist must be a complete explanation.

Let us assume that every event, whether material or mental, has a cause. Given this, it is possible that a certain kind of brain event, call it B, is always followed by a certain kind of nerve impulse, call it N, and also a certain kind of mental event, M. Assume also that the paths ot nerve impulses depend on the relative resistances of synapses and the mental events can causally affect these resistances. Given all this, then we can see how M could be caused by B and how M could, in turn, causally affect the path of N by causally affecting the resistance of certain synapses, as in Figure 3. The consequence of this is that, given the occurrence of B, N always is caused to take a certain path because of the effect of B on M and M on the synapses. From what the neurophysiologist could observe, however, there would seem to be no need of a mental cause in his explanation of the neural events. It would seem that B alone caused N to take a certain path, as in Figure 4.

A neurophysiologist might even take this to be a case of proximate causation and thus consider it to be a brute fact not itself explainable. At any rate he could explain and predict all human behavior to which B, N, and M are causally relevant without any need of a mental cause. Thus his explanation is both purely physiological and scientifically adequate. But in an important sense it is not complete because it omits one causal factor, mental event M. Consequently, dualistic interactionism is, contrary to objection (5), consistent with the observed evidence that no mental causes are needed for explanations of human behavior that meet all the requirements of the physiologist.

Interactionism, then, is consistent with there being no gap in physiological explanations, and it is also consistent with there being gaps that require mental causes. This distinguishes it from many competing theories. Therefore instead of the lack of an observed gap counting against interactionism, the possibility of mental causes would count in favor of interactionism if further examination gives evidence of a gap requiring mental causes. Where people who raise objection (5) may have gone wrong is in thinking that interactionism implies that causal determinism does not apply to mental events, so that there would be no way of establishing how M would affect synapses. Thus, given only B and N, there would be no way of knowing which path N would take. But interactionism is consistent with complete causal determinism.

Let us return to our comparative evaluation of interactionism and epiphenomenalism now that we have rejected objection (5) to interactionism. Although interactionism, unlike epiphenomenalism, requires no nomological danglers and can accommodate the plausible belief that mental events causally affect the body, it avoids these problems only by requiring an inexplicable kind of effect of the mental on the physical. Neither theory, then, is completely satisfactory. But is one more reasonable than the other? We have agreed that the interactionist can at least neutralize the damage of objection (6) by claiming what is inexplicable is the immediate effect of mental events on the brain, and no cases of immediate causation are explicable. On the basis of this we can conclude that interactionism faces less serious objections than epiphenomenalism, and so interactionism is the more reasonable of the two. We can further conclude that interactionism is the most plausible dualism, because we have previously rejected both versions of parallelism. Nevertheless, it still faces objections which some different theory may be able to avoid. If we are to find such a theory, we must turn to monistic theories, the most well known of which is reductive materialism.


Materialism is generally considered to be the chief opponent of dualistic interactionism. It is the theory that whatever exists is material and that what is taken to be mental, and thus immaterial, either does not exist or is really identical with something material. The classical exposition of this theory occurs in the philosophy of Hobbes, although Hobbes, like many other materialists as we shall see, has trouble being completely consistent. At the center of Hobbes' materialism is his conception of sense, which he claims is the source of all human beings' thoughts, imaginings, dreams, and remembrances, "for there is no conception in a man's mind, which hath not at first, totally or by parts, been begotten upon the organs of sense. The rest are derived from that original."20 His materialism becomes clear when he says that sense is "some internal motion in the sentient, generated by some internal motion, of the parts of the object, and propagated through all the media to the innermost part of the organ."21 Thus for Hobbes all that exists is either a material object or some physical event consisting of some material objects in motion. Certain of these physical motions are what constitute sense and, consequently, the whole realm of the mental. Hobbes, then, does not deny the existence of mental phenomena. Rather, he appears to be reducing them to motion and thus to material phenomena.

Because of his reduction of the mental to physical motion, Hobbes can go beyond his claim of materialism to an assertion of mechanism. In his introduction to the Leviathan he says:

For seeing life is but a motion of limbs, the beginning whereof is in some principle part within; why may we not say, that all automata (engines that move themselves by springs and wheels as doth a watch) have an artificial life? For what is the heart, but a smine: and the nerves but so many strings, and the joints but so many wheels, giving motion to the whole body such as was intended by the artificer?22
On this view, living things, including human beings, are no different from nonliving things. They are in principle just like a machine such as a watch, although much more complicated. We can explain and predict all the motions of machines and their parts by applying the laws of mechanics to our knowledge of the spatial locations and masses of the relevant material objects and the forces acting upon them. By similar uses of these laws we can, according to Hobbes, explain all the behavior of living things. According to Hobbes, then, everything is some kind of material object, and the science of mechanics is sufficient to explain and predict the behavior of everything, living and nonliving. Hobbes, then, is not only a materialist but also a mechanist. However, Hobbes' mechanism is not essential to his materialism, because materialism does not imply mechanism. It is possible that everything is material and some events happen by chance and, consequently, are neither explainable nor predictable by the science of mechanics. Because we are here interested only in materialism we need not consider mechanism further.

It follows from Hobbes' materialistic solution to the mind-body problem that the science of psychology is reducible to, or replaceable by physics though the converse is not true -- that is, if psychology is reducible to physics, it does not follow that living hings are no different in principle from nonliving things. What follows is only that the data of psychology are no different from the data of physics. For example, if it is claimed that the data of psychology are only behavior, that is, overt motions and sounds of human bodies, then psychology might well be reducible to physics, in the sense that we could explain and predict with physical laws all the behavior that we could explain and predict with psychological laws.

One view clearly consistent with this view of psychology as behavioristic but inconsistent with materialism is epiphenomenalism, which, as we have just seen, states that whereas certain material processes cause and indeed give rise to mental states and events, these states and events have no effect on any material processes or even on other mental processes. Consequently, if epiphenomenalism is true, then a behavioristic psychology is sufficient to explain and predict all human behavior, but materialism is false. Other vews consistent with a behavioristic psychology but not with materialism are a neutral identity theory which will be discussed later in Chapter 4, and parallelism, which, although dualistic, denies mind-body interaction and thereby is compatible with a completely physicalistic explanation of human behavior.

Nevertheless, although there are good reasons to class Hobbes as a materialist, there are passages in his writings where he sounds more like a dualist of the epiphenomenalist variety. This is apparent when he says that "sense, in all cases, is nothing else but original fancy, caused, as I have said, by the pressure, that is by the motion, of external things upon our eyes, ears, and other organs there unto ordained."23 Sense, then, is fancy, and fancy is, according to Hobbes, the appearance of motion rather than motion itself, contrary to Hobbes' previous characterization of sense as motion. But if sense is appearance, then it would seem that there are not only material objects in motion or at rest, but also appearances that are quite different. Hobbes is thus faced with the central problem tor materialists: how to incorporate into their theory that which seems to be completely alien to it, that is, appearances such as hallucinations, dreams, and mental images, and such other phenomena as sensations, emotions, and thoughts.

At the beginning of our discussion of materialism it was stated that a materialist could try to handle mental phenomena in either of two ways. He might admit, as did Hobbes, that there are mental phenomena, such as sensations, but claim that they are really reducible to something material. If he were to do this, he would be what we will call a "reductive" materialist. However, he might instead deny that there are any mental entities at all. If he were to do this, he would be what we will call an "eliminative" materialist. On the face of it, neither approach looks very promising. After all, we noted in the beginning of this chapter, mental phenomena certainly seem to be radically different from material phenomena, so radically different, in fact, that dualism seems initially very plausible. Thus, in saying that all mental phenomena are really material phenomena, the reductive materialist seems plainly to be saying that some things which are radically different from material phenomena are really material phenomena after all. This claim seems to be incoherent. The second way seems to be little better, however, because if he denies that there are mental phenomena he seems clearly to be denying that there are thoughts, feelings, desires, hopes, dreams, and even pains. But surely any theory that denies that existence of such phenomena is false, because if there is anything each of us can be certain about, it is that we are aware of feelings, desires, pains and the like. Furthermore, because just being aware or conscious is a mental state, we would have to deny that anyone is conscious or aware of things. But as Descartes pointed out, although each of us can doubt the existence of almost everything including other minds, he cannot doubt that he is doubting, and if he is doubting, he is conscious. Thus that at least one being is in the mental state of consciousness seems to be an undeniable fact. If a theory implies the contrary, then we have good reason for rejecting the theory. Consequently, the second way the materialist could try to handle mental phenomena seems to be no more helpful than the first.

One way to help indicate the prospects for materialist theories is by contrasting them with alternatives. We can do this with the following chart:

  Dualism (all types) Materialism
Mental objects such as minds Purely immaterial entities such as souls; wholly lacking in material properties. Either: (1) purely material entities such as brains;
or (2) completely denies that there are any minds.
Other mental objects, such as sensations Purely immaterial entities, wholly lacking in material properties. Either: (1) purely material entities such as brain parts, or perhaps bodily movements;
or (2) denies that there are any such mental objects.
Mental events Events, such as dreams, which consist wholly of immaterial mental objects. Either: (1) Events, such as dreams, which consist wholly of material objects such as brain parts;
or (2) completely denies that there are such mental events.
Mental states, such as a psychosis A condition or situation of an immaterial mental object. Either: (1) a condition or situation of some material object such as a brain or a human body;
or (2) completely denies that there are any such states.

