Talk at SIU Philosophy Colloquium, Feb. 19, 2009

Reductionism and Discourse Relativity

D. S. Clarke

We find throughout Hilary Putnam’s writings many insightful remarks. One of the most important of them questions the meaningfulness of mind/body identity assertions as the contemporary way of formulating a physicalist metaphysics. To say that an experience “is ‘identical’ with a cortical event,” Putnam says, “trades on a notion of ‘identity’ that seems to me entirely meaningless.”[1] This claim is puzzling, for it has the effect of repudiating the reductionist project known as “functionalism” that we associate with Putnam through his earlier writings. To my knowledge, Putnam never explains why he has changed his mind. This paper is an attempt to provide the explanation he never gives and some qualifications that need to be added. I do this by first outlining in a preliminary way some basic concepts, then applying them to various kinds of identity assertions, including the metaphysical identities questioned by Putnam, and finally drawing some general conclusions.

Some Preliminaries

First, a quick review of the reasons philosophers have had for asserting the type of mind/body identities now being rejected by Putnam. I touch a hot stove, feel an intense pain, and withdraw my hand. It seems obvious that the pain I feel is the intervening middle term of a causal sequence that begins with my touching the stove and ends with my hand withdrawal. Touching the stove causes the pain, which in turn causes the withdrawal. Now neurophysiologists can describe in some detail the brain processes activated by stimulation of pain receptors and causing the processes within efferent nerves that activate muscular responses. By identifying the pain I feel with these brain processes, we can provide an explanation of why the pain can be said to cause my hand withdrawal. Without this identification, the causal relation would remain the mysterious interaction postulated by dualist theories. If we let X be the description provided by neurophysiologists of the brain processes occurring when we feel pain, it seems plausible to assert that the pains we feel when touching hot objects are identical with brain process of type X, or more generally, that sensations are identical with brain processes, with the type of brain process varying with the type of sensation, whether a pain, a pleasure, or the having of some sensory image. Such identities are what we understand as “reductions” of the sensations to the physical processes. This at least was the original formulation of the mind/body identity thesis by U. T. Place and J. J. C. Smart. It was subsequently modified by Putnam and others on the grounds that it is conceivable that a person could undergo a surgical replacement of the portions of his brain processing signals from pain receptors with silicone transistor substitutes. Such a person could conceivably experience the same type of pain after surgery as before, though the physical processes would be very different. At most we can say that the particular sensation a person feels is identical with a particular brain process, but that pain as a type of sensation is subject to different physical realizations. If so, a “token-token” identity between a particular sensation and a particular brain process can be asserted, but we can only say that a type of sensation “supervenes” on a type of brain process, where the relation of supervenience assumes the existence of particular token-token identities.[2] The complications introduced by this revision of the identity theory can be ignored for our purposes, however, for it seems that Putnam is branding as meaningless all assertions of identity between the mental and the physical, whether between types of events or processes, or between particular events or processes.

His rejection of metaphysical identities would seem to also extend to extensions of the identity thesis to more stable mental attitudes such as beliefs, hopes, and wants. Why did Jones carry his umbrella this morning? One explanation is that he believed it would rain, and this explanation can become a causal explanation if Jones’s belief is identified with a brain state as part of a complex of causes of his behavior. As for sensations, to assert an identity between a belief that p and a brain state of type Z is in some sense to “reduce” the mental state to that brain state, and the reduction is thought to be important in enabling causal explanations of behavior. An identity used for this reductionist purpose would also be meaningless for Putnam.

Identities used to reduce the mental to the physical in the way just described have one feature that distinguishes them from standard identities. A standard identity of the form a = b (‘Mark Twain is Sam Clemens’) is symmetric: a = b if and only if b = a. In contrast, reductionist identities establish a priority for the physical over the mental. The sensation a person reports or an attitude she expresses is claimed to be nothing but what is described by the appropriate description of a brain process or state, and the relation of being nothing but is clearly asymmetric. We can say that a pain is nothing but a brain process of type X, but no one would say that the brain process is nothing but the pain. This contrast to standard identities should initially give us pause. The more fundamental reasons for problems with metaphysical identities, however, become clearer when we consider features of sentences within discourse contexts.

