Wilfrid Sellars, "The Identity Approach to the Mind-Body Problem," Review of Metaphysics 18 (1965): 430-51. Presented at the Boston Colloquium for the Philosophy of
Science, April, 1963. Reprinted in Wilfrid Sellars, Philosophical Perspectives (1967).
THE IDENTITY APPROACH TO THE MIND-BODY PROBLEM1
M y primary aim in this paper is to set the stage for a discussion of some of the central themes in the so-called "identity" approach to the mind-body problem. I particularly have in mind Herbert Feigl's elaborate statement and defense of this approach in Volume II of the Minnesota Studies. A secondary but more constructive purpose is to bring out some of the reasons which cause me to think that the theory is either very exciting but false, or true but relatively uninteresting.
I shall begin with a preliminary formulation of the identity theory which will highlight the topics I propose to discuss. Roughly put, the theory claims that what it calls "raw feels" -- a technical expression which is intended to cover impressions and images pertaining to the external senses, as well as bodily sensations and feelings in a more usual sense -- are identical with "brain states." It hastens to add that in speaking of "raw feels" as identical with "brain states" it does not simply mean that the very same logical subjects which have "raw feel" characteristics also have "brain state" characteristics, or that "raw feel" characteristics do not occur apart from "brain state" characteristics, but rather that the very characteristics themselves are identical. As Feigl puts it,
"raw feel" universals are identical with certain "brain state" universals.
This rough and ready formulation of what is actually a highly sophisticated philosophical thesis blocks out three topics with \
which any attempt to assess the identity theory must come to grips. Each of these topics turns out on the most cursory inspection to involve highly controversial issues which are at the very center of the philosophical stage. I shall not attempt to resolve all or, indeed, any of these issues. My aim will rather be to thread my
way through them in such a way as to bring out the common ground I share with the identity theory and thus make possible a meaningful joining of issues.
It will not have passed unnoticed in this particular climate of opinion that the identity theory as formulated above is committed
to the idea that it makes sense to speak of the identity of attributes or universals. This is the first of the thorny topics on which something must be said. This may be the place, but it is not the time to develop a theory of abstract entities. I shall simply mobilize some of the pre-analytic strands which any theory must take into account, and develop them in a way which gives the claim that "raw feel" universals are identical with certain "brain state" universals at least the appearance of being in keeping with the spirit of a scientifically oriented philosophy.
Universals, then, are a subset of abstract entities. Their distinctive feature is that they are expressed in language by predicates (e.g., "red") or by predicative expressions (e.g., "three feet long," "between red and yellow in color"). I shall say that predicates (under which term I shall usually include predicative expressions) "stand for" or "express" universals. Universals may be referred to as well as stood for or expressed. But predicates do not refer to universals; indeed, they are not referring expressions at all. Among the expressions which refer to universals, a particularly important role is played by those which are formed from predicates or predicative expressions which stand for or express the universals
to which reference is made; thus,
Being three feet long
Being between red and yellow in color.
Universals are public objects. They are identities not only with respect to their many instances, but also with respect to the many minds which think in terms of them, and the many languages, which give expression to them. This inter-subjective and inter-linguistic character must be accounted for by any adequate theory of abstract entities. Equally important, and even more "platonistic" in tone, is the distinction which must be drawn between those universals which have been "discovered" or have come to be "known" and those which have not, and, within the sphere of
the former, between those which are effectively taken account of
by our language, and those which are not. To unpack this a bit,
I shall assume that a universal is "discovered" or comes to be
"known" in the course of coming to know what use a predicate
would have to have in order to stand for or express it. The universal is effectively taken account of by our language if our language contains a predicative expression which actually has this use.
Notice, therefore, that, while we can refer to unknown or undiscovered universals (I drop the quotation marks from these metaphorically used terms) and to universals which are not effectively taken account of by our language, only universals which are effectively taken account of by our language can be referred to by referring expressions formed from predicates which stand for or express them. Thus, although we can refer to the unknown property of persons which would explain their telekinetic powers, our language contains no predicate which stands for or expresses this property.
Two universals are identical if, were a language to contain predicative expressions which stand for or express them, these predicative expressions would either independently have the same use, or one would be a definitional abbreviation of the other.