There are a few things to notice about this chart. First, we see that neither dualism nor materialism simply repeats the definitions of terms with which we began the chapter. Thus, both types of theory go beyond the minimal definitions of terms such as 'mental object,' 'mental event ' and, 'mental state.' Second, note how the initial implausibility of eliminative materialism shows up: The various types of mental entities we typically think exist are rejected outright. This comes in clause (2) of each of the entries on the materialist side of the chart. It must also be noted that when the eliminative materialist denies that there are any mental phenomena, he is not merely saying that there are no mental phenomena as the dualist construes them. That is, it is not simply being said that there are no immaterial mental phenomena. All materialists claim that. The very bold thesis that there are no mental phenomena, no matter whether they are interpreted as immaterial or as material entities, is being advanced. Once we see just how bold the thesis of eliminative materialism really is, we also see just why many philosophers are apt to reject it without a further thought. They regard eliminative materialism as preposterous.

Even so, eliminative materialism has been defended; we will consider two theories proposed as ways to save eliminative materialism. The first, known as analytical or logical behaviorism, attempts to avoid the problem facing eliminative materialism by analyzing the meaning of psychological expressions of language in terms of purely physicalistic expressions. This elimination of the need for psychological terms is thought to justify a corresponding elimination of psychological entities. A second theory concentrates on the reference or denotation of psychological terms, instead of their meaning, in order to eliminate psychological entities. This attempt utilizes what is called the "double-language theory," because it claims that there are two quite different sorts of ways to refer to certain physical entities. Psychological terms and certain physicalistic terms refer to, or denote, or name the very same entities, namely, certain physical processes in human bodies. As can be seen, both theories approach a substantive philosophical issue by an explicit examination of certain features of language. In this respect they exemplify the recent trend in Great Britain and the United States to take a linguistic approach to philosophy. It is the view of many of these linguistic philosophers that language holds the key to the final termination the problems and puzzles that have perplexed philosophers for centuries.


Analytical behaviorism is the theory that all sentences using psychological or mentalistic terms are transformable by analyses of what they mean into sentences using no psychological terms, but using only terms that refer to some kind of bodily behavior. This theory, then, claims that although there are many true sentences using psychological terms, we do not have to infer from this that these terms refer to mental objects, events, and states, because we can reformulate every one of these sentences in such a way that we use only terms that refer to material objects, events, and states. Consequently, the analytical behaviorist admits that sentences such as 'I like you,' 'Smith believes that it is raining,' and 'Jones suffers from inferiority feelings' are in many cases true. Therefore he is not committed to defending the implausible sentence 'There are no mental phenomena such as beliefs and feelings.' But having made this admission, he claims that he can still consistently be a materialist because to admit that a sentence is true is not to commit oneself to what it refers to. The analytical behaviorist says that psychological sentences really refer to human bodily behavior, and he attempts to show this by the way he analyzes their meanings. It seems, then, that by considering language, by "operating on a semantical plane," the materialist may be able to avoid the predicament just described.

Before we move on to evaluate critically analytical behaviorism, there are two other things we should do. The first is to distinguish analytical behaviorism from methodological behaviorism, a distinction often ignored, and the second is to explain the concept of analysis relevant to analytical behaviorism. In discussing Hobbes' position we claimed that a behavioristic psychology is consistent with epiphenomenalism and also parallelism. It is also consistent with the deterministic version of interactionism developed when we discussed the objection to interactionism based upon an allegedly required gap in physiological explanation. Thus, a behavioristic psychology is consistent with all the mind-body dualisms we have discussed. This is because a behavioristic psychology, as we saw, takes the only subject matter of psychology to be human bodily behavior, but it need not decree that there are no mental phenomena. The theory that proposes this kind of psychology has been called behaviorism. But because it is a theory only about the methodology of the science of psychology, it would be better to call it "methodological behaviorism." We can see, then, that methodological behaviorism, the behaviorism relevant to the science of psychology, is consistent with a mind-body dualism as well as with materialism. Thus it differs markedly from analytical behaviorism. Analytical behaviorism, as interpreted here, entails materialism, but methodological behaviorism does not, because it is consistent with dualism. It is true that many methodological behaviorists, especially such early ones as Watson, seem to have thought that materialism is implied by methodological behaviorism, but this is because they did not distinguish sufficiently between statements about the methodological requirements of science and statements that assert metaphysical positions.

In discussing analytical behaviorism we are interested in analyzing what certain linguistic expressions mean, and therefore we are interested in what is called meaning analysis. This can be defined as the linguistic method that analyzes the meaning of a linguistic expression (the analysandum) in either of two ways. The first is by providing another linguistic expression (the analysans) synonymous with the analysandum. The second way is by providing expressions such that (1) each is synonymous with certain key expressions containing the analysandum; and (2) none contains any expression synonymous with the analysandum. The first kind of meaning analysis is explicit definition and the second is contextual definition. The distinction between these two is important because only the latter is relevant to analytical behaviorism, as some examples will show. We can give an explicit definition of 'human' by saying that 'human' equals by definition 'rational animal,' or, as we shall state it:

'human' = df. 'rational animal'
We would explicitly define 'bachelor' as follows:
'bachelor' = df. 'unmarried male'
On the other hand, we could begin to give a contextual definition of the term 'existent' by seeing that a sentence such as, 'Many strange things are existent,' can be analyzed as:
'Many strange things are existent' = df. 'There are many strange things'
Here in the analysans there is no word or phrase synonymous with 'existent.'

Let us now see why only contextual definitions are relevant to analytical behaviorism. Consider the following sentence:

The average American family has 1.3 cars.
Let us assume that it is true, that we convince someone it is true, and that he then exclaims that he had never before realized that there was in the United States a family with a fraction of a car. We would of course try [to] explain to him that he had misunderstood what we meant. We were not talking about -- referring to -- a real family. Although the sentence is true, there really is no such family. This might leave our friend completely puzzled. How can that sentence be true and there not be such a family? What we would have to do is show him that the average families are in an important way eliminable unlike ordinary families. The problem in eliminating this average-family is like that of the eliminative materialist. We cannot identify this average family with some ordinary family, for that would seem to imply that somewhere there really is a family with 1.3 cars. What would we do to eliminate this average-family? We would try to restate the whole sentence in such a way that no expression in it seems to refer to an average family, but only to ordinary families. For this purpose an explicit definition will not help. Suppose in the preceding sentence we replace 'the average American family,' with the following analysans:
the American family which has the average number of cars.
Here we have attempted to provide an explicit definition. Will it help our mistaken friend? Not at all, because the analysans of 'the average American family' we have provided, is an expression that seems to refer to that same strange family. We can help our friend, however, if we contextually define 'the average American family' by providing a sentence synonymous with the puzzling sentence but containing no phrase synonymous with 'the average American family.' Consider the following:
The number of family-run cars in the United States divided by the number of American families equals 1.3.
Here we do not have 'the average American family' or any expression synonymous with it. We only have expressions that refer to ordinary families and cars, and no one need wonder about the strange family with its fractional car. We have "analyzed away" a very strange kind of entity by a contextual definition because we have shown that no expression that seems to refer to such an entity needs to be used. We need only use expressions that refer to ordinary entities. Thus if we are to analyze away certain entities, we cannot use explicit definitions. Only contextual definitions can help. Let us see if they can help the analytical behaviorist.

An Attempt to Justify Analytical Behaviorism: Verifiability Criterion of Meaning

Many people would doubt that sentences involving psychological terms could be contextually defined in terms of sentences containing only behavioral terms. There are others, however, who say that no matter how difficult it may be to find adequate contextual definitions of this kind, it can nevertheless be done. This confidence in analytical behaviorism was expressed by Carl Hempel, who at one time claimed:

All psychological statements which are meaningful, that is to say, which are in principle verifiable, are translatable into propositions which do not involve psychological concepts, but only the concepts of physics. The propositions of psychology are consequently physicalistic propositions. Psychology is an integral part of physics.24
An example of a psychological sentence which Hempel claims to be verifiable, thus meaningful and translatable into a physicalistic sentence, is a statement "that Mr. Jones suffers from intense inferiority feelings of such and such kinds. . . ."25 Because this sentence can only be confirmed or falsified by observing Jones' behavior, the sentence "means only this: such and such happenings take place in Mr. Jones' body in such and such circumstances."26

It is important to note that although a statement such as 'Jones suffers from inferiority feelings a, b, and c,' which we can call sentence J, does not seem to mean or to be translatable into any kind of physicalistic sentence, there is nothing else such a sentence can mean, given Hempel's criterion of meaning, if it is to be considered a meaningful sentence. We can put Hempel's point into a deductive argument as follows:

  1. The conditions of the verification of J are Jones' behavior under such and such conditions.
  2. The meanings of sentences are the conditions of their verification.
  3. The meaning of J is Jones' behavior under such and such conditions.

