Herbert Feigl, The "Mental" and the "Physical": The Essay and a Postscript (1967).
A. Review of the More Basic Meanings and Connotations of "Mental" and "Physical". Conclusions regarding their Respective Merits and Demerits.
The surveys and discussions of the preceding sections have paved the way for a summary and systematic appraisal of various characteristics which have been proposed as defining criteria of the mental and the physical. Outstanding candidates among the criteria of mind are (1) direct experience and (2) intelligence. "Direct experience" is synonymous with one sense of "subjectivity", viz. sentience, raw feels, or phenomenal givenness. "Intelligence" connotes learning capacity, purposive (goal directed), behavior and -- on the human level -- intentionality (symbolic behavior). Although the two criteria have in fact a certain area of coincidence, this coincidence (or overlap) is not a matter of logical necessity. By and large then, the two criteria of mentality define two entirely different concepts.
"Mind" as we have come to suspect all along, is an ambiguous term, or at best a group of concepts with family resemblances (in Wittgenstein's sense). The major components of the connotation of "intelligence" may be attributed not only to the higher animals but also to the "thinking machines" which we generally consider not only as lifeless but also as devoid of sentience. Direct experience, on the other hand, may well be attributed to some of the lower animals, babies, idiots, and to the severely insane; but in each of these classes at least some, if not all, of the marks of intelligence are lacking. Furthermore, it is customary in contemporary psychology to classify the unconscious (deeply repressed) traumata, anxieties, wishes, conflicts, etc. as mental. This again indicates that direct experience is not the criterion here, even if -- according to the psychoanalytic doctrine -- deeply repressed matters are potentially conscious, in that they can be brought to the fore of awareness by special techniques. Hypnotic and posthypnotic phenomena also often involve deeply unconscious processes, which because of their other similarities with the conscious processes are unhesitatingly classified as mental.
One might suppose that the term "physical" (to which we have paid thus far only sporadic attention) is much more definite in meaning than the term "mental". Unfortunately, the contrary is the case. There are some superficial and entirely inadequate definitions of "physical" which need only be mentioned in order to be promptly dismissed. For example, to define "physical" as the "outer" aspect (in contradistinction to the "inner" mental life) is to use misleading metaphors. "Inside" and "outside", "internal" and "external" have a good clear meaning in ordinary usage. What is literally inside, e.g., the skin of a person is most of his body (i.e., the body minus the skin) and that's "physical" in at least one very good sense of the term. After all, anatomy and physiology are concerned with the physical structure and the functions of organisms. Inside the skull is the brain of man, and that is "physical" in the same well understood sense.
Similarly unhelpful is the definition of the "physical" as the mechanical-compositional, as contrasted with the purposive-holistic. We have already repudiated this sort of definition-by-contrast, by pointing out that "mechanical" in the strict sense of "characterizable by the concepts and laws of Newtonian mechanics" designates only a narrow subclass of the class of physical events or processes, using "physical" (comprising also electrodynamic, relativistic, and quantum-theoretical characteristics) in the sense of modern physics. And if by "purposive" we mean no more than by "teleological" and "holistic", then there are innumerable teleological mechanisms, many of them with typical features of organic wholeness, both in nature and among the artifacts of technology. If "purposive" is understood in the narrower and more fruitful sense, then it involves intelligence (and this, on the human level, includes intentionality).
But the fact that there are (human) organisms functioning intelligently and displaying (symbolic) behavior which indicates intentional acts is describable in an intersubjective ("physicali") manner and therefore again does not support a definition-by-contrast between the physical (in this case physical) and the mental. It remains true, however, that among the objects and processes describable in physical terms, there are differences at least of degree (often of very considerable degree) if not of a fundamental, evolutionary-emergent type, as between the structure and the dynamics of electrons, atoms, molecules, genes, viruses, and unicellular and multicellular organisms. The tremendous differences between, e.g., a simple inorganic structure and a human being are therefore not in the least denied. As Castell (74) puts it, the solar system and an astronomer thinking about it, are in many essential respects very dissimilar indeed. (But the dualistic conclusions drawn by Castell seem to me nevertheless non sequiturs.)
