Herbert Feigl, The "Mental" and the "Physical": The Essay and a Postscript (1967).
A. "Subjective" versus "Objective".
The juxtaposition of "subjective" and "objective" has been the source of endless and badly confused controversies throughout the ages. There is nevertheless something significant and worth preserving in this distinction. To say that a twinge of pain experienced by person A is "subjective" or "private" to him may simply mean that another person B, observing A's behavior, may infer A's pain, but does not have it, i.e. he does not directly experience it. Dentists do not have the toothaches of their patients. In one sense this is clearly analytic (tautological).1 It is analytic for reasons analogous to those which make it self-contradictory to say that I am growing my wife's hair. (Schizophrenics are known to make assertions of this sort.) "I am eating with my wife's teeth" is merely funny, but not self-contradictory. "Dentists always suffer toothaches when their drill comes near the pulpa of their patient's tooth" is synthetic, but empirically false. "I am listening through my wife's ears" if meant literally (not metaphorically) is a border line case, depending on specific detailed interpretation. "I am enjoying Mozart's music exactly as my wife does" is synthetic and may even be rendered as "I have the same musical experience as does my wife." (Remarks about the two meanings of "same" will follow presently.)
The case is a trifle more complex for perception. Two persons sitting next to each other in the concert hall are said to hear the same music, or at a given moment the same tones or chords, produced by the pianist on the stage. But the facts of the case are really not fundamentally different from the first example. A does not have B's musical experience (or vice versa), even if their auditory discrimination, musical appreciation, etc., does not differ in any discernible way. They may be said to hear the same sounds, to be both equally impressed or thrilled by them; but common sense as well as scientific reasoning clearly indicates that their experiences are numerically different. Fundamentally this case does not differ from, e.g., the case of two thermometers immersed next to each other in the same glass of water. It is perfectly proper to say that these instruments indicate the same physical condition. It is also perfectly proper to say that the two thermometers not only indicate but also "have" the "same" temperature. (This is logically quite like saying that two marbles have the same color.) But it would be most improper and paradoxical to say that the events taking place in the one thermometer are identical with those in the other. This is not the place for a discussion of Plato's problem of the "one and the many." Suffice it to point out that the phrases "the same as" and "identical with" are ambiguously used. "Sameness" or "identity" may mean complete similarity, as in the case of the two musical experiences, or in the case of the two thermometric indications. But "sameness" or "identity" in other contexts means the numerical oneness of the individual referent of, e.g., two different names, or of two different unique characterizations (Russellian descriptions). I conclude then that it makes perfectly good sense to speak of the subjectivity or privacy of immediate experience. Numerically different but qualitatively identical (indistinguishable) experiences may be had by two or more persons, the experiential events being "private" to each of the distinct persons.
Terminological trouble, however, arises immediately when we take a scientific attitude toward direct experience and try to confirm, describe, or explain it "objectively." Is it not an "objective" fact of the world that Eisenhower experienced severe pain when he had his heart attack? Is it not a public item of the world's history that Churchill during a certain speech experienced intense sentiments of indignation and contempt for Hitler? Of course! What is meant here is simply that statements about facts of this sort are in principle intersub/ectively confirm-able and could thus be incorporated in a complete historical account of the events of our universe. To be sure, there are cases in which confirmation is practically outright impossible. The last thoughts and feelings of a man immediately before his death, especially in a case of complete paralysis, or of death occurring through electrocution, may be inferable only with scant reliability. But this is not different from the case (cf. Carnap, 67, p. 419f) of the confirmation of the electric charge of a specific raindrop that fell into the Pacific Ocean in a place far removed from any observers. Our current liberal formulation of the empiricist meaning criterion countenances all statements of this sort as perfectly meaningful. They do not fundamentally differ from other less difficult-to-confirm statements about, e.g., the "true thoughts" of a liar or play actor. Modern devices, such as the lie detector, and various clinical-psychological techniques enable us to test for such "private" events with increasing (though generally only relatively low) reliability.
The foregoing considerations suggest that the terms "subjective" and "private" at least in one of their commonly proper and serviceable usages are not to be considered as logically incompatible with "objective" or "public" in the sense of "in-principle-intersubjectively confirmable". Private states in this philosophically quite innocuous sense are then simply central states. (Whether these are ultimately to be conceived mentalistically or neurophysiologjcally may be disregarded for the moment; but this will of course be discussed quite fully later.) "Subjective" or "private" in this sense may then designate the referents of direct introspective reports, and it will be understood that these same referents may well be more indirectly characterized by descriptions involving inference from behavioral symptoms or test results of experiments on behavior. In those cases of subhuman animal behavior in which we don't hesitate to speak of experienced pains, gratifications, rage, expectations, etc., there are of course no introspective reports. But other aspects of such behavior are in many respects so similar to the human case that the ascription of raw feels is usually justified on the basis of analogy. Here again, the "private" means the central state which causally effects (or at least affects) the overt and publicly observable behavior.