Because J is no different from other psychological sentences, this argument can be generalized to conclude that the meaning of any psychological sentence is the behavior of some person or persons under certain conditions. Consequently, given the preceding argument, it follows that for each psychological sentence we can find a physicalistic sentence having the same meaning. And because the relevant physicalistic sentences are about certain bodily events and states, we can conclude that all psychological sentences can be analyzed into sentences using only behavioral terms -- that is, analytical behaviorism is true. But, of course, the crucial question is whether Hempel's argument is sound. It surely seems acceptable if there is good reason to accept premise (2), because all we need do to justify premise (1) for any particular psychological sentence is to find out the particular way we actually verify it when we observe human behavior. But premise (2) is by no means obvious.

If we look at premise (2) it can be seen that it embodies a specific theory about the meaning of sentences. This is the kind of theory that any linguistic approach on the semantic level must consider. The particular theory offered by Hempel is the one proposed over the last forty years by those philosophers known as logical positivists or logical empiricists.27 It is the position of these philosophers that there are only two different kinds of sentences which are literally either true or false. All other sentences lack truth-value. Everyone agrees that there are kinds of sentences which lack truth-value, that is, are neither true nor false, but the logical positivist rejects more sentences than most others. We would all agree that sentences which express commands (such as 'Shut the door!'), sentences which are used to ask questions (such as 'Where are you going?'), sentences which express feelings (such as 'Hurrah for the home team!'), and several other kinds of sentences are neither true nor false. But most of us think that sentences such as 'God created heaven and Earth and all things,' 'We ought to help others,' 'This is a beautiful picture,' and 'The mind is distinct from the body' are either true or false. That is, we think that religious, ethical, aesthetic, and metaphysical utterances are, by and large, either true or false. But here the logical positivist disagrees, because he thinks that the only kinds of sentences that have truth-values are those that are empirically verifiable and those that are analytically true or false. The positivist, then, holds the view that if a sentence is not analytically true or false and there is no possible way to verify it by observation, then we should conclude that the sentence is neither true nor false, but rather plays some different role in language. Such a theory has been called the verifiability criterion of meaning. Thus many ethical, aesthetic, religious, and metaphysical utterances are not analytically true or false and are not verifiable by observation, and hence positivists have claimed that they function to express certain feelings or desires or hopes of speakers rather than to assert something either true or false.

Hempel has claimed in his second premise that the meanings of sentences are the conditions of their verification, so that if there is no possible way to verify a sentence, then it has no truth-value and is what we can call cognitively meaningless. Thus premise (2) implies the verifiability criterion of meaning, because if the meaning of sentences are the conditions of their verification, then all cognitively meaningful sentences are verifiable. Consequently, if there is reason to reject the verifiability criterion, then there is reason to reject Hempel's premise. Incidentally, it should be noted that the verifiability criterion does not imply Hempel's premise, because it might be true that sentences are empirically verifiable and thereby cognitively meaningful, but false that the meaning of sentences are the conditions of their verification. Consequently, even if the verifiability criterion is acceptable, Hempel's premise is still confronted with the additional problem of justifying its claim about what the meaning of a sentence is. However, because we shall find enough reasons to cast doubt on the verifiability criterion, we need not examine the additional problem here.

There is one problem for the verifiability criterion of meaning that we shall not examine even though it is critical. It is this. The criterion is supposed to separate certain sentences from others on the basis of empirical verifiability, and although it seems intuitively evident where the division should be made, no one has yet provided a definition of verifiability that is at all adequate for the iob. Each attempt at an adequate definition has either been so broad that it allows obvious nonsense to count as meaningful, or it has been so narrow that it has excluded many sentences vital to empirical sciences. Thus the criterion is of no use in deciding whether or not a particular sentence is meaningful. But because it has not yet been shown that there can be no adequate definition, we should not rest the case against the criterion on this problem.28

The most serious problem for the verifiability criterion of meaning is that it seems to be self-defeating. It claims that the only true sentences are analytic sentences and empirically verifiable sentences. Consequently, the criterion itself, if it is true, must be either analytic or empirically verifiable. But it is not analytic, because there is nothing self-contradictory about the claim that some nonanalytic, nonverifiable sentences are true. Indeed it would seem that most people untutored in theories of meaning would reject the criterion as false because they think many religious and ethical utterances, among others, are true. Consequently, it does not seem to be a generalization based upon empirical observation of the actual ways in which people use and respond to sentences. It seems, then, to be neither analytic nor empirically verifiable. Some positivists, recognizing this problem, have claimed that this is merely a proposal about what we should consider meaningful and have backed up their proposal by saying that it is surely necessary for a meaningful language of empirical science. But although it may well be that the language of science should meet an adequate verifiability criterion of meaning, this provides no reason to think that any other meaningful area of language must meet similar requirements.

In short, we have found good reason to reject the verifiability criterion. It is acceptable only if there is reason to think that it is true of the way things are or that it is a sound proposal about the way things should be. But we have found no reason to accept it as a proposal and good reason to reject its truth, because it is not either analytic or empirically verifiable, as the criterion itself requires of all true sentences. Consequently, because Hempel's premise (2) implies the verfiability criterion, we should reject the premise as well as the criterion.

We must give up the short road to analytical behaviorism and try the longer, more difficult route. That is, we must see whether analytical behaviorism is justified by trying to provide some specific contextual definitions of particular psychological sentences. If we meet with some success, then there is reason to accept analytical behaviorism; if we have no success, then we should reject it.

An Objection to Analytical Behaviorism: Cannot Analyze Belief Sentences

One philosopher who claims that the program of analytical behaviorism will not be accomplished is Roderick M. Chisholm. He has been willing to rest his whole case against the analysis of psychological sentences in terms of behavioral sentences upon the inability of anyone to analyze satisfactorily sentences containing 'believe.' In various articles he has shown that all the attempts made so far have failed to provide adequate analyses of belief sentences. The consequence of this, of course, is that because 'believe' is a psychological term, analytical behaviorism must be rejected. To see Chisholm's reasoning we must turn to examples of how he criticizes certain specific analyses. In each case he shows either that the analysans is not synonymous with the belief sentence, or that it has been made synonymous only by using some technical term that is not needed to describe merely bodily phenomena. The second prong of Chisholm's attack is as important as the first, because many people have tried to avoid psychological language not by translating it into behavioral language, but by coining scientific-sounding terms which seem to have only one function, namely, to avoid psychological terms. For example, some psychologists, instead of saying,

The subject of the experiment expects food.
The subject of the experiment has an F-expectancy.
As Chisholm points out, such ploys cannot be considered as providing behavioral analyses of psychological sentences because "In all probability . . . the psychologist has only one means of conveying what such expressions as 'F-expectancy' or even 'food-expectancy' might mean; namely, he can tell us that an animal may be said to have food-expectancy if and only if the animal expects food."29 Thus if certain technical terms require the use of certain usual psychological terms to explain their meanings, then analyses of psychological sentences containing such technical terms should not be used as cases supporting analytical behaviorism.

Let us examine the four main kinds of behavioral analysis Chisholm has considered, for in doing so we shall begin to see that behavioral analyses of belief sentences seem doomed to failure.30 He considers in turn what he calls the "specific response" analysis, the "appropriate behavior" analysis, the "satisfaction" analysis, the "verbal response" analysis. In each case we shall consider an analysis of the belief sentence:

Jones believes that there is a fire nearby.
According to the specific-response analysis we might try to analyze this sentence as:
Jones exhibits fire-responses to his immediate environment.
But we have a technical term, 'fire-responses,' in this analysis. How are we to explain what it means? An analytical behaviorist might say that:
Jones exhibits fire-responses.
Jones is exhibiting that behavior which he exhibits when and only when there is a fire.
But this will not do, because it entails that Jones believes that there is a fire when and only when there really is a fire. Jones, however, like all the rest of us, often believes things that are false. And, of course, it will do no good to patch up the analysis by saying 'when and only when he thinks that there is a fire' because 'thinks' is a psychological term. In this way, Chisholm refutes the specific-response analysis.

The appropriate-behavior analysis fares no better. This kind of analysis would analyze:

Jones believes that there is a fire nearby.
Under circumstances relevant to there being a fire nearby, Jones would behave in a way appropriate to there being a fire nearby.
Here we have a purely behavioristic analysis which includes no special technical terms. Thus, unless there is reason to think that it is possible that one of the sentences is true and the other false, we can accept this analysis. What we need to do, then, is see whether we can think of a situation in which one sentence would be true and the other false. This is not hard to do. We can conceive of a case where Jones is involved in a fire drill and behaves just the way he would in a real fire. His behavior is certainly appropriate to there being a fire nearby. Let us also take this to be a case in which, unknown to anyone, a fire has broken out in the building just before the scheduled drill. Surely, then, Jones is behaving in circumstances relevant to there being a fire nearby. Thus the analysans is true. But Jones, knowing that this was merely the scheduled drill, believes that there is no fire nearby. Thus the analysandum is false and the analysis fails. In such a way Chisholm refutes the appropriate-behavior analysis.

The satisfaction analysis proposes to analyze:

Jones believes that there is a fire nearby.
Jones is in a bodily state which would be satisfied if and only if a fire were to occur nearby.
Here again, although we have a purely behavioral analysis, we can conceive of numerous counterexamples to the claim that the two sentences are synonymous. We can conceive of a situation in which Jones has an uncontrollable urge to toast marsh-mallows over an open fire. He has the marshmallows on a stick and needs only a fire. Thus Jones is in a bodily state that would be satisfied if a fire were to occur nearby and that would be satisfied only if a fire were to occur nearby. But let us construct the situation in such a way that Jones has no means to start a fire and has searched everywhere for a fire but found none. He consequently believes that there is no fire nearby. Here the analysandum is false and the analysans true, so that once again an attempt at a behavioral analysis of a psychological sentence fails.

