There is an odd experience that Plato may have had. If light filters into a room through a small enough aperture, anything moving on the street outside will cast its shadow on the ceiling and back wall, and the shadow may have only the most abstract resemblance to the original. Perhaps the human predicament is really like that. The truth about the world may be difficult or even impossible to attain by ordinary methods.
The way to develop this kind of speculation is to take a part of human experience that is known to provide us with inadequate representations and to suggest that the whole may really be inadequate in a similar way. A skull is quite like a camera obscura and Plato’s Cave is a powerful image. Hilary Putnam’s version of the speculation is more frightening. He imagines brains in a vat of nutrient liquid with a computer giving their sensory nerve-sockets all the stimulations of daily life and collecting and taking account of all the impulses coming from their motor-nerve outlets.
Plato’s prisoners could escape from the cave, but Putnam supposes that his brains in a vat will never find their way into skulls placed on the shoulders of bodies that walk the world. Is there any other way in which they might discover their predicament? If not, then, given that the microcosm of their experiences is exactly like ours, how can we tell that ours is not produced in the same way as theirs? So how do we know that we attain the objective truth about the macrocosm?
The theme of Putnam’s book is that truth is attainable only if it is partly dependent on the mind and so partly constituted by the methods by which it is attained. If this is a viable strategy for avoiding scepticism, it is one that needs careful planning, because it is easy to go too far and to fall into subjectivism.
The speculation that we might be brains in a vat seems to be endemic at Harvard. Robert Nozick, in his recent book, argues ingeniously that, though he does not know that it is false, he does know something incompatible with it – namely, that he is in Cambridge, Massachusetts. Putnam does not let things get so far, because he thinks that the speculation is necessarily false. His argument is that, if the brains in a vat described their predicament in our vocabulary, the words would have to be interpreted not in our way but in theirs. The two nouns, ‘brains’ and ‘vat’, as used by them (impulses picked up by the computer from their motor-nerve outlets), would refer, not to the two types of thing to which they refer on our lips, but only to two types of stimulation of their sensory nerve sockets or to related phantasmata in their microcosms. He then claims that, on this interpretation, their description of their predicament could not be true, because (to put the reason in our way) it would imply that they were hallucinating that they were in a vat, and not that they were really in one. So if we actually are in this predicament, our description of it must be false.
This argument seems to fall short of proving that the speculation is necessarily false. Certainly, the use of the word ‘hallucinate’ is not justified. It is true that the macrocosmic description of the predicament of the isolated brains could not be produced by them, but why should we suppose that it would be contradicted by the microcosmic description that they could produce? However, this leaves Putnam’s overall strategy unaffected. For it seems to be enough for his purpose that, if the isolated brains tried to speculate about their predicament in a way that made the truth entirely independent of their methods of attaining it, they would lack the resources to formulate the speculation. Wittgenstein made this point in his discussion of solipsism in the Tractatus: if sense is based on reference, intelligible hypotheses cannot reach quite so far into the unknown.
But what is reference? When someone uses the word ‘cat’, we do not doubt that we know the kind of thing to which he refers. But what focuses his reference onto that particular species? Nothing, Putnam argues, within the microcosm of the speaker’s language. If he has a theory about what he calls ‘cats’, he will be able to tell us a lot about its interpretation by specifying the worlds in which it would be true and the worlds in which it would be false. However, even if he fixes the truth-value of his theory in every possible world, the reference of the word ‘cat’ will still remain indeterminate. This is a more extreme version of a thesis of Quine’s.
The natural reaction is to protest that we all know how to use reference as a line thrown out from the microcosm of language to establish connections with definite types of thing in the macrocosm. So at least this is not like the speculation about the isolated brains, which pushed us towards scepticism. The boundary between microcosm and macrocosm is drawn quite differently in this case. It divides language from the world and the point is not that our experience leaves us in any doubt about the targets of our references, but that it is difficult and perhaps impossible to explain what makes them that.
Evidently, a theory is needed to support the protest of common sense. Maybe the rival candidates for the reference of the word ‘cat’ would introduce recognisable anomalies into the instructions for applying it. Putnam dismisses this suggestion, perhaps too hastily. In any case, we need something more than a theory about the internal structure of the microcosm of language: we need a theory that will tell us what reference itself is. He mentions Hartry Field’s idea, that it is a specific causal relation, R, but observes that his original argument shows that the reference of R will be as indeterminate as that of any other word. When we advance the boundary of the microcosm of language in order to theorise about reference, we take the problem of indeterminacy with us. Here again he moves close to an early idea of Wittgenstein’s: there are certain links between language and the world that can never be explained in words. However, he finds this unacceptable, both because it is too mysterious and because it makes the world of objects too independent of the world of words.
It is difficult to be sure what kind of dependence he requires, or even that it is a single kind. If all our beliefs and theories about the world are the product of some sort of interaction between the world and the mind, it would be unreasonable to ask for a specification of the world’s contribution that would not itself be a product of the same interaction. However, in the case of sense-perception there is something more modest that we can do: we can use our sensory input to construct a physical theory about its external causes. Berkeley, of course, was not convinced, but maybe his scepticism can be answered. Anyway, we seem to understand this controversy, because it is generated by a kind of shrinkage: we shrink the microcosm within the macrocosm and then we seem able to argue intelligibly about the vacated space.
Now, as Putnam explains, the case of sense-perception provided the model for the subsequent controversy about another contribution made by the mind – the concepts that it applies to its sensory input. Let us adopt the word ‘mentalisation’ for this digestive process. Then the questions posed by 19th-century idealists and in a different form by modern philosophers of language concern the unmentalised world. Does it make any contribution to the final product, truth, and, if so, what contribution does it make? What is the raw material really like?