Clearly, much of the burden of the above distinctions is borne by the word "use" and the phrase "the same use." My general strategy is clear. It is to connect "realistic" talk about universals with "nominalistic" talk about linguistic expressions. My further strategy would be to connect talk about the use of expressions with talk about uniformities in the occurrence of linguistic inscriptions, and, therefore, to build a bridge to "behavioral criteria of synonymy." But that is strategy for a war and not a battle. Here I shall limit myself to pointing out that the patterns of use I have primarily in mind are (1) the reporting or observation pattern, and (2) the consequence pattern. The latter is, roughly, the pattern which would find its explicit formulation in what Carnap calls "transformation rules," L-transformation rules, P-transformation rules, and others. I add "and others" to Carnap's list because it is not clear that it is an exhaustive classification. Other possible candidates are "bridge laws" and "correspondence rules."
I pointed out above that we can refer to universals for which we
have no corresponding predicates. There are two types of case, one of which is, for our purposes, uninteresting. Thus there is a sense in which it can be said that there are color universals for which we have no predicates. We can imagine that we had no predicate for the color between red and yellow. It should be noticed, however, that while we might not have had the predicate "orange," we might well have had the predicate expression "between red and yellow in color." And, indeed, to be in the logical space of color is to have predicate expressions adequate to the job of introducing predicates in the narrower sense, thus "orange."
The interpretation of statements asserting the identity of universals, where the logical space of the universals is in this sense familiar, is relatively straightforward. Consider, for example:
The universal which. . . . = the universal which. . .
We can distinguish two forms the descriptions might take: (1) Each locates a universal with respect to a point outside the logical space of the universal located. Thus,
The color of Plato's beard = the color of your father's moustache.
Here, if we have the relevant information, we can go from one of the descriptions either directly to an illustrating name, of the universal (i.e., a name formed appropriately from the predicative expression which stands for it), thus
The color of Plato's beard = orange (i.e., being orange)
or to a description which locates the universal with respect to the logical space to which it belongs, thus
The color of Plato's beard = the color between red and yellow.
(2) In the second type of case, at which we have just arrived, at least one of the descriptions locates the universal with respect to the logical space to which it belongs. Where we have a predicate which expresses the color between red and yellow, we can move from the above (supposing that predicate to be "orange") to
the color of Plato's beard = orange (i.e., being orange).
If we do not have such a predicate, we at least have the predicate expression "between red and yellow in color," and can say
The color of Plato's beard = being between red and yellow in color,
and could introduce a predicate having the use of "orange."
The important case of referring to universals for which we have no corresponding predicates is that of referring to what I have called unknown or undiscovered universals. Consider, thus,
1. The property which an adequate theory of telekinesis -- if
we but had it -- would ascribe to persons having this
Contrast this with
2. The property which the theory (current) of chemical interactions assigns to catalysts.
In (2) it is implied that we have a predicate in our language which stands for or expresses the property in question. Not so in case (1). There the property in question is referred to by relating it to the properties expressed by the predicates of the science of telekinesis at its operational and instantially inductive level. The logical space of these empirical constructs is not that of the properties to which access would be gained by constructing a sound theory of telekinetic phenomena.
In general, then, the universals which it is the task of theoretical science to "discover" are referred to via a reference to the unborn or undeveloped theory, the predicates of which would stand for or express them, and, therefore, via a reference to the logical space of the empirical properties of the phenomena to be explained by the theory.
Yet the predicates of even sketchily developed theories express or stand for universals. Here it is essential to note that as a theory develops, its predicates cannot, in general, be said to continue to stand for or express the same universals.2 This brings me to a fundamental point which adds an element of symmetry to our previous classification of universals. To the classification (which highlights the temporal dimension);
we must now add a third heading under (2)
- Not yet discovered.
- Discovered or known:
- Not yet effectively taken account of by our language
- Effectively taken account of by our language
and a new major category:
- No longer effectively taken account of by our language
- Lost, or, so to speak, undiscovered universals.
These considerations strongly suggest that the objective or "platonistic" status I am ascribing to universals might be construed in a Peircean way as relative to the continuing scientific community. Thus, if on hearing the above proliferation of universals, one is tempted to expostulate, "which of these universals really exist?" I would reply by recalling Peirce's characterization of a true proposition as one that the continuing scientific community would ultimately accept -- and then I would change the subject.
Now if the claims of the identity theory are placed in the frame-word of the above distinctions, it is clear that the theory does not assert the identity of "raw feel" universals with certain "brain state" universals which are effectively taken account of by existing language. For on the above analysis, this would involve that some "brain state"predicates have the same use as "raw feel" predicates. And this is obviously not the case. The claim is, rather, that among the universals which would find expression in the predicates of a to be developed "brain state" theory, some are identical with "raw feel" universals.
