Before trying to fit the vocabulary of the mind to the model of ‘voice’, we should examine our model more closely. Of particular importance is the question of what ontological distinctions to associate with the distinctions of verbal function we have examined. In short, are there voices? One is inclined to answer ‘Of course! We hear and enjoy and recall and recognize voices, so there are voices’, but why the conclusion? Are there sakes? I can do something for Sam’s sake, and he can want something for the sake of the nation, but then must there be sakes? There seems to be point, and truth, in saying that there really are no sakes or dints, rather less point and truth in denying the existence of miles and degrees Fahrenheit, and a great deal that is implausible in denying the existence of voices. Where then should we draw the line?
It may seem that the line is naturally drawn by determining whether the existence contexts, ‘there are …’, ‘there was …’, and so forth, are legitimate contexts for the noun in question. This course would establish that there are no sakes (or better: ‘sake’ does not denote or name or refer to anything; if ‘there are sakes’ is to be a solecism, its negation must be as well). ‘Mile’ would also be ruled to have no ontic force, since, for example, ‘there are seven miles between …’, ‘there once was a mile …’ and ‘there is a mile …’ are all improper. Voices would be admitted, however, on the strength of sentences like ‘there was a voice in the dark and I recognized it’.7 The justification for this course would be slender enough even if our grammatical intuitions in particular cases were strong and unanimous, but they are not. Our intuitions are poor witnesses just when they would be most heavily relied on. Consider the claims:
(1) ‘there is a mile between them’ is deviant usage
(2) ‘there is a mile to go’ is not deviant usage
(3) ‘there are five miles of hard hiking between the peaks’ counts as affirming the existence of hiking, not miles.
Are these any easier to assess than the question they would be expected to settle, viz., should we say, in the context of discussing or choosing ontologies: there are miles? As these claims get harder to assess, their value as criteria wanes on two fronts: relevance and decidability.
A more lenient ontology could be loosely put: something for every noun (and noun phrase, etc.). This most relaxed course cheerfully admits the existence of dints and sakes, and treats the whole question of ontology with what some may hold to be deserved disrespect. But consider the following exchange:
‘How old is Smith’s sake?’
‘Sakes don’t exist in time.’
‘But they do exist, don’t they?’
‘Why not?’
‘Then if Smith’s sake is timeless, we’ll be able to do things for it after he’s dead.’
‘No; although a sake is timeless, it can no longer receive benefits after the death of its owner.’
‘Then I might only think I was doing something for Smith’s sake, if all along he was dead without my knowing it?’
‘No, you’d be doing it for Smith’s sake, only his sake would no longer have any use for whatever you were doing.’
This sort of nonsense should be blocked one way or another. If one merely forbids the word to appear in the various syntactical roles it appears in above, what is one clinging to when one refuses to admit that the word has no ontological role? Asserting existence under these conditions is as empty as denying it. Answering yes or no to an ontological question only begins to have some point when we have decided that granting the existence of something licenses us to ask (and expect answers to) certain very general questions about it, e.g., what sort of thing is it?, does it exist in time?, and especially, is it identical with x? This last question is indispensable. It is hard to see what anyone could have in mind by affirming the existence of something, if he then disallowed this question. For, let us divide our universe into as many different ontological categories or types as we wish, if we assert that x is a thing existing in sense A, or in category A, and y is a thing existing in sense B, or in category B, then at the very least we must acknowledge that we have just spoken of two things, x and y, not just one – or in other words that x is not identical with y, but is another thing.
Consider voices again. We entertained the proposal to admit voices into our ontology because under some circumstances ‘there is a voice …’ rings true in the ear, but there are better reasons for denying them. If the anatomist or physiologist or acoustician were to be concerned because among all the things encompassed by his theories there still were no voices; if he were to suppose this meant he had left something out, something perhaps even inaccessible to science, he would have been confused by our admitting voices in our ontology. He assumed this meant he could safely reason: Is the voice identical with the larynx? No. Then is it the lungs? No. Is it a stream of air? No. Is it a sound? No. Then it must be some other thing I have not yet examined. We must rule out this series of questions, but if we must, it cannot be on the grounds that voices are logically (or ‘by meaning’) non-identical with physical entities, for we cannot rule out a question simply because its answer is ‘No (as a matter of logic)’. We can rule out the questions only by declaring them ill-formed, and hence admitting no answer.8 So the ontological question is of a piece with the question whether the odd sentences (e.g., ‘I can sit on an opportunity’, ‘The voice is identical with the larynx’) are logically false or ill-formed, for although whenever two things exist they may well be logically (as opposed to contingently) non-identical, so long as we do hold there to be two things we cannot burke the question of identity by declaring it ill-formed. The point, then, in denying the existence of voices is to permit the claim that physicalists need not identify voices with any physical thing (talk of such identities being ill-formed). That such a denial is to some extent counterintuitive is not contested, but then it is also counterintuitive to suppose there are dints and sakes. Since no drawing of the line is clearly superior in intuitiveness, we may turn to other criteria. The denial of voices has at least the systematic advantage of providing a reason for ruling out the physiologist’s questions, which are, intuitively, wrongheaded.9
Certainly no one interested in voices ever fell into the misunderstanding just described, but it is tempting to suppose that not only philosophers but also psychologists, neurophysiologists and cyberneticians are bedevilled at times by a parallel confusion over the ontological status of the mental vocabulary. The outcome of our analysis of ‘voice’ was the adoption of a relatively restrictive sense of ‘exists’, and this allows a clarifying reformulation of the mind-body problem: when the neurophysiologist – or his armchair counterpart, the physicalistic philosopher – asks whether he has left anything out of his theory of the mind, or whether anything relevant to the operations he is studying is outside the domain of his science, he is asking whether there exist (in this strong, restrictive sense) any such things. On the one hand the answer that such things do not exist may come as a relief to the neurophysiologist or physicalist, but may also come as a surprise, for even in this strong sense of ‘exist’ it seems that pains exist as surely as pins, desires and ideas as surely as electrons. On the other hand the answer that such things do exist will mean that the Rylean strategy has led us back to our starting point; we are back on the pendulum and must decide between the old alternatives of interactionism, parallelism, identity theory and so on. No amount of talk of categories and category mistakes will keep us from the snares of dualism unless we are prepared to grant ontological priority to one category at the expense of another.
