Herbert Feigl, The "Mental" and the "Physical": The Essay and a Postscript (1967).
A. Is There a Phenomenal Language? The Relations of Meaning, Evidence, and Reference.
The central core of the proposed solution rests upon the distinction between evidence and reference. No matter what indirect (behavioral) evidence we use for the ascription of mental states, the mental state ascribed is not to be confused with the evidence which only lends support to the ascription. A fortiori, we must eliminate the still worse confusion of the pictorial appeals (attached to evidential terms) with the conceptual meaning or the reference of neurophysiological concepts. The only case in which pictorial appeals or imagery may be thought to play an essential role in knowledge claims is at the ultimate phenomenal basis of the confirmation of all knowledge claims. And, as we have pointed out, if and only if these knowledge claims are so extremely restricted as to refer exclusively to a currently experienced datum, then -- in this very special case -- evidence and reference coincide. "Now green", "now anger", "now green spot on a gray background", "stinging pain suddenly increasing", etc. might be examples. The last example shows that the indexical term "now" need not appear in the phenomenal sentence; but of course the sentence is in the present tense, and this is presumably equivalent with the occurrence of the indexical "now".
It is difficult to decide whether indexical terms (i.e., egocentric particulars like "now", "I", "here", "this") are indispensable constituents of singular phenomenal sentences. There are, of course, many examples of universal statements which contain only phenomenal terms as descriptive signs (in addition to purely logical signs): "Orange is more similar to red than it is to green"; "Whatever is colored is extended (in the visual field)"; "Anger always subsides after some time"; etc. There is also the difficult question whether phenomenal sentences can contain proper names (or something like topological coordinates) for elements in the phenomenal fields. One of my examples suggested that one might use proper names for the small bright spots on the dark background of a visual field and thus describe their relative positions in terms of such relations as "to the left of", "above", and "far below". It seems clear that there is a danger of logical paradoxes, engendered by category mistakes, if we try to mix phenomenal sentences of this sort with the usual behaviorally based ascriptions of mental states to organisms. In these behavioral ascriptions the organism (or the person?) is the individual which is represented by the subject term of the sentence; the predicate is then something like "sees green", "sees an array of bright spots on a dark background". There can then be no direct translation of sentences in which the subject terms denote elements in a phenomenal field, into sentences in which the subject terms denote individual organisms. But perhaps there can be an empirical coreference between statements about some (configurational) aspects of neural fields and those about phenomenal fields.
The precise logical explication of empirical identity or coreference is fraught with many difficulties. Some of these stem from the tendency to think of meaning as intension, and then to conceive of intension in terms of its simplest picturable examples. Blueness is an intension indeed, but what are the intensions of "energy", "entropy", "electric field strength", "electric charge", "neuron discharge", "reverberating neural circuit"? In all these other cases the intensions are non-intuitive and can be specified only by postulates and correspondence rules. Similarly non-intuitive are the elements of the corresponding extensions, or the denotata. It does seem to me that we can rightly say that both the intension and the extension of the theoretical concepts of the physical sciences are largely unknown by acquaintance, and that only a very small selection of them can therefore be identified with the intensions and extensions of concepts-by-acquaintance. But of course the latter presuppose the existence of a phenomenal language. It has indeed been seriously questioned as to whether there is a phenomenal language at all. In the usual, and full-fledged sense, "language" means a symbolic system with specifiable syntactical (formation and transformation) rules, semantical (designation) rules, and pragmatic (verification) rules. Scraps and bits of phenomenal phraseology seem to fulfill these requirements, but an overall system like that of the physical language does not seem attainable.
The difficulties are further complicated by the question on which level of analysis we are to specify elements and relations described by phenomenal sentences. There is a long history of objections against the Hume-Mach-Russell-Price analysis of experience into "hard" and "soft" data. Phenomenologists, Gestalt psychologists, and more recently many analytic philosophers have raised serious objections not only against the atomism or elementarism of the sense-data doctrine, but also against any doctrine of immediacy or of the given.1
I have throughout this essay maintained and argued that genuinely phenomenal or acquaintance terms are indispensable, not only for the reconstruction of the indirect confirmation of practically all our knowledge claims, but also as labels for the referents of some knowledge claims -- whether they are about my own raw feels or those of other humans or animals. I have allowed for the possibility that the "hard data" (i.e., those data which we can talk about with a minimum of inference) are not preanalyticallv but only postanalytically "given." But on just what level of psychological, introspective, phenomenological, or logical analysis we find those data which stand in the required one-one correspondence to neural events, is an open question. With W. Köhler I am inclined to think that an analysis which stops at a relatively simple configurational level (but does not proceed further to "atomize" the given) may well yield the desired items on the ψ-side of the ψ-φ isomorphism. But phenomenal description, even of the Gestalt type, is no easy matter.
Notes 1. For some impressive arguments against atomism see W. Köhler (183, 184, 185), Brunswik (56), Wallraff (340); and against immediacy, Lean (193), Chisholm (75), Wittgenstein (357), Rhees (278), Quinton (270), W. Sellars (315). Others like Ryle (294), Black (38) and Quine (268) have denied the possibility of a phenomenal language altogether. W. Sellars admits phenomenal concepts only as theoretical terms in a language of behavior theory.