The terms "subjective" and "objective" are indeed mutually exclusive if they are used in a quite familiar but different way. In designating some impressions, opinions, beliefs, value judgments, etc. as "subjective," we sometimes contrast them with the "objective truth," or "objective reality." If, e.g., my friend maintains that the room is cold, I am inclined to argue with him by pointing to the thermometer (which reads, say, 74°); and perhaps by explaining his "impression" by the fact that he is too scantily dressed, or that he is sick, or suffers from anxieties, etc. Similarly in the more drastic cases of dreams, illusion, delusion, etc. we criticize some (interpretive) judgments as based on "merely subjective" evidence. And it should go without saying that disagreements in aesthetic value judgments may often be explained on the basis of individual or cultural differences. "De gustibus non est disputandum" is our final resort if no objectively justifiable standard can be agreed upon.2 But wherever beliefs can be criticized as, e.g., "biased," "too optimistic," "too pessimistic," etc., there are standards, such as those of normal inductive inference, which may indeed justify the rejection or correction of such "all too subjective" convictions. Here "subjective" and "objective" are indeed incompatible, although of course there may well again be an "objective" explanation of the genesis of "subjective beliefs."
There is, however, also a philosophical and speculatively extended sense of "subjective" or "private". In this very special and highly problematic sense it is assumed that there may be subjective states which are in principle inaccessible to intersubjective confirmation. Here we had better speak explicitly of "absolute subjectivity" or "absolute privacy." It is this sense which is entertained in some of the more radically interactionistic forms of dualism. And it is this sense which by definition is incompatible with "objectivity" understood as intersubjective confirmability. As I have indicated before, I no longer insist that a doctrine involving the notion of absolute privacy is entirely devoid of cognitive meaning. But I am inclined to regard it as scientifically meaningless. To recapitulate: if the scientific enterprise is defined as necessarily requiring inrersub/ective confirmability of knowledge claims, then this follows immediately and quite trivially.
Now, I think it is an essential aspect of the basic working program and of the working hypotheses of science that there is nothing in existence which would in principle escape intersubjective confirmation. Allowances have already been made for the (sometimes) insuperable practical difficulties of even the most incomplete and indirect confirmations. But the optimistic outlook that inspires the advance of science and informs its heuristic principles,9 does not tolerate the (objectively) unknowable or "un-get-at-able." No matter how distant, complicated, or indirect the connection of scientific concepts with some (intersubjective) evidential bases may be, they would not be concepts of empirical science (as contrasted with the concepts of pure logic or mathematics) unless they could in some such fashion be "fixed" by "triangulation in logical space." The "fix" we are able to obtain may be as indefinite as it is when theoretical concepts (like those of the positron, the neutrino, or the meson in physics; that of the unit of heredity; or of memory traces; of the superego, of general paresis, or of schizophrenia in biology, psychology, or psychiatry) were first tentatively introduced by only very sketchily formulated postulates. The concepts of absolutely subjective or completely private data, however, are so conceived that they can be applied only on the basis of the direct experiences contained in a given stream of consciousness. A completely "captive mind"10 might experience senselike qualities, thoughts, emotions, volitions, etc., but they would (ex hypothesi) not in any way, i.e., not even through weak statistical correlations, be connected with the publicly observable behavior or the neurophysiological processes of an organism.
While it is difficult to spin out this yarn in a consistent (let alone plausible) fashion, I do not think it impossible, in the sense that it would necessarily involve some self-contradictions. There are philosophers who have been concerned with an analysis of the meaning of the "continuance of a pure (immaterial) stream of experience after bodily death"; or with the problem of the "inverted spectrum" (Could pure sensory-like qualia like red and green, blue and yellow, be systematically interchanged for different persons, despite a complete similarity in their discriminatory and linguistic behavior, as well as in their neurophysiological processes?). Speculations of this sort were declared taboo and absolutely meaningless by the early logical positivists. They were compared with assertions about absolute space and time, the (Lorentzian) ether, the "bond" between cause and effect, or the existence of a metaphysical substance, over and above anything that could be verifiably known by science about spatio-temporal relations, coordinate transformation, functional relations between observable properties or measurable magnitudes, or relations of compresence of various observable properties. There is no doubt that this positivistic cleansing of the Augean stables of metaphysics had a most salutary effect. But positivists (temperamentally often negativists), in their zeal and eagerness to purge the scientific enterprise of meaningless as well as superfluous elements, have often overshot their goal. A redressing of the balance has become necessary, and we pursue nowadays responsible analyses of, e.g., causal necessity11 which are perfectly compatible with the basic antimetaphysical insights of Hume. Similarly as regards the notion of absolute privacy, it is illuminating to conceive it at least as a logical possibility, and then to state as clearly as feasible the reasons which can be adduced for rejecting the idea for our world as we have come to conceive of it in our science to date.