The last attempt we shall, consider is the verbal-response analysis, which is favored by many linguistic philosophers, such as Rudolf Carnap. Carnap has analyzed such sentences as:

Jones believes that there is a fire nearby.
Jones has relation B to 'There is a fire nearby' as a sentence in English.

The immediate reaction to this analysis is to point out that it uses a technical term 'relation B,' which is needed only for analyzing away psychological terms. In reply to this attack on the analysis, however, it has been pointed out that people have relationships of various kinds to different verbal utterances and that it is part of the job of natural science to study them. Thus the preceding analysis, because it uses an expression which describes a relationship between people and language, is using an expression required whether or not it is used in an analysis of psychological sentences. Nevertheless, although this first objection may have been refuted, there remains the problem of explaining the meaning of 'relation B' without having to rely on psychological terms. This has not been done. Furthermore, because it is not necessary that someone understand English if she is to believe that there is a fire nearby, the explanation of what 'relation B' means cannot imply that Jones knows English. Thus an equally good analysans should be

Jones has relation B to 'II y a un feu pres d'ici' as a sentence in French.
We might try to explain the analysans by claiming that it means:
Jones has B-responses to a sentence in his language synonymous with the English sentence, 'There is a fire nearby.'
However, not only does this sentence take us right back to the specific-response analysis, with its problems, but it would seem that the phrase 'in his language' means 'in the language he understands,' and 'understands' is surely a psychological term. Perhaps 'understands' can be avoided and 'B-response' elucidated without relying on psychological terms, but so far no such attempts have succeeded. The verbal-response analysis, then, appears also headed for failure even though there are philosophers who still seek to perfect it.

We have examined four of the most plausible attempts to provide a behavioral analysis of belief sentences and have found reasons to reject the analysis offered in each case. With Chisholm we can conclude that it seems highly unlikely that any other behavioral analyses will fare better. Thus although we have not examined, and cannot examine, all possible behavioral analyses of psychological sentences, we have examined a sufficiently good sample to conclude that we should reject analytical behaviorism because it claims that belief sentences and all other psychological sentences can be given behavioral analyses. Consequently, if eliminative materialism is to be saved as a viable alternative to dualistic interactionism, we must depend upon the double-language theory.


The double-language theory has been summarized as follows by Feigl, who has been its chief proponent:

Certain neurophysiological terms denote (refer to) the very same events that are also denoted (referred to) by certain phenomenal terms. The identification of the objects of this twofold reference is of course logically contingent, although it constitutes a very fundamental feature of our world as we have come to conceive it in the modern scientific outlook. [We can say] that neurophysiological terms and the corresponding phenomenal terms, though widely differing in [meaning], and hence in the modes of confirmation of statements containing them, do have identical referents. I take these referents to be the immediately experienced qualities, or their configurations in various phenomenal fields.31
J. J. C. Smart maintains a thesis much the same as Feigl's, but he states it specifically in terms of sensations and brain nrocesses. He says that
insofar as "after-image" or "ache" is a report of a process it is a report of a process that happens to be a brain process. It follows that the thesis does not claim that sensation statements can be translated into statements about brain processes. Nor does it claim that the logic of a sensation statement is the same as that of a brain-process statement. All it claims is that insofar as a sensation statement is a report of something, that something is in fact a brain process. Sensations are nothing over and above brain processes.32
There are three important features of the theories of these two men which we should carefully note.

First, both Feigl and Smart discuss certain expressions of language and what these expressions refer to or are used to talk about. Thus both men are taking a linguistic approach to the mind-body problem. But this should not be taken to mean that they are concerned only with language, because what the expressions refer to are not other linguistic expressions but some kinds of nonlinguistic phenomena. Thus both men are interested in getting from certain facts about language to certain conclusions about nonlinguistic reality.

Second, both men stress that the psychological expressions and the physiological expressions they are considering differ widely in meaning, so that the psychological sentences are not analyzable or translatable into physiological sentences. They claim only that these two different kinds of terms have the same referents, not that they have the same meanings, They claim, for example, that the expression 'brain process' has the same referents as the expression 'sensation,' but the two are clearly different in meaning. Consequently, although they are like analytical behaviorists in being linguistic philosophers, they are not analytical behaviorists, because they deny that psychological sentences are synonymous with behavioral sentences.

Third, although both Feigl and Smart are double-language theorists, there is one important point on which they disagree. As brought out in the preceding quotations, they have very different views about the common referents of psychological and certain physiological terms. Feigl. claims that the common referents are immediately felt qualities, that is, feelings in their uninterpreted or raw state. Smart, on the other hand, claims that sensation terms refer to entities that are nothing but brain processes. These surely do not seem to be uninterpreted feelings. This is a most important difference for our purposes, because Smart's thesis, if generalized to include all psychological terms, becomes the theory of materialism, but Feigl's thesis seems to be inconsistent with materialism, because on his view the referents are mental entities. Because we are here discussing materialism, we shall concentrate primarily on Smart's materialistic version of the theory, which we can now see consists of two claims. The first is that psychological and certain physiological expressions have common referents, and the second is that this common referent is in every case material.

Any double-language theorist who holds the two preceding claims is a materialist, but this is not enough to determine whether he is an eliminative materialist or a reductive materialist. An example will help distinguish between these two kinds of double-language materialists. Consider a materialist who claims that, first, the psychological term 'experience of pain' denotes, or refers to, the very same things as the neurophysiological term 'firing of C-fibers,' and, second, these common referents are nothing but firings of C-fibers, that is, certain neural processes in brains that consist entirely of certain nerve cells being triggered to fire. Such a materialist is a double-language theorist. If he further claims that there really are no experiences of pains or other sensations, he is an eliminative materialist. If, however, he claims instead that the reason the two terms have common referents is because each experience of pain is identical with firing of C-fibers, then he is not an eliminative materialist because if experiences of pains are identical with actual brain processes, then there are experiences af pain. He is instead a reductive materialist. Smart, for example, is a reductive materialist. He claims not only that 'sensation' and 'brain process' have common referents, but also that sensations are nothing over and above brain processes, that is, sensations are really reducible to, and thereby identical with -- the same thing as -- certain brain processes. We shall examine this thesis later in Chapter 4, but let us first consider the double-language theorist who is also an eliminative materialist.

The chief objection to an eliminative materialist is that he must claim that whenever anyone says, even with complete sincerity and conviction, "I have a sharp, throbbing, aching pain," what he is reporting or actually referring to, is never a pain which is aching, throbbing, or sharp, but rather something extremely different from a pain, namely, a purely physiological neural process. But this is initially extremely implausible, indeed."even quite absurd. Could there possibly be a way to defend such an initially implausible thesis?

Defense of Eliminative Materialism: in Analogy Between Demons and Pains

Richard Rorty, who is an eliminative materialist, provides a defense by drawing an analogy between the use of strange demons by a primitive tribe to explain illnesses and our present use of pains and other sensations to explain certain human behavior. He first describes this imaginary tribe: A certain primitive tribe holds the view that illnesses are caused by demons -- a different demon for each sort of illness. When asked what more is known about these demons than that they cause illness, they reply that certain members of the tribe -- the witch-doctors -- can see, after a meal of sacred mushrooms, various (intangible) humanoid forms on or near the bodies of patients. The witch-doctors have noted, for example, that a

blue demon with a long nose accompanies epileptics, a fat red one accompanies sufferers from pneumonia, etc. They know such further facts as that the fat red demon dislikes a certain sort of mold which the witchdoctors give people who have pneumonia. If we encountered such a tribe, we would be inclined to tell them that there are no demons. We would tell them that diseases were caused by germs, viruses, and the like. We would add that the witch-doctors were not seeing demons, but merely having hallucinations.33
He then uses the analogy to dispel the initial absurdity of claiming that no one has pains:
The absurdity of saying "Nobody has ever felt a pain" is no greater than that of saying "Nobody has ever seen a demon," if we have a suitable answer to the question "what was I reporting when I said I felt a pain?" To this question, the science of the future may reply "You were reporting the occurrence of a certain brain-process, and it would make life simpler for us if you would, in the future, say 'My C-fibers are firing' instead of saying 'I'm in pain'." In so saying, he has as good a prima facie case as the scientist who answers the witch-doctors' question "What was I reporting when I reported a demon?" by saying "You were reporting the content of your hallucination, and it would make life simpler if, in the future, you would describe your experiences in those terms."34
On this view, then, it will be reasonable to eliminate sensations in the future if, as is likely, the somewhat primitive, explanatory function of sensation terms is replaced by the more advanced terms of physiology. At that point, we will be justified in casting sensations away as we already have done with demons.