Sentences are typically used within discourses as combinations of sentences about some common topic or topics, whether the discourses of natural (ordinary) language conversations or the specialized discourses used to communicate within the social institutions of science, law, religion, and the arts and literature. Every discourse is in turn formulated within a discourse framework from which it derives the standards used to evaluate what is endorsed. To take some obvious examples, the empirical standards used to evaluate the truth or falsity of a scientific theory are very different from those used to assess whether a theorem follows from the axioms of a mathematical system, from the moral conclusion that a person ought to perform a certain action, the judicial conclusion that a person has a right to some area of privacy, or a minister’s declaration to a congregation that God loves humanity. These differing standards of evaluation are a function of the different purposes served by science in describing matters of fact, mathematics in deriving conclusions from axioms and definitions, morality in guiding conduct, the judiciary in applying laws, and religious institutions in providing hope and a sense of security and self-worth. It is common for writers to describe discourse frameworks as different languages, and discourse frameworks are characterized by their distinctive vocabularies. This description can sometimes be useful, but ignores the fact that there is much overlap in the vocabularies of different frameworks, for expressions migrate from languages of specializations to the shared natural language of the linguistic community and then are incorporated into other specialized forms of discourse. More distinctive of frameworks are the standards of evaluation we apply and the manner in which nouns are introduced into the vocabulary characteristic of a framework.

Within a given framework, we can often distinguish its sub-frameworks, and these sub-frameworks will usually have distinctive vocabularies, while sharing the evaluative standards of the general framework to which it is subordinate. For example, the framework of empirical discourse descriptive of our environment includes as sub-frameworks the standard descriptive discourse framework of everyday life and those of the specialized natural sciences. Each has its distinctive vocabulary, though with overlaps between them. In addition, sub-frameworks may be distinguished by their mode of term introduction. Theoretical terms are often introduced in the sciences by postulation for explanatory purposes, with experimental identification of referents coming later. Such introduction is different from the baptismal introduction of nouns such as ‘gold’ and ‘water’ of natural (ordinary) language. But despite these differences, everyday descriptive discourse and the sciences share observational standards for reaching consensus about truth falsity, and this establishes the more fundamental contrast between them and discourses within frameworks such as those of law, morality, and religion employing their own unique standards. This sharing of standards makes possible so-called “theoretical identities” such as ‘Water is H₂O’ and ‘Gold is the chemical element with atomic number 79’ as identities across sub-frameworks. Their pairing of terms like ‘water’ and ‘gold’ of natural language with theoretical terms combines the vocabulary and type of term introduction of everyday descriptions with those of the natural sciences. For such identities there is a breakdown of the law of the indiscernibility of identicals, but the fact that predicates like ‘tasteless’ and ‘yellow’ ascribed to water and gold are different from those ascribed to molecules and atoms does not prevent the assertion of the identities. [3]

These differences of evaluative standards for general frameworks provide examples of the types of meaningless identities that Putnam seems to compare to mind/body metaphysical identities. We consider now some of these examples.

Some Examples Applied to Putnam’s Thesis

Meaningful Identities. First, some examples of meaningful identities, some true, some false, formulated within the general empirical fact-stating framework. They include true identities with names as ‘Mark Twain is Sam Clemens’ and ‘The Evening Star is the Morning Star’ with its definite descriptions referring to the planet Venus, as well as false identities and non-identities such as ‘Jefferson was the first U. S. President’ and ‘Washington was not the first President’. Whether true or false, all such identities of the form a = b or a =/ b are obviously meaningful because formulated within a framework for which there are empirical standards for assessing truth or falsity. As noted above, we also accept as meaningful theoretical identities pairing natural language terms with terms used in the descriptions provided by the specialized natural sciences. We thus have identities between sub-frameworks such as ‘Water is H₂O’, ‘Gold is the chemical element with atomic number 79’, and ‘Lightning is an electric discharge’. Though the terms of these identities come from different sub-frameworks and the means of identifying their referents are different, they are evaluated as true or false within a general discourse framework with the common purpose of describing our environment. For this reason, we regard them as meaningful.

Consider now some identities from other discourse frameworks. Within the framework of mathematics we have identities and non-identities such as ‘7+5 = 12’ and ‘The sum of the angles of a Euclidian triangle is not equal to 180 degrees’ whose truth or falsity can be established by the evaluative procedures of mathematicians. Also meaningful within the framework of fiction is ‘The son of Jocasta is the husband of Jocasta’ whose terms can be used to identify Oedipus, the moral identity ‘A person A’s right to truthful information from B is the duty for B to tell A the truth’, the judicial identity ‘The right to bear arms is the right protected by the Second Amendment of the U. S. Constitution’ and the Judeo/Christian ‘The Ten Commandments are God’s commandments transmitted through Moses at Mt. Sinai’. All such inter-framework identities – some still controversial – are not formulated within the empirical fact-stating framework, but they are nonetheless meaningful, since the frameworks of fiction, morality, and the law provide standards of evaluation by which consensus can be reached, albeit not in the way achieved by experimental sciences. To claim they are meaningless is to simply repeat the dogma of logical positivism as derived from the early Wittgenstein, Schlick, Carnap, and Ayer.