At first sight, this is hardly much better. For, it might be urged, how could any predicates in a "brain state" theory have the same use as "raw feel" predicates? After all, the latter doesn't even presuppose the knowledge that there are such things as brains! But before we take up this and other objections, we must explore the notion of a "raw feel" universal.
The "rawness" of "raw feels" is their non-conceptual character. The sense in which "raw feels" are "of something" is not to be assimilated to the intentionality of thoughts. To say that they are non-conceptual is, of course, not to deny that they can be referred
to and characterized by the use of concepts, or even directly responded to by concepts in direct self-knowledge.
The word "feel" in the expression "raw feel" is an extension to all sense modalities of a use of the word "feel" which has its ultimate source in such contexts as,
1. He felt the hair on the back of his neck bristle.
In this primary context, "to feel" is clearly a cousin of "to see,"
and feeling in this sense can properly be classified as a mode of
perception. Notice that feeling in this sense is conceptual; a
propositional attitude. One would, perhaps, be more comfortable
about this remark if the example had been,
2. He felt that the hair on the back of his neck was bristling.
The relation between (1) and (2) is an interesting and important topic in the philosophy of perception. I shall simply assume on the present occasion that (1) is a stronger form of (2) which emphasizes the noninferential character of the experience.
Notice that to ascribe a perceptual propositional attitude to a person in the form,
3. He perceived that-p
is to endorse the proposition involved in the attitude. We can, however, ascribe the same propositional attitude in a non-endorsing way by using such locutions as
4. He thought he perceived that-p
None of these is completely neutral with respect to endorsement. They all tend to imply the falsity of the proposition involved in the attitude, and have other overtones which are irrelevant to my purpose. I shall make a technical use of (7) in which it will imply neither the truth nor the falsity of the proposition involved
in the perceptual propositional attitude. Thus the statements
5. It seemed to him that he perceived that-p
6. It appeared to him that-p
7. He was under the (perceptual) impression that-p.
8. He was under the tactual impression that the hair at the back of his neck was bristling
ascribe perceptual propositional attitudes while making no commitment concerning the truth or falsity of the proposition involved in the attitude.
9. He was under the visual impression that there was a red and triangular physical object in front of him
Now a classical theme in the philosophy of perception is that the truth of statements such as (9) implies the occurrence of something which is variously called a "(visual) sensation" (a sensation of a red triangle) or a "(visual) impression" (an impression of a red triangle), where this occurrence is understood to be a non-conceptual episode which somehow has the perceptible qualities which the propositional attitude takes to be exemplified in the world of perceptible things. Thus, the fact that a person is under the visual impression that a certain stick in water is bent is taken to imply that he is having a visual impression of a bent object. I shall
assume that this is true. This does not mean that I accept the
".sense datum inference," for it should not be assumed that to have a visual impression is to sense a sense datum as these terms are
used in classical sense-datum theories.
Notice that visual impressions are classified by the use of the word "of" followed by the phrase which would appear in the statement of the propositional attitudes which imply their occurrence, thus
Impression of a red and triangular object
Impression that there is a red and triangular object in
front of one.
The idea that there are such non-conceptual episodes was put to use in explaining, for example, how a straight stick (in water) can look bent and a red object (in green light) look black. It was postulated that the propositional attitude ascribed by
He is under the visual impression that there is a black object in a certain place
involves, among other things, (1) the occurrence of an impression of a black object, (2) the occurrence of the thought that there is a
black object in a certain place, the thought (or perceptual judgment, as it was called) being evoked, by the impression. Roughly
speaking, impressions that were construed as conceptual responses to impressions of. To this was added the idea that, while in standard
conditions viewing red objects results in an impression of a red object and viewing bent objects results in an impression of a bent object, in non-standard conditions (e.g., viewing a straight stick in water) the viewing of an object that is not bent may result in an impression of a bent object and the viewing of an object that is not red may result in an impression of a red object.
Although the examples I have been using come from vision, exactly the same distinctions were drawn in the case of feeling. Here "feeling of. . ." is the counterpart of "visual impression of. . . ." We can therefore understand the philosophical use of the expression "raw feel" as an extension to all modes of perception of an expression which stands for the non-conceptual kind of episode, which explains why a person can be under the impression that he is being pricked by something sharp when this is not the case.