Determining the ontological status of the mental vocabulary will not be simple. Instead of a general argument there must be detailed investigations of individual words and families of words, and we must be alert to the possibility that only a partial case can be made. To expedite this investigation – which will be concentrated in Part I, but will cast lines through all of Part II as well – I wish to introduce a technical term. I shall call nouns or nominalizations that do denote
or name or refer to existing things (in the strong sense developed above) referential, and other nouns and nominalizations, such as ‘sake’, ‘mile’ and ‘voice’, non-referential.10 Non-referential words and phrases are then those which are highly dependent on certain restricted contexts, in particular cannot appear properly in identity contexts and concomitantly have no ontic force or significance. That is, their occurrence embedded in an asserted sentence never commits the asserter to the existence of any entities presumed denoted or named or referred to by the term. Our prospects can now be outlined with the use of this new term. We may find that no mental entity terms can be plausibly claimed to be non-referential, in which case we are thrown back on the old alternatives: either minds are identical with physical entities or they are not, in which case we must put together some sort of dualism. Or we may find that the entire vocabulary of the mind succumbs to non-referentiality; this would allay all our fears of ontic bulge and leave the neurophysiologist in the same relatively uncomplicated position as our voice-investigator. Or we may find that the mental vocabulary is a mixed bag; in this event the crucial question will become whether or not the referential terms in the mental vocabulary refer to things identical or non-identical with physical things. In this way we might be able to put together a theory that was throughout physicalistic in import, but only an identity theory with respect to some of the mental terms: viz., the terms that refer to things that actually exist.
Once we decide that a term is best viewed as non-referential, we fuse it in its proper contexts, as we noted earlier with ‘sake’ in the irreducible idiom ‘for-the-sake-of’. The contexts maintain their significance but are not subject to further logical analysis; their parts become like the ‘table’ in ‘potable’. The chief advantage of fusion is the ontological absolution we gain, but there is a price to pay. For example, were we to take the voice problem seriously and proceed with a zealous and rigorous analysis of ‘voice’-idioms, we should have to accept that ‘John strained his voice’ is not to be treated as an instantiation of the formula ‘x strained y’, for now ‘strained-his-voice’ is a fused context not open to further analysis. This means we shall no longer have any logical licence for the apparently sound inference:
Anyone who strained anything was doing something excessive
John strained his voice
John was doing something excessive.
But we must be willing to pay this price if we are to deny a licence to the inference:
The only thing John strained was his vocal cords
John strained his voice
John’s voice is identical with his vocal cords.
Even more awkwardly, fusion will often extend beyond a few words to left and right of our non-referential term. For example, fusion must extend to any pronominal cross-references to voices. Most implausibly, for example, the whole sentence
‘The first thing about his voice that struck me was that I had heard it before’
becomes impenetrable to logical or semantical analysis. This conclusion will seem preposterous until we reflect on just what is and is not being prohibited by fusion. Obviously fusion does not prohibit analysis the way a dictator prohibits free assembly; it merely forbids certain sorts of interpretations being put on the results of analysis. It may well be possible to produce a ‘semantics’ for the ‘voice’-idioms, and a ‘logic’ as well; were this accomplished the only thing fusion would prohibit would be any attempt to treat this ‘semantics’ as an extension of the semantics of our referential vocabulary, with voices as a sort of thing in addition to the sorts of things referred to by referential terms. From the vantage point of our base camp in the midst of existing things and referential terms, ‘voice’ must forever be non-referential; only in this way can the alternative of identity or non-identity be denied. Provided this crucial bit of insulation is maintained, however, there is no limit to the sort or number of systems one may erect for dealing with the parts of fused expressions, and it is even to be expected that any systems discovered will be virtually parallel to the semantics and logic on the referential side of the divide.11 Fusion, then, is from a point of view; it renders contexts impenetrable only from certain angles, as it were. This impenetrability is not just a hindrance. It also provides a degree of freedom by excusing the analyst from finding all our logical and semantical rules obeyed on the far side of the fusion barrier.