Meaningless Identities. Let’s now contrast the sentences just considered with some obviously meaningless identities. These include ‘7+5 = the Evening Star’, ‘The chemical element with atomic number 79 is the son of Jocasta’, and ‘Mark Twain is the right to bear arms’. Unlike ‘7+5 = 13’ and ‘Mark Twain is not Sam Clemens’, these are clearly not false, but ill-formed, examples of what Ryle called “category mistakes.” We shall refer to such ill-formed identities as cross-framework identities. They are different from meaningful identities because their paired terms are from discourse frameworks used for very different purposes and employ different standards of evaluation. In the absence of common standards for evaluation and means of reaching consensus, they are clearly meaningless.

With such examples as a background, we are now able to state what seems to be Putnam’s thesis. It is that both the metaphysical mind/body identities of contemporary materialists and the non-identities of earlier dualism are neither true nor false, but meaningless. They are meaningless because the terms of the identities are derived from discourse frameworks used for different purposes and employing contrasting standards of evaluation. Metaphysical identities are thus to be compared to meaningless identities such as ‘7+5 = the Evening Star’ and contrasted to meaningful inter-framework identities. J. J. C. Smart’s early formulation of the identity thesis had appealed to the analogy between theoretical identities such as ‘Lightning is an electric discharge’ and the identity ‘My pain is a brain process of type X’. Just as ‘electric discharge’ is a technical term applied by physicists, Smart argued, so ‘brain process of type X’ is applied by neurophysiologists, and if physics can use ‘electric discharge’ to identify lightning, so can the terms of neurophysiology be used to identity pains. The fact that we ascribe predicates such as ‘intense’ to pain and describe brain processes in very different terms does not preclude our asserting the identity, for acceptable theoretical identities provide parallel examples of violations of the indiscernibility of identicals. To accept Putnam’s thesis is to reject this analogy and claim that metaphysical identities are more like the cross-framework identities of our examples than like meaningful inter-framework theoretical identities.

Stated in this way, I think we can see why the thesis is basically correct. It is correct because our mental language has discursive transactional uses related to addressing uses of personal pronouns that differ in crucial ways from the uses of standard descriptions, and these differences undermine analogies to theoretical identities. These uses often include solicitations for help. Consider a person coming to an optometrist to improve her vision. Asked what she sees when looking at a letter chart, she could report her visual image by ‘The letters look blurred to me’. Clearly the only basis the optometrist has for accepting what the patient says as true is trust in her veracity; there is no reliable independent means of evaluation. Suppose that correlations are established between this patient’s reports of blurred letters and some characteristic brain process of type Y that normally occurs while she makes the reports. The fact that she makes the report and the type Y brain process fails to occur would not seem to provide a means of rejecting her report. Her testimony would be accorded priority, and lead to a revision of generalizations based past correlation between reports and brain processes. Trust in someone who has proved to be reliable in the past about matters that we can independently test would seem to override conflicting evidence. Similar considerations apply to pain reports. We accept on trust such reports from reliable informants, as a report is typically a solicitation for relief that relies on the sincerity of the report. Again, trust would seem to override independent criteria for pain that we may apply. If a person were to place his hand on a hot stove and report feeling no pain, we would be puzzled, but I think we would accept the reports as authoritative if the person had a history of reliability, and we would do this even if the pain reports were inconsistent with a correlation between past reports of pain and a type of brain process. Now contrast this situation to that occurring when we apply a theoretical identity such as ‘Water is H₂O’. The chemist’s description could be used to discriminate between standard light water and heavy deuteronium water of structure D₂O, and used also to correct an assertion that a liquid is water based on observations that it is translucent and tasteless. In such cases, the chemist’s description would invariably override descriptions based on normal observations. With its overriding reliance on trust, discourse using mental terms has thus very standards of evaluation from those of empirical descriptive discourse, and is not simply a sub-framework of empirical descriptive discourse. This contrast is the basis for assessing metaphysical mind/body identities as meaningless.