If Rortv's analogy with demons succeeds, then he has a rebuttal to the strong objection to rejecting his theory; if it does not succeed, then we shall have reason to reject this theory. Rorty realizes that there is an important disanalogy between the two cases, but he fails to see it destroys his defense. He has provided the eliminative materialist with an answer to the question of what I report when I say I have a pain, but, unlike the demon case, he has not provided any plausible way to explain why, there is such a widespread mistake of believing and reporting that there are pains. And he must provide such an explanation instead of merely an answer if he is to rebut the charge that his theory implies an absurdity. Scientists explain why the witch doctors believe they see demons by stating that eating sacred mushrooms causes them to have hallucinations of demons, and these hallucinations fool the witch doctors who believe they are actual demons. An eliminative materialist, however, cannot use hallucinations to explain why we mistakenly believe we experience pains, because even if it made sense to talk of an hallucination of a pain, these hallucinations would be mental objects rather than neural states, and some mental entities would not be eliminated. Furthermore, there seems to be nothing else a materialist could use to explain our common mistake of believing and reporting certain occurrences, which are merely neural processes, to be sharp, throbbing, and aching pains. Thus the absurdity remains, and this attempt to defend eliminative materialism fails.

With the failure of this last attempt, we can conclude that eliminative materialism should be rejected. There seems to be no way to make plausible the complete elimination of sensations. Nevertheless, materialism is not thereby refuted, because reductive materialism remains unscathed. By claiming that sensations are nothing but, and therefore are identical with brain processes, it rejects the claim that no one has pains, and thus avoids the objection that is fatal to the eliminative version.


The crucial claim of a reductive materialist is that mental entities, such as sensations, are nothing over and above certain physical entities, such as brain processes. This reductive claim states more than that each sensation is the same thing as some brain process, because the use of the phrase 'nothing over and above' also implies that sensations have only the physiological properties of certain brain processes. Thus they are said to have no psychological properties, in spite of the way it may seem. This is clearly essential to a reductive materialist, because he claims that sensations are really material entities.

Notice that someone can hold the identity theory without also being a reductionist. We have seen that Feigl is not a materialist. There is also the seventeenth-century shilosopher, Spinoza, who sounds much like a reductive materialist when he says, "the mind and the body are one and the same thing conceived at one time under the attribute of thought [that is, when understood as mental], and at another under the attribute of extension [that is, when understood as physical].38 But Spinoza disagrees with materialists when he goes on to claim that what is conceived in these two different ways is neither mental nor material because it has both physical properties and mental properties. Let us call this version of the identity theory the "neutral theory'' because it proposes entities that are neither mental nor material. Spinoza, then, is an identity theorist, but not a materialist.

To help understand the difference between the neutral identity theory and reductive materialism, it will be helpful to revert to the chart used earlier (see p. 175). We lave already disposed of one species of materialism, namely the eliminative version. Reductive materialism remains. To better understand it, consider dualism as represented on the chart. There it is pointed out that a dualist will maintain that mental objects are totally immaterial; they are objects which are totally lacking in material properties. All properties that they have are either mental properties, or else are iroperties that are neither mental nor material. Examples of the latter sort of property would include normative properties such as moral ones. Thus, the properties being morally good, or being morally wicked, are neither mental properties nor material properties. Presumably, if dualism is true, some mental objects, namely minds, might have such normative properties. What about mental properties? Examples of these would include, say, the achiness of a pain, or the yellowish color of a visual afterimage. We have already mentioned examples of material properties several times. These include such properties as mass, weight, spatial location, specific electrical charge, and the like.

Now an eliminative mterialist insists that there are no mental phenomena of any sort. Hence, such a theorist denies that there are any mental properties. But a reductive materialist does not take this extreme position. Instead, he adopts the very opposite of the dualist position. While the dualist says that mental objects have purely mental properties, and perhaps some that are neither mental nor material, and no others, the reductive materialist claims that mental objects have purely material properties (and perhaps some that are neither mental nor material properties), and no others. Although this is not an eliminative position, it is still an exceptional contention. Consider how reductive materialists will handle a specific mental entity, say, a pain. Unlike an eliminative materialist, he will agree that there surely are pains. Pains, a reductive materialist contends, are material entities of some sort, presumably neural in nature. What about pain that is especially achy; what will be said about this mental property of achiness? The reductive materialist has two options here. He can deny that there are such properties, or he can insist that the mental property of achiness is really nothing more than some material property. These two options are forced on him by his contention that the only properties any concrete thing has are material properties, and perhaps some other nonmental, nonmaterial properties as described earlier on page 188.

We now are in a position to see that the distinction between eliminative and reductive materialism has surfaced in two ways. Thus, as depicted on the chart (see p. 175), a materialist has the options of eliminating mental objects, events, and states, or else saying that each is really nothing more than a material object, event, or state, respectively. We have lately introduced the notion of property, and said that there are, or at least seem to be, both mental properties and material properties. Once again, a materialist has the same two options. He can either maintain that there are no mental properties, or else claim that mental properties are really nothing more than material properties. Of course, if one thinks that there are no mental phenomena whatever, then one will also claim that there are no mental properties. But if, like the reductive materialist, it is allowed that there are mental objects, events, and states, then something has to be done with their mental properties. And, as just noted, the only options available to a materialist at this point, are either to reject mental properties outright, or to maintain that mental properties are nothing more than material properties.

The foregoing remarks explain the reductive materialist theory in somewhat more detail. Let us now consider the neutral identity theory. To do this, consider again our example of an especially aching pain. The neutral identity theory contends that the pain is in fact a neural entity of some sort, presumably a brain entity. Hence, it identities the mental thine (pain) with something else (a brain entity). This, obviously, is the factor that makes the theory an identity theory. But now let us reconsider what we have described as the mental property of achiness. A neutral identity theor insists that the pain has this mental property, but contends further that this property of achiness is not something one can eliminate or reduce to a material property. It is the latter view which differentiates the neutral theory from reductive materialism. It is a consequence of this theory that things such as pain have both sorts of properties: mental properties such as achiness, as well as material properties such as a specific electrical charge. Since, on this account, mental phenomena would have both sorts of properties, things such as pain would not be purely mental things as dualists say, nor would they be purely material things, as reductive materialists claim. It is this factor which renders mental phenomena such as pain neutral; they have both sorts of properties, and so are neither purely mental nor purely material.

We need to be clear about this latter point. Someone might say that there is something very odd about the manner in which we have described the neutral identity theory. After all, if we say a mental thing such as pain is identical to a material thing such as a brain entity, are we not then saying that the pain is partly mental (it has mental properties) and also partly material (it likewise has material properties)? Since pain and other things we call 'mental phenomena' would thus be partly mental and partly material, it would be wrong to call them 'neutral'. Neutrality typically is taken to amount to being neither one thing nor another, and this would not be so given what we have just said.

There is some force in this objection. However, what it really shows is the weight being borne by the word 'purely'. Since on the neutral identity theory pain has both mental and material properties, it is not purely rnental and it is not purely material either. It is this and this alone which justifies us in speaking of the neutral identity theory. It must also be emphasized that the neutral identity theory is not a materialist theory. The neutral theory allows that pain, say, has some mental properties such as achiness, and that these mental properties are not elimmable or reducible to purely material properties. Materialism though, in any of its forms, is inconsistent with the claim that there are mental properties of this sort. The reductive materialist must claim that whenever anyone says, even with complete sincerity and conviction, "I have a sharp, throbbing, aching pain," what he is reporting is never a pain which is sharp, throbbing, and aching, but rather a pain which has only the purely material properties of the neural brain phenomenon with which it is identical. Thus although, unlike an eliminative materialist, he can agree that we often report pains, he must deny that we ever report something which has the purely psychological properties of being sharp, throbbing, and aching, because pains never have those purely nonmaterial properties. But this is initially extremely implausible, indeed, even quite absurd. Could there possibly be a way to defend such an initially implausible thesis?

A Defense: Central State Materialism

The only remotely plausible defense of the claim that there are sensations but that they lack all purely psychological properties, is one that construes sensations as entities postulated to explain certain sorts of bodily behavior. For example, when we see someone injured, writhing, and screaming, we postulate that the injury has caused him to experience pain, and we explain the writhing and screaming as caused by this experience. On this view, such experiences are essentially understood in terms of the roles they have in these causal explanations of behavior. As one materialist says, "The definitive characteristic of any experience, as such is its causal role."36 Given the additional plausible premise that all bodily behavior is in principle explainable in terms of the physiological processes and states of a person's central nervous system, this materialist goes on to conclude that these central physiological states have the definitive causal explanatory roles of experiences. Consequently, he argues, each experience is nothing over and above a materialistic state of a central nervous system. This is central-state materialism.

There are three objections to this argument. First, although we generally do assign causal explanatory roles to sensations and experiences, we certainly do believe that in our own cases we immediately experience certain properties of sensations which they have whether or not they have any causal relationships to any of our bodily behavior. To think that all the properties that pains actually have are to be discovered or postulated by science is to treat one's own sensations as if they belonged to someone else. Surely each person is directly aware of some properties of his own sensations and experiences.

But, second, even if we assume that each psychological state has certain definitive causal relationships and that physiology is in principle adequate to explain all human behavior, we need not conclude that these psychological states are identical with central states. We have seen how a dualistic interactionist can accept the explanatory adequacy of physiology if he postulates a certain sort of causal role for mental entities. Each one would be caused by some brain event and would in turn causally affect another brain event in a way that leaves no gap in the adequacy of physiological explanations of behavior.