It has not yet been decided that all or any of our mental entity terms are non-referential, but only that we should investigate to see. To this end I propose to employ a tactic which can be called tentative fusion. We wish to proceed with no ontological presuppositions to the effect that mental entity terms either are or are not referential, and this can be accomplished by treating all sentences containing mental entity terms as tentatively fused, subject to further discoveries which will lead us to confirm the fusion or relax it. We do not assume from the start, that is, that certain sorts of questions have answers, that certain sorts of implications hold, that certain sorts of parities exist between physical entity nouns and mental entity nouns. What we start with, then, are sentences containing the mental entity words to be examined. We may say these sentences are ‘in mental language’, and we acknowledge that as wholes they are significant and hence true or false. Part of what is then at issue is whether or not the parts of these sentences should be construed to fall under our standard semantics – whether or not to relax the fusion.12 The broader question of which this forms a part is whether or not these sentences, accepted either as wholes or as analysed, can be correlated in an explanatory way with sentences solely from the referential domain of the physical sciences. Our model here is the case of voices; the explanation of vocal phenomena may contain no reference to voices; can the explanation of mental phenomena similarly avoid reference to minds, thoughts, pains? By taking whole sentences as our initial units we avoid making the one presupposition that leads irresistibly to the pendulum of old-fashioned alternatives: the presupposition that ‘mind’, ‘thought’, ‘pain’ are referential, or in other words, the presupposition that there are minds and mental events on the one hand, and bodies and physical events on the other.
Not just any mapping of sentences on sentences will constitute an explanatory correlation, of course. One could associate each true sentence of the mental language with a sentence which catalogued as exhaustively as possible the entire physical state of the person or persons in question, but this would explain nothing. At the very least the sentences associated with the mental language sentences must describe conditions which vary in systematic ways related to distinctions in the mental language sentences. The degree of freedom, however, which we obtain by tentatively fusing the mental language sentences, will allow us to avoid one preposterous requirement of crude identity theory. As Putnam points out, the supposition that a particular mental experience, e.g., thinking of Spain, is identical with a particular physical state requires that all beings truly said to be thinking of Spain must be in this particular physical state, which rules out, most implausibly, the possibility that beings with a different biochemistry from ours, or a differently embodied nervous system, could think of Spain.13 Even among homo sapiens it is not plausible to insist that when two of them are both thinking of Spain they must share some unique physically describable state.
Rather than attempt to characterize in an abstract fashion the minimum requirements of a satisfactory explanation, let us proceed to see what correlations we can find. Once they have been set out we can ask whether or not they constitute an adequate explanation of mental phenomena. In most general terms our task is to provide a scientific explanation of the differences and similarities in what is the case in virtue of which different mental language sentences are true and false. Thus, for example, our task is not to identify Tom’s thought of Spain with some physical state of his brain, but to pinpoint those conditions that can be relied upon to render the whole sentence ‘Tom is thinking of Spain’ true or false. This way of proceeding still chara
cterizes the task as finding an explanation of the mind which is unified with, consistent with, indeed a part of science as a whole, but eschews – at least initially – the obligation to find among the things of science any referents for the terms of the mental vocabulary. This obligation will only be taken on in the event that some or all of the mental terms resist all efforts to treat them as non-referential.
The first obstacle thrown in the way of our attempt to achieve explanatory correlations is a very general, but very powerful, argument to the effect that those features of the world in virtue of which certain mental language sentences are true or false are outside the domain of the physical sciences, and not describable or subject to explanation within the scientific framework. If the argument is sound then our having reached an appropriately noncommittal stance for dealing with the ontological problem will be to no avail, for this argument is concerned with relations between sentences. This argument will be presented and examined in Chapter 2.
2
INTENTIONALITY
III THE PROBLEM OF INTENTIONALITY
In the previous chapter we formulated a stance that enables us to ask what the relation is between the physical sciences and the truths expressed in our mental language while carrying the minimum of metaphysical baggage. We avoid all ontological presuppositions about mental entities by tentatively treating all sentences of the mental language as containing no referential terms. Thus for at least the time being we absolve the scientist from the responsibility of discovering physical events, states or processes which deserve to be called thoughts, ideas, mental images and so forth. No entity on his side of the fence need line up with mental language in such a way that we would say he has discovered what thoughts are, or isolated a mental image or even the experience of having-a-mental-image. We have the mental language, and since the suggestion that all the things we say in the mental language might be false is incoherent, we also have the truths expressed in mental language. The task is to relate these truths to the scientific corpus, and further to explain the relations. Since we cannot very well claim to have explained a mental phenomenon if we are unable to say (in the scientific language of our explanation) when a sentence heralding the occurrence of the phenomenon is true and when not, our task will involve at least this much: framing within the scientific language the criteria – the necessary and sufficient conditions – for the truth of mental language sentences. At this point we face a very general argument designed to show that this is impossible, that no criteria for mental truths can be expressed in the language of science.
Content and Consciousness Page 3