The foregoing considerations suggest some of the more fruitful definitions of "physical". "Physical1 may be defined as the sort of objects or processes which can be described (and possibly explained or predicted) in the concepts of a language with an intersubjective observation basis. This language or conceptual system is -- in our sort of world -- characterized by its spatio-temporal-causal structure. This is so fundamental a feature of our world that it is extremely difficult to imagine an alternative kind of world in which intersubjectivity is not connected with this feature. One can understand, but need not concede, Kant's contentions regarding the synthetic a priori character of this "presupposition."1 The concept of "physical1" is closely related to but by no means equivalent with one of the primary meanings of "physical" in ordinary language, viz. observable by sense perception. In its most natural usage "observable by sense perception" clearly comprises the solid and liquid objects of our environment; it includes of course our own bodies; it includes a trifle less clearly the air (which can be felt if it moves with sufficient speed; or other gases if they can be smelled); it includes less obviously some of the dispositional properties of various sorts of matter (such as their hardness, elasticity, solubility, fusibility, etc.); and it scarcely includes electric or magnetic fields, atoms and electrons, or the secret thoughts of other persons.
But in one usage "observable by sense perception" does comprise the feelings, emotions, and even some of the (dispositional) personality traits of other persons. For example, we say, "I could see how disappointed he was." "I can see that he is a depressive person," etc. But these are usages which from the point of view of logical analysis are perhaps not fundamentally different from the case of a physicist, who (looking at a cloud chamber photograph of condensation tracks) says, "Here I see the collision of an electron with a photon." Such (extended) "observation statements" urgently demand a logical analysis into their directly verifiable, as contrasted with interpretive and inferential components. Logical analysis, pursuing as it should, an epistemological reconstruction, must therefore be distinguished from phenomenological description.
From the point of view of a phenomenological description, the "pre-analytic data" of the clinical psychologist contain his direct impression of (some of) the personality traits of his clients; just as the experienced physician's judgments may be based on his direct impression of the disease (diabetes, multiple sclerosis, Parkinson's disease, etc.) of his patient. Phenomenological description is a subtle and interesting matter, but philosophically much less relevant than it is often supposed to be. By a little exercise of our analytic abilities we can, and for epistemological purposes we must, separate the directly verifiable situation (the patient is very slow in all his movements, hangs his head, speaks with a very low voice; or: he has dry skin; his breath has a fruity smell; his hands tremble; etc.) from the inferential interpretations, i.e., the conclusions regarding his mental or physical illness.
Inasmuch as the use of terms like "psychoneurosis" is established, and diagnoses of psychoneuroses can hence be confirmed, on an intersubjective basis, the concept of psychoneurosis is evidently a physical concept. At least partial explanations of the behavior and the subjective experience of psychoneurotics have also been given on a physical (roughly: behavioristic) basis. We can plausibly explain neurotic dispositions by tracing them causally to the childhood situations of the patient (not necessarily neglecting some of his biologically inherited constitutional traits). And we can predict his anxieties, depressed moods, etc. on the basis of such intersubjectively confirmable information as, e.g., about a preceding period of highly "id-indulgent," overbearing, or hostile behavior. These "physical1" explanations do not differ fundamentally from explanations of, e.g., the growth of plants or the behavior of lower animals. That a plant grows poorly may be explained by the sandy soil in which it is rooted, the lack of rainfall, etc. The behavior (or some aspects of it) of an amoeba may be explained by the thermal and chemical conditions of its immediate environment.
The distinction between psychoneuroses and "physical" nervous disorders originates from the same commonsense considerations that have traditionally led to the contrast of "states of mind" and states of the body. No matter whether normal or abnormal processes are concerned, whenever scientifically or philosophically innocent people speak of something as being "in the mind" or "merely in the mind," this means apparently that it is not directly accessible to sensory observation. But, it is also positively characterized by the fact that these "states of mind" can (usually) be reported by those who have them, and that they can (sometimes) be influenced by talking. Sticks and stones cannot be made to move by merely talking to them.2 Persons (having minds!) can be made to do things by suggestions, propaganda, requests, commands, etc., often by just giving them certain bits of information.
But important and interesting as is this sort of difference, in its scientific aspects it no longer establishes a fundamental difference between inanimate things and minded persons. Modern robots have been constructed which emit information about their "inner" (physical!) states, and they can be made to do things by speaking to them. But if intellectually acute and learned men3 discuss seriously the problem as to whether robots really have a mental life (involving thoughts and/or feelings), there must be a question here that clearly transcends the obviously scientific and technological issue as to whether robots can be constructed which in their behavior duplicate all essential features (of course, one must ask: which ones and how completely?) of human behavior. If by "thinking" one means a performance starting with "input" premises and culminating in "output" conclusions which are deductively or inductively implied by the premises, and if the performance consists (at least) in certain observable relations between input and output; then there is no doubt that certain types of robots or computers do think. If one means by "feeling" what the logical (or illogical?) behaviorists mean, then it is at least conceivable (cf. Scriven, 304) that there might be machinelike structures (artificially made, or even naturally existing on some other stars) which behave (respond, etc.) in every way as it they had feelings and emotions.4 Here the question is clearly of the same logical nature as the queries: "Do butterflies feel?" "Do fishworms, when put on the hook, feel pain?" "Do plants have feelings?" "Do human embryos, four months old, have any direct experience?" I shall try to clarify the nature of these questions in the following subsection. For the moment it must suffice to point out that here we have to do with the distinction between "mental" (in the sense of sentience) and physical. Intelligence, in contrast to sentience, is clearly definable in physical1 terms. But as to whether sentience is so definable is perhaps the central perplexity among the mind-body puzzles.