The notion of absolutely private data of experience, if such data are to be described, would require a purely phenomenal or absolutely private language. Such a language, by definition and ex hyporhesi, could not serve as an instrument of communication. Even a completely solitary humanlike individual could not engage in audible (or visible, etc.) symbolic activities. Not even soliloquies in this physically expressible form would then be possible. For ordinary soliloquies, amounting to more than the unexpressed thoughts of a private thinker, are expressible, and the very expressions would provide (no matter how unreliable) clues to the "inner" thought processes.12
Now, of course, if by a "language" one means what is customarily meant by it (viz., an instrument or vehicle of intersubjective communication), then an absolutely private language is ruled out by definition. Language as we know it and use it is indeed not absolutely private in the sense explained. But that it is intersubjective reflects a basic empirical feature of our world, or at least a basic feature of our-world-as-we-conceive-it in common life and in science. But I must postpone discussion of the fuller implications of this feature until I present my denouement of the "world knot" in the final section. For the present I submit that by a "language" one is not compelled to mean an instrument of interpersonal communication. The idea of the soliloquy (intra- personal communication) may be restricted and modified in such a manner that it refers to unexpressed and inexpressible thoughts. This preserves a sufficient "family resemblance" with the ordinary notion of language. Such an absolutely private language would still enable the solitary thinker silently to label the qualities of his direct experience and to think silent thoughts which have the logical form of declarative (singular, universal, etc.) statements. I could, for example, with the help of remembrance, think that extreme anger always gradually subsides, that a given tone-as-heard is increasing in intensity, etc. Knowledge thus formulated in a private language may well be called "knowledge by acquaintance." It is true that ordinary discourse entertains a much wider conception of knowledge by acquaintance. There it covers knowledge based on, and not essentially transcending, the observations (amplified by very moderate and limited inferences). Thus we can quite properly say that we know the properties of sticks and stones, of apples and oranges, the manners and mannerisms of our close friends "by acquaintance."
But this ordinary concept of acquaintance is not very sharply defined. Having actually seen Winston Churchill for a few seconds (when on July 10, 1954, he emerged from 10 Downing Street in London and entered his black limousine, holding his cigar and waving to the assembled small crowd), am I entitled to say that I know him "by acquaintance"? Would I know Churchill "by acquaintance" if I had seen him (or rather his image) only in the cinema newsreels? I leave it to the linguistically more sensitive and subtle Oxford analytic philosophers to decide these questions, or else to tell me that "knowledge by acquaintance" is a hazy notion, involving "slippery slopes" in various directions. (Anyway, the latter alternative is what I consider the best analysis of the ordinary usage of the term.)
For a philosophical usage of the term, however, I suggest that "knowledge by acquaintance" be understood as knowledge involving no inferential components -- or, if this be chimerical, then knowledge involving only that minimum of inference which is present when only memory is^utilized for the recognition of similarities and differences. It is in this sense that I could assert on the basis of acquaintance, "Ah, there is that peculiar smell again; I don't know what causes it, I don't ever know how to label it; it is so different from any fragrances of flowers perfumes, cigar smoke, burnt toast, tangerines, etc. that I can't ever place it in a multidimensional scheme of the rank orders of smells; but I know I have experienced this smell before and I am (subjectively) sure I would recognize it in the future if I were to experience it again."
As I have said earlier, I make no claim for the infallibility of knowledge by acquaintance. Our world, being what it is, is such that corrections of subjective-experience judgments (knowledge claims made on the basis of direct acquaintance) are definitely possible from the vantage point of intersubjective observation. Moreover, it should require no reminder that I quite emphatically want to distinguish acquaintance from knowledge by acquaintance. "Acquaintance as such" (in the philosophically restricted sense) is to mean simply the direct experience itself, as lived through, enjoyed, or suffered; knowledge by acquaintance, however, is propositional. Knowledge claims of any sort may be valid or invalid; the statements which formulate such knowledge claims are either true or false. In the case of practically all knowledge claims which have scientific status, the confirmation of their truth is incomplete and indirect. Knowledge by acquaintance, however, is direct and complete in the following sense: it seems utterly inappropriate to ask someone what his evidence is for asserting that he, e.g., feels at the moment elated, depressed, anxious, dizzy, hot, cold, and so on through the various modalities and qualities.