The contrast is just as evident when we turn to the identification of attitudes such as beliefs with brain states. We do explain and predict on the basis of beliefs, as when we predict a person will carry his umbrella by citing his belief it will rain or explain why someone crashed his car by citing a belief the street was not icy. But the beliefs we attribute as mental states are clearly not to be understood as causal intermediaries between environmental circumstances and behavioral responses. Like sensation words, ‘belief’ has primarily a transactional use in providing aid, with the aid coming through evaluation of what is believed. To say that someone believes that p implies that the belief may be mistaken; otherwise, we usually say he knows that p, thus endorsing what he says. The person who is described as believing it will rain is one who may be mistaken, and the belief can be corrected if we have better information. In this way, we may help the person believing it will rain avoid carrying the umbrella as a needless burden. For beliefs used to explain past actions this transactional aspect is just as obvious. If Jones crashed his car because he believed the streets were not icy, correcting the mistake becomes a means of alerting others to the importance of not sharing the mistake on wintery days. It may be too late to help poor Jones, but others may be saved a similar fate.

Richard Behling has remarked to me that the conclusion that metaphysical mind/body identities are meaningless seems too strong. They may be very different from the fact-stating identities of empirical discourse, and from theoretical identities such as ‘Lightning is an electric discharge’. Because metaphysical reductions have been the source of never-ending disputes, they are in some sense “odd,” and their identities are certainly very different from those within the discourse frameworks of social institutions that enable eventual consensus. The analogy between mental/physical identities and theoretical sub-framework identities may not hold. But the metaphysical identities are not “meaningless” if properly understood. I think Behling’s reservation is important, and this calls for a qualification of Putnam’s thesis.

Sensations and Intentional Attitudes

To state this qualification requires noting some differences between metaphysical identities stated for sensations and those for propositional attitudes like beliefs. Intuitively, we find more plausible, I think, the identity between a sensation and brain process than we do for one stating an identity between a belief and brain state. This is partly due to the fact that experimental psychology and neurophysiology have established quite detailed correlations between reports by subjects of what they experience and chemical and electrical processes in their brains, and these correlations have practical applications. Where we do have a correlation between a sensation and brain process, we have the possibility of aiding those in distress. A pain can be relieved by an aspirin altering the chemical state of the brain; blurred vision can be corrected by taking measures to alter the signals sent by the optical afferent nerves to the brain; and moods of depression can be altered by appropriate drugs. If indeed the primary use of our mental language is to solicit aid, then we seem to have a way of understanding what is asserted by metaphysical mind/body identities. As cross-framework combinations of terms, they are indeed ill-formed and odd. But as reductions and asymmetric identities they establish a priority of the physical descriptions of the sciences over mental avowals of everyday language, a priority that can be useful in directing attention towards what is usually the most effective ways of relieving distress. Such metaphysical identities may be meaningless as assertions of matters of fact, for this is the domain of the descriptive discourse framework whose justification is based on public observation. But understood as indicators of means of assistance, they can be useful and lead to helpful actions, and to this extent are meaningful.

Now contrast this situation to that for beliefs. It is true that our descriptions of our propositional attitudes are not authoritative in the way we think they are for sensations. Self-deception about our beliefs, wants, and hopes is common, and can be corrected by others on the basis of behavior. But such correction will not come through applications of neurophysiology. It may be possible in the future for neurology to have detailed knowledge sufficient to distinguish the brain state of a person who expresses the belief it is raining from the brain state of one who says it is cloudy and from that of another who says it is snowing. But of what possible use could we put such information? It can certainly be of no realistic help to the persons having these beliefs. If they are mistaken, then surely the way to help them is to simply inform them of their error or perhaps to convey this to others who may share in it. This is certainly to be preferred to altering their brain states with technology that may someday become available. There may be some sense in which it is meaningful to say that a belief that p is a brain state Z, for we could understand this as asserting that a neurophysiological description has priority over any expression of a belief by ourselves or a belief ascription of another. But unlike the case for sensations, there is no such priority, as it fails to direct attention in a way that could lead to practical benefits. In this sense, the belief/brain state identity is not meaningless, but instead false, though in the odd way characteristic of metaphysical identities.

Finally, we must recognize that these metaphysical identities were formulated as part of the general project of extending the methods of the natural sciences to areas previously thought accessible only to intuition and introspection. The reductionist identities used are meaningful as means of testing the possibility of certain extensions. Consider, for example, attempts to provide reductionist theories of meaning. Like mental words, the word ‘meaning’ has a primary transactional use in interpretation, as when we say that the word ‘bachelor’ means ‘unmarried man’, that ‘rosebud’ in Orson Wells’s Citizen Kane means the genitalia of Kane’s wife, or that Hamlet’s hatred of his step-father means that he considers this step-father as a rival for the affections of his mother. We learn much from criticisms of early behavioral reductions of meaning formulated by Quine and Carnap and from criticisms of more recent functionalist reductions formulated by Fodor, for this enables us to evaluate these extensions of experimental science to transactional interpretation.