Furthermore, third, even if we reject dualism and agree, that each psychological state is identical with a state of a person's central nervous system, it does not follow from this that these states are material states. They would have certain physiological properties because of their causal role in physiological explanations, but they also could have those purely psychological properties that each person so intimately experiences them to have in his own case. For these three reasons, then, this argument in defense of reductive materialism fails. It does not succeed in dispelling the initial absurdity in the thesis of a reductive materialist about pains and other sensations.

These considerations do not suffice to refute reductive materialism; they serve only to undermine one plausible argument for that theory. However, two further factors are relevant. First, this argument is as strong as any we know of in favor of reductive materialism. Moreover, reductive materialism seems quite implausible independently of criticisms of arguments in its behalf. Consider, again, the options such a materialist faces with regard to purely mental or psychological properties such as the achiness of pain. Complete elimination of such properties is not a reasonable option, given that we surely seem to experience such properties all too often. But reduction of mental properties to purely material properties of, say, the central nervous system, is not plausible either. A mental property of achiness is not the same as those purely material properties which physiologists discover or ascribe to brain phenomena such as neural impulses or molecular activity.37 Reductive materialism, then, should be rejected.

We can also conclude that materialism, whether eliminative or reductive, should be rejected. There seems to be no way to make plausible either the complete elimination of sensations with their psychological properties or the stripping of all psychological properties from sensations. Nevertheless, although this monistic theory is refuted the double-language theory has not been refuted, because a neutral or nonreductive version of the identify theory remains unscathed. By agreeing that psychological states are brain states having both psychological and physiological properties, this neutral theory avoids the defeating objections to the three materialistic theories we examined. By relying on the double-language theory, it does not require a behavioral analysis of psychological sentences. Thus it avoids what refutes analytical behaviorism. And by retaining sensations and their psychological properties, it avoids the objections to Rorty's version of eliminative materialism and to central-state materialism. Furthermore, as we shall see, it also avoids all but one of the objections to the various dualistic theories. This neutral theory is, then, most appealing.


Let us see how easily the neutral identity theory avoids the objections that have plagued the various forms of mind-body dualism. This will clearly show the advantages of the theory.

1. Epiphenomenalism. The neutral identity theory avoids both of the objections that we have seen uniquely face epiphenomenalism. First, epiphenomenalism denies what seems to be true, that mental events causally affect the body. The identity theory, however, by identifying sensations and the like with brain phenomena, asserts that mental events do causally affect the body because brain processes obviously affect the body. Second, as Feigl points out, epiphenomenalism requires causal laws involving danglers, which makes such laws uniquely different from all other scientific laws. That is, such laws involve mental effects which do not causally affect anything at all. The identity theory, however, requires no nomological danglers because, as Feigl stresses, the sentences that express relationships between mental and physical phenomena, such as 'Sensation S1 occurs if and only if brain process B1 occurs,' are not causal laws. Causal laws express causal relationships between different events. They state that certain events are the causes of other different events. But if the identity theory is correct, those sentences relating mental and physical events do not relate different events which are causally related. They express identity, which, of course, is not a relationship between different events. Thus, such psychophysical sentences are not causal laws and will not involve danglers if the identity theory is correct. We can see how the identity theory avoids danglers in another way. A dangler is an effect which is not part of some continuing causal process. But brain processes surely are parts of causal processes and if, as the identity theory claims, mental events are identical with brain processes, then mental events are parts of causal processes.

2. Parallelism. The crushing objection to parallelism is that is requires a deus ex. machina to explain mind-body regularities. Furthermore, by denying mind-body interaction, it goes contrary to what seems to be true. The identity theory easily avoids these two objections. Because mental events are identical with brain processes and because brain processes causally interact with other parts of the body, it follows that mental events causally interact with parts of the body. And obviously no deus ex machina is needed to explain mind-body regularities, because they are completely explained by pointing out that underlying the regularities are identities. Where there is identity there must be regularity.

3. Dualistic Interactionsim. Dualistic interactionism is the theory that we have found to be preferable to those we have so far compared with it. We saw six objections leveled at the theory, although most of them turned out to be of little consequence after we carefully refined what the theory required. The only three objections left that seem to have any force are the objection from the problem of other minds, the objection from the apparent lack of characteristics relevant to causal interaction, and the objection concerning the inexplicability of how thr mental affects the body. We shall examine how the neutral theory fares with these three objections, but let us first briefly indicate how easily it avoids the other three. The problem of deciding where mind-body interaction takes place is solved by discovering where those brain processes that are identical with mental events occur. This poses no insurmountable problem. The problem of a gap in physiological explanations and causal chains is quite obviously avoided, because if mental events are identical with certain brain events, then nothing nonphysiological is required either for full explanations or continuous causal chains. In this regard, a neutral theorist could even agree with our central-state materialist that certain causal roles are partly definitive of experiences. But, of course, he could also claim that they have these roles merely as a matter of fact.

The objections stemming from evolution are countered by saying that although no new nonphysical entities evolve through time as the dualist claims, certain bodies have evolved in such a way that they come to have properties of a new kind, namely, psychological properties. Such properties are often called "emergent" properties, because they emerge only when certain sorts of complex physical systems have evolved from simpler material stuff. So far, then, the identity theory easily dispenses with problems that have bothered other mind-body theories. Let us turn to the three more serious problems facing dualistic interactionism.

The objection from an apparent lack of characteristics relevant to causal interaction between mental and physical events is no more troublesome for the identity theory than any of those objections discussed previously. The identity theory need only point out that if it can be granted that brain events have characteristics relevant to causally interacting with other bodily events -- something we all grant -- then there should be no worry about mental events because they are identical with brain processes. This same point, furthermore, dispels the mystery of how the mental affects the body without violating any scientific principles. Each mental event affects a brain event in just the way any physiological event affects another. We have, then, two important points at which the neutral version of the identity theory is clearly superior to dualistic interactionism. If it can avoid the problem of other minds and faces no objections uniquely its own, it is clearly the theory we should accept.

At first glance it may seem that the identity theory can dispense with the objection from other minds as easily as it has with all the others directed at dualistic interactionism. It will be remembered that this problem arises for dualism because if dualism is true, then no one perceives other minds, nor can he deductively or inductively infer that there are other minds from premises describing what he perceives. And because the belief in other minds can be justified only in one of these three ways, it follows that the belief cannot be justified. Now it seems that if the identity theory is correct, we can perceive mental events because we can perceive, or at least detect by perception, those brain processes that are identical with mental events. Thus, it seems, each of us can justify his belief that there are mental events other than his own, and consequently, that there are other minds. Although we may not be able to justify the belief by a sound inference, we do not need one, because if the identity theory is true, we can justify it noninferentially by perceiving mental events.

However, before the identity theorists claim victory, they should look more carefully at this way out of the problem of other minds. It is true that if mental events are identical with certain brain processes, then we perceive mental events when we perceive brain processes. But this is more like seeing a conglomeration of H2O molecules when we see water, than it is like seeing a white sheet of paper when we see the page of a book. We both see a white sheet and see that there is a white sheet. But although we may see H2O molecules, we do not see that there are H2O molecules here. We must infer this from what we see. Similarly, we may be seeing a mental event occur when we observe a brain process, but we do not see that a mental event is occurring. We must infer that a mental event is occurring, and this is obviously not noninferential perceiving. Thus, we can conclude that someone else has mental phenomena by using a statement of the identity theorist as a premise:

  1. Brain process B1 is occurring in Jones (because I perceive that it occurs).
  2. Sensation S1 is identical with brain process B1.
  3. Jones is having sensation S1.
We can, however, reach the same conclusion by using a statement of the interactionist instead of premise (2):

2a. Sensation S1 is caused by brain process B1.
But, of course, premises such as (2) and (2a) are just the key one a skeptic about other minds attacks, and neither theory is able to handle his attack better than the other. Consequently, we cannot use the problem of other minds to help choose between the two theories.

Incidentally, of all the positions we have examined, only analytical behaviorism avoids this problem. If that position is true, then psychological sentences are entailed by behavioral sentences and so conclusions about other minds can be deduced from premises describing certain behavior that we observe. But although this is an attractive feature of analytical behaviorism, it is of little help in salvaging the theory because it seems that no such entailments hold.

Although we have not found that either the most plausible dualism or what surely seems to be the most plausible alternative theory avoids the problem of other minds, we have found that the neutral identity theory does handle quite easily two objections that interactionism is unable to rebut in a completely satisfactory way. Does this clear advantage, then, allow us to proclaim the neutral theory as the most plausible mind-body theory? This would be, unfortunately, premature because there is one important objection to the neutral version of the identity theory that we have not yet considered.

An Objection to the Identity Theory: The Nonidentity of Discemibles

The crucial objection to the identity theory has already been mentioned. It surely is wrong to claim that mental phenomena, such as sensations or thoughts, are identical with certain physical phenomena, such as brain processes, because we cannot say of nental phenomena many of the things we say of physical phenomena, and vice versa. This objection can be made more specific by using the principle of the identity of indiscernibles. According to this principle, objects that may seem to be different from each other are really identical if "both" have all the same properties, and if they are identical, then both have all the same properties. For example, the forty-ninth state of the United States is identical with Alaska. Both the forty-ninth state and Alaska have all the same properties, such as the properties of being the most northerly state, the largest state, and the state nearest Russia. However, the fiftieth state is not identical with Alaska, because the fiftieth state has the property of being an island, which Alaska does not have. The two are discernible and so nonidentical.