Putnam, like Kant, believes that truth would be unattainable if it were any kind of correspondence with the supposed raw material. But before we can assess the effect of such a theory of truth, we need to understand it and compare it with his alternative theory. Here, unfortunately, we encounter a radical difficulty. We may think we prefer the picture offered by one theory to the picture offered by the other, but there does not seem to be any rational way of choosing between them. They are both so general and they both speculate about a space that seems to be inaccessible to rational inquiry. How can we tell whether truth is correspondence with the raw material or coherence with the mentalised product? In the case of sense-perception, we are able to produce a theory about what lies beyond the boundary of the microcosm, but in the case of truth there is no support for any parallel theory. The microcosm of language is not produced by a shrinkage that leaves us apparently able to argue intelligibly about the vacated space, and it really is not clear how discussions of what lies beyond its boundary can proceed. Certainly, if discussable truth were coherence with the mentalised product, it would not be worth discussing without undiscussable truth: i.e. without correspondence between the whole system, including that product, and the raw material. Perhaps the real question here is not one of theory but of definition: when we review the stages by which truth is produced, how far back should we go in our demarcation of the relation that we are going to call ‘truth’?
If we must have a theory, perhaps we should turn to the details and try to develop a physical account of reference, like Field, undeterred by the need to use language to explain its own basis. Or maybe we should start at the other end and go to cognitive psychology for an explanation of the mentalisation of sensory input. But it seems obvious that whatever successes we achieved along these lines could not be sufficiently general to count as a theory about the input into the microcosm of language. Is this then a real problem, or only an illusory one, produced by the misleading analogy with sense-perception?
Putnam, like other contemporary philosophers who adopt his position, does not argue for it in a schematic way, like Kant. He claims that certain features of our theories about the world show that truth is ‘some sort of (idealised) rational acceptability – some sort of ideal coherence of our beliefs with each other and with our experiences as those experiences are themselves represented in our belief system – and not correspondence with mind-independent or discourse-independent states of affairs’. The raw material is mentalised at the point of sensory input and our beliefs only have to fit the product of the mentalisation. Now there is no reason to suppose that there is only one way to mentalise a given piece of raw material, and so we cannot infer its nature from the nature of the product. Of course, we know why ‘snow is white’ is true, and we can always say that the material is the material that yields the product, but that adds nothing to our knowledge. Moreover, if we move back from the boundary of the microcosm of language, we find that the construction of our theories allows us further options, which, unlike the options at the boundary, can be discussed informatively. So how can truth be construed as correspondence with anything independent of the truth-seeking mind?
These arguments seem to establish two points. First, if truth is some kind of correspondence, it is not a one-to-one correspondence that picks out a single theory for each collection of raw material. Second, the raw material itself cannot be specified without being mentalised. But do they establish the third point, that truth is not correspondence with anything completely independent of the mind?
Perhaps the third point is only an exaggerated reaction against the cruder notions of correspondence. It seems obvious that the target of truth must be the unmentalised world, and this does not seem to be controverted by the fact that we cannot describe it without mentalising it, and so cannot produce any general hypothesis about the nature of the raw material or its contribution to our system of beliefs. It may be an illusion to suppose that, if the target of truth is independent of the mind, some general hypothesis of this kind ought to be constructible. If so, the illusion is probably produced by the misleading analogy with the case of sense-perception. Certainly, this diagnosis gets some support from the history of 19th-century idealism, which is partly recapitulated by recent theories about the relation between language and the world.
There is also the theory that lies at the opposite extreme, subjectivism or relativism. Putnam, of course, rejects this theory, but his acceptance of the third point makes it difficult to see how he distinguishes his own theory from it. Few philosophers have adopted a subjective stance towards science (Feyerabend and perhaps Kuhn and Foucault), and such writings are best taken as an antidote to the more extreme versions of objectivism. Ethics is a more suitable case for relativistic treatment, but towards the end of his book Putnam claims the same mind-dependent objectivity for the system of human values that is all that he has conceded to science.
No doubt we do tend to take too subjective a view of ethics, because our standard is set too high by our exaggerated estimate of the objectivity of science. But can the two really be brought so close together? His main argument is that, since truth is rational acceptability, our empirical world depends on our criteria of rational acceptability and that gives us one value as objective as the vaunted facts of science. This adds something new to his previous position: the idea that a certain standard of rationality corresponds to a basic human need. Scientific rationality is not justified pragmatically by its success in predicting developments in an independent empirical world. On the contrary, the empirical world is, in large part, a product of the human need to think in that way. The inquiry then shifts to other human needs and the project of building a balanced system of values on them.
Putnam’s chapters are like cavalry charges. Each has a limited aim, pursued with force and panache. But he succeeds both in presenting some of the deepest problems in philosophy in a way that will be intelligible to the layman and in advancing towards solutions. It would be interesting to know how Hume or Russell would have reacted to some of his more provocative conclusions. Hume’s philosophy was very anthropocentric, but it is unlikely that he would have agreed with such a dramatic extension of the mind’s ‘propensity to spread itself on things’. Russell would probably have complained that anyone with ‘a robust sense of reality’ would have given the raw material a bigger role. But what is the flaw in Putnam’s reasoning? If there is one, it must be in his treatment of the outer edge of the microcosm of language. Perhaps his theory that truth is coherence with the mentalised product exploits the method of production without including it in the truth-relation simply because it cannot be explained in words. That would hardly be fair to the raw material, because it is only by conceding the constraints put by it on the mentalised product that his theory can maintain its distinctness from total relativism.