I pointed out that sense impressions or raw feels are classified according to the perceptible qualities which are ascribed to some part of the world by the perceptual propositional attitudes which they evoke and which characterize their standard causes. As I see it, the "of" phrases in
Sense impression of a red triangle
are adjectives which, in addition to classifying raw feels extrinsically by their causes and effects, also classify them with reference to their intrinsic character.
Raw feel of being pricked by a sharp object
How are we to understand the intrinsic character of raw feels? Obviously the sense impression of a red triangle is not, in the literal sense, either red or triangular; nor is the raw feel of being
pricked by a sharp object a being pricked by a sharp object.
The most that can be said is that the families of qualities and relations which intrinsically characterize raw feels or sense impressions correspond in a certain way to the families of qualities and relations which characterize perceptible objects and processes. (The scholastics took the different, and ultimately unsatisfactory.
tack of holding that the characteristics were the same, but the mode of exemplification different.) I shall return to this point later. For the moment I shall simply say that the logical space of the qualities and relations which characterize raw feels is, in certain respects, isomorphic with the logical space of the perceptible qualities and relations of physical objects and processes. It would be useful, therefore, to introduce predicates for raw feels which are formed from predicates which stand for perceptible qualities and relations by adding the subscript "s" Thus a triangulars impression or raw feel would be one which in standard conditions is brought about by viewing a triangular object and which, ceteris paribus, results in being under the impression that a triangular object is before one.
It will have been noticed that even my characterization of the intrinsic properties of raw feels has been, so to speak, extrinsic. For I characterized them in terms of their correspondence with the perceptible qualities and relations of physical objects and processes. It might be inferred from this that I think of our access to the logical space of impressions as indirect, as based upon a prior access to the logical space of perceptible qualities and relations. I shall postpone taking a stand of my own on this matter and limit myself for the moment to pointing out that the type of identity theory I am examining rejects this suggestion and insists that our access to the logical space of sense impressions or raw feels is direct and, indeed, is the presupposition of our access to the
logical space of physical objects and processes. It insists, indeed, that the qualities and relations of "raw feels" are "directly given" and that physical objects and their properties are "existential hypotheses" whose reality is guaranteed by the fine job they do of saving the appearances.
Finally, a categorial point about raw feels which is implicit in
the preceding remarks. They are construed as "pure episodes"
and are contrasted with dispositions and mongrel categorical-hypothetical states. It should be noticed that the fact that one
has in some sense "privileged access" to a state of oneself doesn't by
itself imply that this state is a pure episode. Children can be
trained to respond linguistically to Skinnerian states of their
organism. Nor, as this point in turn suggests, need "privileged
access" be construed in terms of classical theories of the given.
The identity, theory we are examining, however, is committed to the idea that raw feels are pure episodes and that raw feel facts are "given" in something like the classical sense
Before taking the bull by the horns, a word or two about the other terms of the identities envisaged by the identity theory. It will be remembered that, according to the theory, raw-feel universals are identical with certain brain-state universals. Which brain-state universals? Indeed, which brains?
For there is, in the first place, the brain as an empirical object to which empirical properties definable in observation terms can be ascribed. Can raw-feel universals be identical with universals which characterize the empirical brain? They cannot, of course, be identical with any universals expressed by empirical predicates defined in terms of the publicly observable features of the brain, for raw feels are pure episodes which are public only in the sense that others can infer that which is given to oneself. What authorizes the inference is, of course, a classic question.) Nevertheless it is important to see that there is a sense in which it is perfectly legitimate to suppose that raw feels are identical with certain states of the empirical brain. This, for the simple reason that it makes sense to suppose that they are states of the empirical brain. Imagine a person who has been defleshed and deboned, but whose nervous system is alive, intact, and in functioning order. Imagine its sensory nerves hooked up with input devices and its motor nerves hooked up with an electronic system which enables it to communicate. Without expanding on this familiar science fiction, let me simply suggest that we think of what we ordinarily call a person as a nervous system clothed in flesh and bones. In view of what we know, it makes perfectly good sense to introduce the term "core person" for the empirical nervous system, and to introduce
a way of talking according to which raw feels and, for that matter, thoughts are in the first instance states of "core persons" and only derivatively of the clothed person. \
I submit that in this sense most scientifically-oriented philosophers think of raw feels and thoughts as brain states. But while the thesis that raw feel universals are, in this sense, brain states
and therefore trivially3 identical with certain brain-state universals is almost undoubtedly true, it is relatively non-controversial and unexciting. Only a Cartesian dualist would demur.