Let us now apply the principle of what the mind-body identity theory claims are identical.38 On one side of the identity claim, we have things such as sensations, pains, afterimages, beliefs, and desires. On the other side are certain physical entities, especially brain processes and constituents of brains such as nerve fibers. We ascribe properties to all these things. For example, we describe pains as intense, sharp, throbbing, aching, and unbearable; and we describe nerve fibers as located in the brain, publicly observable, conducting neural impulses, and constituted of molecules. Consequently, if the neutral identity theory is correct about the identity of pains with firing C-fibers, then by applying the principle of the identity of indiscernibles, we can conclude that sensations such as pains are located in the brain, are publicly observable, conduct nerve impulses, and are constituted of molecules. And we can also conclude that certain nerve fibers are intense, sharp, throbbing, aching, and unbearable. But surely these conclusions are not true. Therefore, according to this objection the neutral theory is incorrect.

It must be admitted that the two preceding conclusions are extremely odd, perhaps even a misuse of language. Indeed, the sentences 'My pain conducts nerve impulses' and 'My nerve fibers are aching unbearably' seem to be like the sentence 'My birthday is asleep in bed.' This third sentence seems to be clearly meaningless, and therefore neither true nor false, because it makes no sense to say such a thing about the day of someone's birth. The sentence, it is claimed, involves what is called a category mistake. That is, in this sentence the predicate 'is asleep in bed,' which is in one linguistic category, is ascribed to the term 'birthday,' which belongs in a different category. Whenever this occurs the resulting sentence is meaningless.39

If the sentences about pains and nerve fibers involve category mistakes, then they are meaningless and are, therefore, not true. It would seem, then, that this objection would be quite damaging. If, however, the neutral theory can avoid these odd-sounding and seemingly meaningless sentences, it can rebut this objection. Let us assume here that the sentences are meaningless, and see how a neutral theorist could handle the objection on that assumption. Consider the following sentence he might use to express his view: 'The very same entity that is aching and throbbing unbearably is conducting nerve impulses and is constituted of molecules.' This may be an unusual sentence but it is not particularly odd and certainly not meaningless. Furthermore, this theorist might even devise a new term to refer to such entities, for example, 'fibain.' He would then claim that there are fibains, that is, entities with those properties usually associated with pains, and also with the properties usually associated with firing C-fibers. He can, then, avoid category mistakes by saying that what are thought to be pains and what are thought to be firing C-fibers are really fibains. And a double-language theorist could say that 'pain' denotes is what 'firing C-fibers' denotes, namely fibains.

How do these remarks serve to rebut the objection based on the meaninglessness of certain sentences? They do so because the objection depends on the fact that the neutral identity is stated in certain words. All one need do, then, is change the words used to state the theory and the objection is avoided. He need not introduce the new term 'fibain' to do this, of course. Such a tactic is just one way among several of stating the neutral identity theory in different words so that the current objection simply does not arise.

There is another factor that is important. We have assumed that sentences such as 'My nerve fibers are aching unbearably' and 'My pain conducts nerve impulses' are meaningless. It may be, however, that these sentences simply seem or sound odd or unusual to us now, but that they are not meaningless. Thus, consider, 'Would you please pass me the sodium chloride; I want to put some on my French fry potatoes,' or perhaps 'Each motionless, clear, liquid pool of water is identical with a swarm of discrete particles, each of which is in constant motion.' There was a time when sentences such as these would have been considered quite odd and unusual. Nevertheless, we have by now become used to sentences of this type. We now know that they are not at all meaningless. The point is that when a theory is in the initial stages of its development, many sentences will seem odd at first and one will be tempted to suppose that they are without meaning. The passage of time, however, often shows that this supposition would have been mistaken. The analogy with the neutral identity theory should be clear. These sentences concerned with pain and nerve fibers just seem unusual to us now, when the theory has just recently been proposed. But, it is reasonable to think, they will seem less and less odd as time passes. One moral of this and our preceding remarks is that to base an argument about a nonlinguistic theory on linguistic considerations is often to rely on something that is precariously inconstant.


We have discussed dualist and materialist theories extensively. Dualists allow that there are two different types of entities, mental and material. Materialists, however, embrace a monistic theory, that is, one that insists that there is just one type of concrete entity, namely, material entities. What shall we say about the neutral identity theory? There are reasons for saying that it is a dualist theory. After all, it agrees that there are purely mental properties as well as purely material ones, even if there are no purely mental entities (objects, events, or states). It is precisely this fact which, we noted earlier, makes the neutral identity theory inconsistent with materialism.

There are also reasons for saying that the neutral identity theory is a monistic theory. It agrees that there are purely material things, but also denies that there are purely mental things or entities. One might then say that neutral things that are neither purely mental nor purely material, do not make up another distinct kind of thing, over and above material things. Instead, neutral things such as pain and sensations of other sorts, are indefinite as to type. That is why we say that they are neutral entities. Hence, one might say, the neutral identity theory is really a version of monism; the only distinct kinds of entities it allows for are material entities.

We find this to be purely a terminological issue, which we can stipulatively resolve in either direction. We will choose to say that the neutral identity theory is a monistic theory, despite the fact that it is not consistent with materialism. It is recognized, though, that it is not quite the same as the one other monistic theory we have considered, materialism, since the latter excludes neutral entities. On the other hand, someone might choose to claim that the neutral identity theory is a kind of dualistic theory. This is all right, so long as we realize that it merely represents a decision to use the terms 'dualism' and 'monism' so that they cover properties as well as entities. In classifying the neutral theory as a version of monism we have done nothing more than fail to make the latter terminological decision.


With the rejection of the one objection we found to the neutral version of the identity theory, it is easy to see that it is the most plausible of the many proposed solutions to the mind-body problem. It avoids the problems unique to each of the alternative theories and faces none uniquely its own. We have, then, reason to reject the theory that our description of a persor at the beginning of the chapter seemed to support, namely, dualistic interactionism. Nevertheless that theory is correct about there being interaction., although it is wrong about the entities that interact. It is also correct about there being a dualism, although it is wrong about the two sorts of entities involved. There are entities with only physical properties, such as stones and certain human organs, and there are also 'neutral' entities, such as 'fibains,' which have both physical and psychological properties. It is these neutral entities, rather than mental entities, that interact with bodily processes. Human beings, then, are different from material objects, but not because they have minds or spirits in addition to their bodies.

This difference between persons and material objects is important. It is important for deciding the basic ethical question of how we ought to treat other persons, because | it seems that beings with thoughts, sensations, feelings, and emotions should not be treated merely as an unfeeling, nonthinking material object such as a rock. This difference also has religious significance. If immortality, or life after death, requires a mind or spirit that is distinct from a body so that it survives when the body decays after death, then on the neutral theory there is no life after death. The mental activity of a person ceases when the brain activity with which it is identical ceases. However discomforting this may be, it is nevertheless reassuring that with no firings of C-fibers after death there will be no pain or suffering then either.


1 Rene Descartes, The Philosophical Works of Descartes (New York: Dover, 1955), Vol. 1, p.

2 Ibid., p. 192.

3 Ibid.

4 Ibid., p. 345.

5 Ibid., p. 347.

6 C. D. Broad, The Mind and Its Place in Nature (London: Routledge and Kegan Paul, 1962), p. 97.

7 Ibid., p. 98.

8 C. J. Ducasse, "In Defense of Dualism," in S. Hook, ed., Dimensions of Mind (New York: Collier Books, 1961), p. 88.

9 Ibid., pp. 102-103.

10 A. J. Ayer, The Problem of Knowledge (Baltimore: Penguin Books, 1965), p. 222.

11 Ducasse, "In Defense of Dualism," pp. 88-89.

12 Broad, The Mind and Its Place in Nature, pp. 107-108.

13 Ibid., p. 110.

14 Ibid., p. 113.

15 Ibid., p. 112.

16 G. W. von Leibniz, Philosophical Papers and Letters, edited by L. E. Loemker (Chicago: University of Chicago Press, 1965), p. 751.

17 T. H. Huxley, Method and Results (New York: Appleton-Century-Crofts, 1893), p. 244.

18 J. B. Pratt, Matter and Spirit (New York: Macmillan Publishing Co., Inc. 1922).

19 H. Feigl, "Mind-Body, Not a Pseudoproblem," in Hook, Dimensions of Mind, p. 37.

20 Hobbes, Hobbes Selections, edited by F. J. E. Woodbridge (New York: Scribner's, 1930), p. 139.

21Ibid., p. 107.

22 Ibid., p. 136.

23 Ibid., p. 140.

24 C. Hempel, "The Logical Analysis of Psychology," in H. Feigl and W. Sellars, eds., Readings in Philosophical Analysis (New York: Appleton-Century-Crofts, 1949), p. 378.

25 Ibid.

26 Ibid.

27 For the classical statement of logical positivism, see A. J. Ayer, Language, Truth and Logic (New York: Dover 1952).

28 For detailed but difficult discussions of this problem, see I. Scheffler, The Anatomy of Inquiry (New York: Knopf, 1963), pp. 150-54; D. Makinson, "Nidditch's Definition of Verifiability," Mind (April 1965); and J. W. Cornman, "Indirectly Verifiable: Everything or Nothing," Philosophical Studies (June 1967).