The philosophically much misused and over-exploited term "self-evident" might well be redefined and restricted to just such reports of immediate introspection or self-observation. With this, possibly unwise, terminological suggestion I do not wish to imply any doctrine of "incorrigibility" in regard to such protocols of immediate experience. I grant that even such protocol statements may be in error; and not only for the generally admitted reasons such as possible slips of the tongue or the pen; but also because the predicates or relational words used in such statements, if they are what they are intended to be, viz. universals, presuppose for their correct application even in the "absolutely private" language (as fancied above) at least the reliability of memory. This alone would ensure that the same term is applied to an experienced quality of the same kind as before. Otherwise a protocol statement would simply amount to what would in effect be a first introduction of the predicate in question by stipulative-ostensive definition:13 i.e., it would amount to the resolution to use the same term on future occasions sufficiently similar to the present one. But on the occasion of the first use of a new term, the sentence containing it would be true only in the extremely restricted (very much like analytic) sense that "A," the label which I arbitrarily apply to the completely and incomparably new fragrance that I am just experiencing, designates the quality experienced during each of the moments of its temporary occurrence of finite duration.
There are other uncertainties besides the ones mentioned in the use of (available) predicates for the qualia of immediate experience. Am I to describe the way I feel at a given moment as "happy", "joyous", "merry", "gay", "frolicsome", "blithe", "debonair", "light hearted", "buoyant", "bright", "animated", "gleeful", "hilarious", "jolly", or what?
It is time to draw some conclusions from this discussion. There is one meaning of "mental" in which it coincides with one meaning of "subjective". Let us call this meaning "phenomenal". In so calling it we may leave for later the question as to whether what is phenomenally given and phenomenally labeled is always also indirectly characterizable in an intersubjectively meaningful terminology. In any case we have isolated one contrasting (though not necessarily incompatible) pair of meanings for "mental" arrd "physical": the phenomenal (i.e., the subjectively confirmable) and the intersubjectively confirmable (i.e., the physical in the terminology suggested above). The meaning of "mental" (synonymous with "phenomenal") looms large in introspective and phenomenological psychology. It is also prevalent in Gestalt-psychological descriptions of the configurations in phenomenal fields.
But in the "depth-psychological" statements of the psychoanalytic schools of thought, "mental" includes also subconscious, and some unconscious, states and processes. Since these are described largely with the help of metaphors and similes taken from the phenomenal (disregarding here those from the physical, e.g., mechanical, hydraulic, etc.) sphere, and inasmuch as detailed neurophysiological descriptions are still lacking, it will be well to remember that the word "mental" as commonly employed by present-day psychologists covers both phenomenal and non-phenomenal states and events. The justification for the inclusion of the subconscious ("preconscious") and the unconscious in the realm of mind comes of course from some other attributes traditionally considered as criteria of mentality. We shall turn to those other attributes. The one which (for philosophical-historical reasons) will be taken up first is, however, not as essential in this connection as are some of the others further down the list.
Notes 1. This is now even admitted by Ayer (18) who had earlier (15) held it was synthetic. His earlier position was, however, incisively criticized by Pap (243, 248) and Wating (34).
2. On the meaning and the limits of the justification of norms, cf. my essay (109).
3. Some philosophers rather speak of them as "metaphysical presuppositions"; for my criticisms of this interpretation of science cf. (110, 114).
4. The idea and the phrase are Hilary Putnam's.\
5. Cf. Burks (59); W. Sellars (312, 313, 314).
6. For an extremely lucid and succinct discussion of this point cf. P. E. Meehl (219).
7. The notion of "ostensive definition" is of course highly problematic. In contradistinction to what "definition" (explicit, contextual, recursive, abstractive, conditional, coordinative, or even implicit) generally means, ostensive definitions cannot be rendered in speech, writing, or printing. "Definition" in its normal use always means specification of the meaning of some symbol by recourse to the meanings of other symbols. "Ostensive definitions" (if this phrase is to be retained at all) had therefore better be regarded as the establishment or acquisition of a linguistic habit, the inculcation of a bit of rule-governed linguistic behavior. In an absolutely private language it may amount to the stipulation of a rule which associates certain thoughts or images with specific other items or aspects of direct experience.