For the claim that raw feels and thoughts are in this sense identical with brain states simply transfers the episodes and dispositions initially attributed to persons to the central nervous system, now conceived of as a core person. All of the important philosophical problems pertaining to the relation of mental states to physical states remain.
These considerations give proper perspective to the fact that the brain state universals which, according to the identity theory, are identical with raw-feel universals, are universals which would be expressed by certain predicates of an as yet to be elaborated theory of brain activity. Thus, instead of the relatively unexciting claim that raw-feel universals are identical with certain brain-state universals, where this reduces to the claim that raw feel universals are brain-state universals (i.e., ascribable to brains as core persons) the identity theory claims that raw-feel universals are not only brain-state universals in this unexciting sense, but are identical with certain universals to be "discovered" in the course of developing a scientific theory of brains.
Thus the question arises, "Is it reasonable to suppose that the scientific study of brains will lead to the discovery of brain-state universals which are identical with raw-feel universals?" And to this question we are strongly tempted to answer "No1!!" For interpreted along the lines sketched at the beginning of this paper it becomes, "Would an adequate theory of brains contain predicates which had the same use as raw-feel predicates?" And the idea that this might be so has a most implausible ring. It will be useful to formulate some of the objections which this idea tends to arouse.
The first is that, since predicates which would stand for the relevant brain-state universals are ex hypothesi theoretical predicates, they would not have the avowal or reporting use which is characteristic of some, if not all, raw-feel predicates. To this objection the identity theorist replies that, once the theory was developed,
people could be trained to respond to the brain states in question with the predicates of the theory -- which would thus gain an avowal use.
The second objection is that raw-feel predicates do not have a theoretical use, or, to put it in the material mode, raw feels are not theoretical entities. Here the identity theorist might reply that the other-ascriptive use of raw feel predicates is, in effect, a theoretical use. The force of this reply will be explored subsequently.
The third objection is the challenge "How can a predicate which applies to a single logical subject (a person) have the same use as a predicate which applies to a multiplicity of scientific objects?" The effect of this challenge is to make the point that the identity theory involves, not only the identity of raw-feel universals with certain brain-state universals, but of persons with systems of scientific objects. The identity theorist can be expected to reply that it is enough for his purposes if raw-feel universals which differ only in this categorial respect from the raw-feel universals expressed by predicates which apply to persons as single logical subjects are identical with certain brain-state universals. We shall leave this reply untouched, although we shall return to something like it at the end of our argument.
The fourth objection, however, is the most familiar and goes to the heart of the matter. "How," it asks, "can a property which is in the logical space of neurophysiological states be identical with a property which is not?" Otherwise put, "How could a predicate defined in terms of neurophysiological primitives have the same use as (be synonymous with) a predicate which is not?" To this question the inevitable answer is "It could not."
It might seem, as it has to many, that this is the end of the matter. The identity theory is absurd, and that is all there is to it. And, indeed, the identity theory as we have so far described it has no obvious defense against this standard objection. Yet it is not difficult to discern the fundamental strategy of the identity theorist in the face of this objection. It consists in an appeal to a supposed analogy between the speculatively entertained identity
of raw-feel universals with brain-state universals, and the once speculative but now established identity of chemical universals with certain micro-physical universals. The story is a familiar one, and I shall not bore you with the details. The relevant points are quickly made. Suppose Uc is a certain universal which the predicate "Pc" in the chemical theory current at time T stands for. And suppose that this chemical theory has a degree of sophistication essentially that of chemical theory today, but that micro-physics current at T is rudimentary. An "identity theorist" puts forward at T the thesis that chemical universals will turn out to be identical with certain to-be-discovered micro-physical universals (i.e., universals which would be expressed by the predicates of a more sophisticated micro-physics). An opponent raises the following objections:
1. How can micro-physical predicates which are not tied to Chemical Laboratory observables have the same use as chemical predicates which are?
The "identity theorist" replies that once the theory is developed, these defined micro-physical predicates are given this new use, and therefore acquire a chemical-theoretical role.
2. How can the predicates of current chemical theory, which have no definitional tie to micro-physical primitives, have the same use as any predicates of future micro-physical theory which will have such a tie?