But now to complete our analysis of the meanings of "physical": We have distinguished "physical1" and "physical2". By "physical1 terms" I mean all (empirical) terms" whose specification of meaning essentially involves logical (necessary or, more usually, probabilistic) connections with the intersubjective observation language, as well as the terms of this observation language itself. Theoretical concepts in physics, biology, psychology, and the social sciences hence are all -- at least -- physical1 concepts. By "physical2" I mean the kind of theoretical concepts (and statements) which are sufficient for the explanation, i.e., the deductive or probabilistic derivation, of the observation statements regarding the inorganic (lifeless) domain of nature. If my conjecture (discussed above) is correct, then the scopes of theoretical "physical1" and "physical2" terms are the same! If, however, there is genuine emergence, i.e., logical underivability, in the domains of organic, mental, and/or social phenomena, then the scope of "physical2" terms is clearly narrower than that of "physical1" theoretical terms.
Within the category of "physical1" terms, it is clearly important to distinguish observation terms from theoretical terms: and among the latter several levels may methodologically, if not logically, be distinguished. For example, the concepts of classical thermodynamics form one level, and the concepts of statistical or molecular mechanics (in terms of which those of thermodynamics, with certain modifications, can be defined) form a "higher" level. The concepts of molar behavior theory are related analogously to those of the higher level of neurophysiology; and so on mutatis mutandis, throughout the various fields of scientific theories.
We conclude that to say "x is physical" is highly ambiguous. There is first the obvious distinction between the physical languages (physical language designators) and physical objects (physical language designata). This distinction carries through the two further distinctions and does not, for our purposes, require elaborate discussion. To illustrate, an electromagnetic field, just as the planet Jupiter, are designata of physical language terms. However, the observation terms of the physical1 language serve also as the evidential basis of the physical1 or physical2 theoretical languages. Theoretical terms are here conceived as not explicitly definable on the basis of observation terms. (cf. Carnap, 73; Feigl, 110; Sellars, 315), but as specified by postulates and by correspondence rules relating them to the terms of the observation language. And, to restate this in different words, if there is no genuine emergence in the logical sense above the level of lifeless phenomena, then there is no basic distinction between the theoretical terms of the physical1 and physical2 languages. That is to say that the theoretical terms of biology and psychology are explicitly definable on the basis of the theoretical concepts of physics in the same sense as the theoretical terms of chemistry (e.g., the chemical bond) are nowadays explicitly definable on the basis of. the theoretical terms of the physical language (i.e., of the atomic and quantum theories).
The central questions of the mind-body problem then come down to this: are the concepts of introspective psychology -- relating to phenomenal data or phenomenal fields -- definable on the basis of physical1 theoretical terms, and if so, are they also definable on the basis of physical2 (theoretical) terms? The first question is a matter for philosophical analysis. The second question is, at the present level of scientific research, undecided, though my personal (admittedly bold and risky) guess is that future scientific progress will decide it affirmatively. We turn now to a discussion of the first question primarily, but occasional remarks about the second question will also be ventured.
Notes 1. For a criticism of this rationalistic position, cf. Pap (242); Nagel (233); Reichenbach (275); Feigl (114).
2. This still seems safe to assert even in view of the alleged but highly questionable "facts" of psychokinesis.
3. Cf. Turing (338); MacKay (216); Spilsbury (326); Scriven (304).
4. The question in this form is by no means new. William James discussed it in his Principles of Psychology (Vol. I) by means of the example of the "automatic sweetheart." He was severely criticized by E. A. Singer (318) who, ironically enough, appealed to James' own principle of pragmatism (derived from Peirce's meaning criterion which anticipated the essentially equivalent later operationist and logical positivist formulations of thecriterion). But E. A. Singer in turn was incisively criticized by the much more subtly linguistic behaviorism expounded by Gilbert Ryle's The Concept of Mind.