More generally, we can say that philosophy is the discipline that states justification standards used within different discourse frameworks and examines boundaries between them in the form of limits of extension. The idealists’ attempt to reduce material things to complexes of ideas is more than a historical curiosity of the eighteenth and nineteenth centuries. It is of great interest to us because we recognize the idealist systems that resulted as attempts to extend the framework of religious monotheistic belief as a way of counteracting Enlightenment skepticism about religious beliefs. The “Absolute” of Hegel and Bradley became the substitute for the object of monotheists’ religious faith. This use of philosophy has proved unsuccessful, for the language with which we report how things look to us, the primary language for the idealists, we recognize now as parasitic on a public language used to describe how things are. The arguments used by idealists thus must presuppose the existence of what they claim to be denying. But while Berkeley and his successors commit this error, it is an important error from which we learn much about the dependence of one form of language on another. It is a lesson we could not have learned without initially accepting the reductionist identity of the idealists as meaningful.

I know that many will take this concession to the meaningfulness of metaphysical identities as an invitation to claim that there is a metaphysical discourse framework in which the traditional claims of materialism, idealism, and dualism have been formulated. This framework has been thought to have its characteristic vocabulary, its methods of introducing terms, and its special standards of evaluation that have enabled philosophers to correct the errors of their predecessors. To receive a philosophy education is to learn this vocabulary and the types of argumentation used by the philosophical tradition in establishing one of the three basic alternatives of materialism, dualism, and idealism, just as physics students learn to work within the framework of physics, students of the arts and literature learn the discourse framework of interpretive studies, and seminary students learn to express themselves within the discourse framework of their religious traditions. Metaphysics differs from these other institutional specializations only in the scope of its subject matter, for its topics include those of all other disciplines, and its framework is some sort of master discourse framework to which others are subordinate.

This view is, I think, seriously mistaken, but for reasons much too involved to discuss here. Suffice it to say that the enduring nature of metaphysical disputes seems to indicate that there are no standards of evaluation enabling consensus, and this is the mark of a viable discourse framework. And where we have reached consensus, it seems not to be in the form of some positive conclusion, but in the negative rejection of some attempt overreach beyond the boundaries of a certain discourse framework. For the idealist thesis, this required noticing the dependency of our language describing how things appear to us on language describing how things are. It seems we can reach consensus on the boundaries and overlaps of language frameworks and their different standards of evaluation in a way that we cannot if we confront metaphysical problems head-on within some distinctive framework.

This inability to state directly identities and non-identities about our experiencing seems to support Emerson’s poetic intuition in “The Sphinx.” After listening patiently to the poet’s attempts to describe our experiencing, Emerson has the Sphinx, the Keeper of Mysteries, angrily reply,

      “Who taught thee me to name?
     I am the spirit, yoke-fellow;
        Of thine eye I am eyebeam.

     “Thou art the unanswered question;
         Couldst see thy proper eye,
      Always it asketh, asketh;
         And each answer is a lie.”

We don’t seem able to say what consciousness or experiencing is, for each assertion of what it is or is not reflects the assumptions of a favored discourse framework that some philosopher believes has priority. Often in the past, and most notably so for philosophers like Plotinus, Schelling, and Heidegger, it has been a collaboration between philosophers and religious institutions that has biased the choice of favored framework. Recently we have been faced with the choice between the framework of science and that employed in interpretive studies of art and literature, with the first tending to enjoy most prestige in the U.S. and Australia, while the interpretive framework has tended to be most influential in Continental Europe. This choice between widely divergent frameworks used within social institutions and the political factionalism it has inspired should itself make us skeptical about conclusions reached within metaphysical systems. But while I believe we can say nothing meaningful about what consciousness or experiencing is, we do seem to be able to say a good deal about how it operates as sign interpretation and use, and at the human level this takes the form of describing the various uses of language, the evaluative standards of the principal discourse frameworks, and relationships between them. I think this is the true moral to be derived from Putnam’s insight about the suspect nature of mind/body identities, an insight that has important implications for the way those in our profession interpret the history of philosophy, the way we organize our academic departments, and the way we conceive our social role. It would be overly optimistic to suppose this insight is really sufficient to slay the many-headed metaphysical monster. This Cerberus will surely raise another ugly head again sometime in the future. But let’s hope that at least it will not be in the guise of recent reductionism.