This objection corresponds to the fourth and most telling objection to the mind-body identity theorist. And once again the objection is, in a certain sense, decisive. But here the "identity theorist" has available to him a move which is, at first sight, not available in the raw-feel, brain-state case. He can argue that both of the universals involved in the identification are to be discovered universals, the chemical ones as well as the micro-physical ones. Roughly, the identity claim takes the form,
The universals which will be expressed at T' by the predicates of a more adequate theory of chemical processes are identical with the universals which would be expressed at T' by the predicates of a more adequate micro-physical theory,
and, while the universals which the predicates of chemical theory current at T express would not be identical with micro-physical
universals, the universals which would be expressed by its more powerful successor might be.
For just as universals can be "discovered" and "given effective expression in our language" by our coming to use predicates in various ways, so universals can be "abandoned," and even lost, by no longer finding expression in our language. A chemical predicate which at T did not stand for a micro-physical universal may come to do so at T'. And the chemical universal for which it stood at T may be left in the lurch for a more sophisticated face.
The situation can be represented as one in which chemical theoretic predicates cease to stand for universals which are merely constantly co-exemplified with micro-physical universals ("bridge laws") and come to stand for micro-physical universals. The identification is made rather than discovered -- though the possibility of identification is discovered.
Is anything like this move possible in the raw-feel, brain-state case? Can the identity of raw-feel universals with brain-state universals be assimilated to the identity of chemical and micro-physical universals? Can raw-feel predicates and brain-state predicates be regarded as on the move towards a possible synonymity as was correctly predicted for the predicates of chemical and micro-physical theory? Summarily put, can raw feels be reduced to neurophysiological states?
This suggestion runs up against the obvious objection that according to typical identity theories, raw feel predicates, at least in their first-person use, are as untheoretical as predicates can be. Unlike the predicates of chemical theory, they are not on the move towards a more adequate logical space which they might come to express. Like the Bostonian, they are there. This is often put by saying that they "label" directly given qualities and relations.
And even if the identity theorist were to hold that the other-ascriptive use of raw-feel predicates is to be reconstructed as involving two stages:
the further move of reducing the other-oriented theoretical episodes to neurophysiological states would simply conflict with the identification formulated in (2).
- The-postulation of inner episodes to account for the perceptual behavior (including verbal behavior) of others
- The identification of these theoretical inner episodes with the raw feels given to self awareness (i.e., the identification of theoretical kind with the given kind),4
Suppose that at this point the identity theorist switches his tactics to conform to his reductionist strategy and abandons the thesis of given-ness. In other words, he now argues that instead of the use of raw-feel predicates being a confluence of two autonomous uses, a self-ascriptive use in which they "label given universals" and an other-ascriptive use in which they can be compared to theoretical predicates, the theoretical use in the explanation of perceptual behavior is primary and is an anybody-ascriptive use, and that such avowal or reporting use as raw-feel predicates have is a dimension of use which is built on this anybody-ascriptive theoretical use. Would not this complete the parallel with the chemistry-physics model? For if both raw-feel and brain-state predicates are theoretical predicates, can we not conceive of a reduction of raw-feel theory to brain-state theory?
If the concept of reduction is construed on the model of the physics-chemistry case, then, as I see it, the answer is "No." For reduction in this sense is a special case of the identification of universals located with respect to two theoretical structures which are expected to merge. Roughly, instead of the primitive predicates of one theory ending up as defined predicates in the unified theory -- which is the chemistry-physics case -- these primitive predicates may end up as primitive predicates in the unified theory. In effect, the to-be-discovered sense-impression universals would be no more complex than the sense-impression universals expressed by current sense-impression predicates; they would have a different categorial framework, and be logically related to (but not complexes of) universals expressed by other primitive predicates in the to-be-achieved unified sense-impression, brain-state theory. The logical space of sense-impressions would, so to speak, have been transposed into a new key and located in a new context. It would not
however, have become internally more complex in the way in which the logical space of chemical properties becomes internally more complex by virtue of their identification with micro-physical properties. That is to say, there would be no increase in complexity with respect to what might be called the factual content of sense-impression universals; such increased complexity as occurred would be of a logical character. Roughly, the new sense-impression universals would be exemplified, not by single logical subjects (persons), but rather by a manifold of logical subjects which might be called -- borrowing a term without its philosophical commitments -- sensa.
