Herbert Feigl, The "Mental" and the "Physical": The Essay and a Postscript (1967).
F. "Intentional" versus "Non-intentional".
The mental life of (at least) the adult homo sapiens is characterized by the capacity for awareness -- in addition to the occurrence of mere raw feels. (We credit some animals and certainly young children with the latter in any case.) To have an experience, and to be aware of having it, is a distinction which I think cannot be avoided, even if in a given case it may be very difficult to decide whether awareness actually supervened. This is one of the notoriously difficult questions of phenomenological description. But assuming the distinction, it is fairly plausible that awareness is impossible without some sort of symbolism, even if it be the "silent" symbolism of imagery or (if there be such) of imageless thought. It is here where the idea of "intention" (not in the sense of purpose, end-in-view, or resolution, but) in the sense of reference becomes essential.
I shall try to show that the scientifically relevant issues regarding in-teractionism versus parallelism (or epiphenomenalism) should be carefully separated from the philosophical issues which stem from the "intentional" features of mind, stressed by Brentano and the phenomenological schools of thought. According to this point of view the most fundamental difference between the mental and the physical consists in the fact that the mental life consists of acts directed upon objects, no matter whether these objects exist in the world, or are pure concepts, or figments of the imaginations. It is true that dualism in the Cartesian tradition has emphasized the intentional as well as the raw feel features of mind. The mind-body problems in the larger sense therefore have customarily included such questions as, Can we give a physical (1 or 2, in this case) account of how thoughts, beliefs, desires, sentiments, etc. can be about something? Can we give a naturalistic translation of the language of reasoning as it occurs in arguments, i.e., discourse in which we give reasons intended to support knowledge claims, or value judgments? I think it has become increasingly clear1 that the answer must be in the negative; but not because human behavior involving "higher thought processes" is not in principle capable of physical (at least physical1 explanation and prediction; but rather because the problem is one of the logical reducibility or irreducibility of discourse involving aboutness (i.e., intentional terms), to the language of behavioral or neurophysiological description. Now it seems fairly obvious that such discourse, just like discourse involving oughtness (i.e., normative discourse) is not logically translatable into purely factual statements. The relation of designation (formalized in pure semantics) is not an empirical relation, but a construct of semantical discourse.
Personally, I therefore consider the problem of intentionality not as part of the psycho-physical but rather as a part of the psycho-logical problem, i.e., as part of the relation of psychological to the logical forms of discourse. This becomes even more evident because, assuming the ultimate possibility of a full neurophysiological account of behavior (including linguistic behavior), we should then have the problem of relating the physiological to the logical forms of discourse. If many writers permit themselves nowadays to speak of "thinking machines" (electronic computers, chess playing machines, etc.), then it is equally justified to pursue the problem of the relation between the mechanical (or the electrical) and the logical. In the case of the machines, it is ourselves who have built them in such a way that in their functioning they conform to certain rules of logical, mathematical, or semantical operations. In the case of human beings we have nervous systems which through education and training acquire the dispositions toward certain types of symbolic behavior which in actual operation then is more or less in conformity with certain rules.
But the abstract statement of a rule is not to be confused with the formulation of the (statistical) empirical regularity of the symbolic behavior. An illicit inference or a computation mistake is a violation of a rule, it is not an instance which would disconfirm a law of behavior. The recent phase of the clarification of these issues was in essence initiated by Husserl and Frege in their critique of psychologism, i.e., of the confusion of logical with psychological discourse. The pan-empiricist position of, e.g., John Stuart Mill who regarded logical truths as on a par with the truths of the natural sciences, was thus effectively and definitively refuted. Later, very much needed refinements of the anti-psychologistic position were added by Carnap (65, 68, 69, 71, 72), and a full study of the logical status of rules and rule-governed behavior has been contributed by W. Sellars (loc. cit.).
No matter what the most clarifying analysis of rule-governed symbolic behavior in its relation to the rules as such may turn out to be, there can be no doubt that if physical (at least physical) determinism is to be maintained, the following will have to hold: A person's brain state when thinking, e.g., about Napoleon's defeat at Waterloo must qualitatively or structurally differ from the brain state of the same person (or, for that matter, of other persons) when thinking about Caesar's crossing of the Rubicon, This aspect of the psychology and physiology of thought is definitely relevant for our problem.
Notes 1. Cf. especially Wilfrid Sellars (310, 311).