29 R. M. Chisholm, "Intentionality and the Theory of Signs," Philosophical Studies (1952).

30 For Chisholm's discussion of these kinds of analysis, see Perceiving (Ithaca, N.Y.: Cornell University Press, 1957), pp. 168-73.

31 Feigl, in Hook, ed., Dimensions of Mind, op. cit., p. 38.

32 J. J. C. Smart, "Sensations and Brain Processes," The Philosophical Review, 68 (1959), pp. 144-45.

33 R. Rorty, "Mind-Body Identity, Privacy, and Categories," Review of Metaphysics (1965-1966), pp. 28-29.

34 Ibid., pp. 30-31.

35 B. Spinoza, Spinoza Selections, edited by John Wild (New York: Scribner's, 1930), p. 209.

36 D. Lewis, "An Argument for the Identity Theory," The Journal of Philosophy, LXIII (1966),

37 For further discussion and a defense of this claim, see J. Cornman, Perception, Common Sense and Science (New Haven: Yale University Press, 1975), Appendix.

38 This point is discussed in some detail in J. R. Stevenson, "Sensations and Brain Processes: A reply to J. J. C. Smart," The Philosophical Review, 69 (1960), pp. 505-10.

39 See G. Ryle, "Categories," in A. Flew, ed., Logic and Language, 2d series (Oxford: Basil Blackwell, 1955), pp. 65-81 for a discussion of category mistakes; and for a more extensive and detailed study, see J. Cornman, "Types, Categories, and Nonsense," American Philosophical Quarterly, Monograph Series 2 (1968), pp. 73-97.


1. Do you think that the characterizations of mental and material phenomena on page 145 are satisfactory? Using these characterizations, classify the following terms. (Note that some of these are neither mental nor material and thus do not belong in any of the classes.) Give reasons for your choices.

a loud sound
the color blue
bodily pleasure
intellectural pleasure
the number 3
mirror images

2. On page 145 it is claimed that mental objects do not seem to be located in space, yet we talk of mental pictures or images "in one's head," and we speak of pains, which are supposedly mental, as being located at different places in our bodies. Can these "locations" of what seems to be mental, be reconciled with the claim on page 145, or must we revise the claim?

3. In the passage quoted on page 151, C. D. Broad states that one of the causal factors involved in voluntary movements of our bodies is a mental phenomenon called a "volition." Do you think that when someone moves his arm intentionally he has a volition and thereby "an idea of his arm (and not his leg or liver) and an idea of the position which he wants his arm to be"? Does this happen when, for example, you type a paper or ride a bike or swim? How do you think Broad could best answer these questions?

4. Broad claims that mental causes act on neural causal chains by varying the resistances of certain synapses to nerve currents. He concludes from this that there is no gap in physiological causal chains. But if we think of a synapse as like an electrical circuit with a variable resistor, must not energy be expended in "turning the knob" that changes the resistance? That is, how can the "knob be turned" by something such as a mental event that has no energy to expend? How could Broad best reply to this?

5. Reread the quotation from Broad on page 160. Consider that the behavior of ants, oysters, protozoa, and certain internal-guidance missiles is "varied appropriately to meet special circumstances." Must Broad conclude that all of these have minds? Surely he should not conclude that missiles have minds? Is there a reply Broad can make?

6. Can you think of any kind of human behavior that seems to be explicable only on the hypothesis that humans have immaterial minds? Consider falling in love, getting angry, telling a joke, writing poetry, dreaming, seeing a mirage. State whether you think that these or any other human activities require explanations in terms of mental phenomena, or whether you think no such explanation of any human behavior is needed. Give reasons for your answer.

7. Suppose that someone invents a computing machine that expresses itself in verbal utterances which sound much like a person talking. Suppose also that it can learn from past mistakes and improve its ability to arrive at answers to questions of many different kinds. Suppose, furthermore, that we ask it whether it thinks about the questions it is asked and whether it has feelings and desires of its own; its answer is "yes." Should we conclude that it has a mind? If not, should we conclude that humans also lack minds? Justify your answer.

8. There is an objection to the existence of minds that we have not considered. We have seen that it is possible that the mental has evolved from the material, but according to the theory of evolution there is reason to think that it has not. According to this theory, all our physical organs exist because they have contributed to our ability to survive as individuals or as a species. A mind, however, is clearly not at all necessary to enhance these abilities. Consequently, we should conclude that nothing mental has evolved from the original material primordial mass. How could an interactionist reply to this? An epiphenomenalist? What is your reply?

9. Interactionists claim that certain material events are causally related to certain mental events, either as causes or as effects. Parallelists claim that no material events are causally related to mental events. Rather, some material events are constantly accompanied by mental events. Is there some way to decide by observation and experimentation whether there are mind-body causal relations or merely constant correlations? Is there any way to decide between the two theories on the basis of experimental evidence? Can epiphenomenalism be experimentally distinguished from these other two dualistic theories? If not, what do you think are the consequences for the mind-body problem?

10. If the reply to objection (5) on pages 171-172 is true, then it might be claimed that there would be no way to explain scientifically how brain events cause nerve impulses to take certain paths, because such causation would have to be regarded by scientists as cases of proximate causation. But, it can be objected, all physical causation is ultimately to be explained in terms of atomic and subatomic occurrences. Consequently, scientists should not conclude that neurological causation is proximate; thus, it should be concluded that the reply to objection (5) is false. Is this objection sound? Explain your answer.

11. Distinguish between materialism and mechanism, and explain how one may be a materialist but not a mechanist. Can one be a mechanist but not a materialist? Explain.

12. Distinguish between methodological behaviorism and analytical behaviorism, and explain how one can be a methodological behaviorist but not an analytic behavorist. Can one be an analytic behaviorist but not a methodological behaviorist? Explain.

13. Explain in your own words the difference between an explicit and a contextual definition. Give an example of each not given in the text.

14. Evaluate the following argument:

Analytical behaviorism is merely a thesis about the contextual definitions of psychological terms. Consequently, it is not a metaphysical position, nor does it entail any, because such a position is about what there is, not about the definitions of words. Therefore, the text is wrong when it asserts that analytical behaviorism entails materialism.

15. According to the verifiability criterion of meaning, which of the following sentences are cognitively meaningful? Explain your answers.

The planet Pluto is made of green cheese.
Everything in the universe is twice the size it was yesterday.
John is certainly a good son.
Either God exists or He does not exist.
Please drive carefully.
There is life after death.
There is no life after death.

16. Some people claim that the verifiability criterion of meaning is a metaphysical thesis. If this were so what would the consequences be for the criterion itself?

17. [On page 184] it was concluded that it is reasonable to reject analytical behaviorism on the grounds that those attempted analyses of psychological sentences examined in the text all failed. Do you agree that this is sufficient reason to reject analytical behaviorism? Explain. Do you think a scientifically more "penetrating" analysis might succeed? If you do, suggest how one might proceed.

18. Consider the following argument from W. V. O. Quine, Word and Object (New York: Wiley, 1960):

If there is a case for mental events and mental states, it must be just that the positing of them, like the positing of molecules, has some indirect systematic efficacy in the development of theory. But if a certain organization of theory is achieved by thus positing distinctive mental states and events behind physical behavior, surely as much organization could be achieved by positing merely certain correlative physiological states and events instead. . . . The bodily states exist anyway; why add the others?
Is Quine's view compatible with the identity theory? Explain. Evaluate his argument utilizing the discussion in the text about sensations as entities postulated to explain behavior.

19. Evaluate the following objection to the identity theory:

No one can see that I am in pain by looking at my pain and therefore no one can see my pain. But neurosurgeons can see brain processes, so none of my or anyone else's pains are identical with brain processes. The identity theory is false.

20. Consider the following objection to the identity theory from R. Abelson, "A Refutation of Mind-Body Identity," Philosophical Studies, Vol. 21: 85-89 (1970): It is possible that human beings can think of any particular number. But there are infinitely many numbers. Thus it is possible for there to be infinitely many different human thoughts. But there is only a finite number of human beings throughout time, and there is only a finite number of discrete brain states for each person. Thus there are only a finite number of different human brain states. From this Abelson concludes that there are more mental states possible than there are brain states available. Thus someone might be in a mental state with which no discrete brain state is uniquely correlated. But then that mental state would not be identical with any brain state, and so the identity theory would be false.

21. The neutral identity theory is acutally quite implausible because it is "species chauvinistic." That is, it identifies what we usually call 'mental entities' with neural entities. But this conclusion is based solely on what has been found out about humans and higher animals. Lower level creatures such as mollusks certainly feel pains, but they lack the sort of neural structure required by the neutral identity theory. Moreover, for all we know there may be all sorts of extraterrestial beings with all manner of mental states but which lack neural systems altogether. Hence, we should reject the neutral theory.

Evaluate this objection to the neutral identity theory.

22. The conclusion reached in this chapter is that on the basis of the data discussed in the chapter the neutral version of the identity theory is the most reasonable position. Are there any data overlooked? Do you think some other position is more reasonable -- perhaps because of something overlooked here? If you do, give a short defense of your claim.