But if sense-impression or raw-feel theory is to merge with brain-state theory, the latter phrase must be used in its proper sense of "theory adequate to explain the properties of empirical brains as core persons" and freed from any commitment to the idea that brain theory is of necessity a theory the scientific objects of which, nerves, are reducible, along with their properties, to systems of micro-physical particles in a sense which implies that all the predicates of an ideal brain theory would be definable in terms of micro-physical primitives none of which apply exclusively to micro-physical systems which are the theoretical counterparts of brains.
Thus, if the objects of brain-state theory are conceived to be reducible to micro-physical objects (however un-thingish) by an adequate micro-physical theory, the latter phrase must connote not "micro-theory adequate to the explanation of inanimate physical objects" (as it often tends to do), but rather "micro-theory adequate to the explanation of any physical object, animate or inanimate."
Thus it is my conviction that a theory which is to explain the properties of core persons will involve a family of families of
predicates which would be a categorial transformation, but not
substantive reduction, of raw-feel predicates, and which would
apply only to systems of scientific objects which are the theoretical
counterparts of empirical brains. Thus I accept the identity theory
only in its weak form according to which raw feels or sense impressions are states of core persons, and according to which, therefore, the logical space of raw feels will reappear transposed but unreduced in a theoretical framework adequate to the job of explaining
what core persons can do. In my opinion, such a theory is not yet even on the horizon.
The fundamental point I wish to make in defense of this thesis is that, if one thinks of "sense impressions" or "raw feels" as theoretical constructs introduced for the purpose of explaining "discriminative behavior" such as is found in white rats, then there is no reason to suppose that the postulated states might not be identified with neural states and conceived of as reducible along, the lines described previously. It is therefore crucial to my thesis to emphasize that sense impressions or raw feels are common sense theoretical constructs introduced to explain the occurrence, not of white rat type discriminative behavior, but rather of perceptual propositional attitudes, and are therefore bound up with the explanation of why human language contains families of predicates having the logical properties of words for perceptible qualities and relations.
Finally, it must be remembered that, although the framework of raw feels is a theory, the logical space of the attributes of raw feels is modeled on that of the perceptible qualities and relations
of physical objects and processes; and this logical space is, in an important sense, closed. Perceptible qualities and relations are, as the identity theory indirectly acknowledges, pure occurrent qualities and relations. They are neither dispositional nor mongrel states. To say of a physical object that it is red and triangular is
not to ascribe a power or disposition to it, though it is, in a very
strong sense, to imply that it has certain powers and dispositions.
Now it is not the logical spaces of occurrent perceptible qualities and relations which generate the demand for scientific explanation, but rather the logical space of the powers and dispositions of physical objects and processes. Here one must be careful, for there is clearly a sense in which the latter space is an off-shoot of the former. Roughly, it is not such facts, expounded in a "phenomenology" of sensible qualities and relations, as that to be orange is to be between red and yellow in color, which demand scientific explanation, as it is such nomological facts as that black objects sink further into snow than white objects when the sun is shining. And, when physical theory explains these causal powers by "identifying" perceptible things and processes with systems of micro-physical objects, the "identification" is not to be construed
as involving a reduction of perceptible qualities and relations to the qualities and relations of scientific objects, but rather as a correlation of these two sets of qualities and relations by means of "bridge laws." How this correlation is to be construed, and whether, unlike perceptible qualities and relations, the physical objects and processes which exemplify them can be reduced (i.e., identified, rather than merely correlated, with systems of scientific objects), are questions which transcend the scope of this essay, although they must be given an acceptable answer by a realistic interpretation of scientific objects such has been presupposed throughout the entire argument.5
1 This paper was prepared for and presented at the Boston University Colloquium for the Philosophy of Science, April 10, 1963.
2 It should not be assumed that the evolution of the use of a predicate is to be construed as a matter of its successively standing for a series of universals to which the concept of evolution does not apply. If, as I have argued in Chapter III above, universals are linguistic roles, they share in their own way the evolutionary and revolutionary vicissitudes of games and institutions.
3 Compare the trivial move from "shapes are properties of physical objects" to "shapes are identical with certain properties of physical objects."
4 Notice that this approach has the merit, at least, of meeting some of the objections to the instantially inductive form of the argument from analogy.
5 For an exploration of this problem, see Chapters VII and VIII above; also the early chapters in my Science, Perception, and Reality, Routledge and Kegan Paul, 1963.