Cooling it

Donald Davidson is perhaps the most distinguished philosopher in history never to have written a book. Indeed, he did not get round to writing articles until he was into his forties (he is now 76). Yet those articles – short, intense, allusive, hard – have changed the shape of contemporary analytical philosophy. They were in mid spate when I was a graduate student at Oxford in the early Seventies, and they acted as a kind of philosophical IQ test for the young philosophers of my generation. I well remember poring with tormented excitement over ‘Truth and Meaning’ and ‘Mental Events’, two of the most influential (and contested) articles of recent times. These cryptic texts gave the impression of well-honed conjuring tricks, in which the deepest of problems were given tantalisingly rigorous and ingenious solutions. In those days you were either a ‘Davidsonian’ or you weren’t; you certainly had to find out where you stood. But it wasn’t easy, because each Davidson article presupposed the others, and they assumed you were good at logic. It became clear that Davidson had a system, but it needed to be pieced together by the reader, as best he or she could. Puzzlement about a particular Davidson piece would be met with a knowing look from the initiated and the query ‘But have you read “In Defence of Convention T”?’ The very plainness of his name (often transmuted to David Donaldson) lent an aura of mystique to the plosive economy of the Davidson corpus. And the man himself, with his startling blue eyes and precisely articulated mode of speech, his unhurried confidence, his immersion in his own vision, his neatness, certainly encouraged the feeling that he had it all figured out, and all you had to do was figure him out. It did no harm, too, to discover that Davidson had been an enemy aircraft spotter in the US Navy in the Second World War, that he was a trained pilot, that he went gliding for a hobby, that he has climbed mountains, that there are very few places in the world he hasn’t visited. Davidson wasn’t just profound: he was cool (and there aren’t many philosophers you can say that about). Davidson had nerve.

The principal appeal of the Davidsonian system lies in its attempt to combine two conceptions of human beings that have traditionally been taken to be rivals. One conception, advocated by the positivists, though not unique to them, draws inspiration from the physical sciences and formal logic: it seeks to reduce mental discourse to physical discourse, and it offers to replace ordinary language with the kind of formalised language devised by Frege and his successors. Ultimately, there is nothing more to us than an arrangement of physical facts expressed in the notation of the predicate calculus. This conception effectively displaces our common-sense picture of mind and language in favour of a kind of pared-down physical naturalism in which we are represented as continuous with the rest of nature. The other conception, associated with the later Wittgenstein, but by no means unique to him, insists on the autonomy and legitimacy of our ordinary ways of thinking about human psychology and human language: these are not to be replaced by some austere physical theory or gleaming logical apparatus – for they are perfectly in order as they stand. We are, in fact, what we commonsensically take ourselves to be: rational agents with free choice. Man is not just an irregular clump of vibrating particles, nor need he be coached in the language of the logician: he has beliefs and desires and intentions, and his natural mode of expression is not to be improved on. He eludes physical science, at least in his mental and linguistic part: he is separate from the rest of nature and needs to be studied by methods peculiar to himself.

These two conceptions seem to represent radically incompatible ways of thinking of ourselves, and no middle ground appears to be available. But it is not as if either conception can be comfortably adopted to the exclusion of the other. The first view suffers from the problem that no such reduction or translation has ever been carried out, so that limiting ourselves to physical description will inevitably involve abandoning the idea that we have minds at all. Also, there seems to be a lot about natural language that cannot be reconstructed in terms of the usual logical systems, so that we would not be able to say as much if we spoke only Formalese. The price of seeing ourselves in these restricted ways is that what we see is no longer ourselves, but only some desiccated residue. On the other hand, if we remove the mind from the scientific domain completely, as the second view suggests, regarding ourselves as beyond the reach of causation, law and material composition, we run into equally severe problems. Do we not have brains that subserve our minds? Is not our behaviour somehow governed by natural law? Are we not in some clear sense ultimately made of matter? And is not formal logic an object of great beauty and power, giving undeniable insight into the structure of thought, whose services we should solicit and exploit? Hence the classic dilemma: how can we both be and not be an object of natural science?

Davidson’s key idea is that the dilemma is unreal: we can enjoy the benefits of both conceptions without incurring the disadvantages each appears to entail. What we must do is compromise, not pushing either conception beyond its legitimate sphere. Yes, but the question has always been how exactly that is to be achieved. The beauty of Davidson’s philosophy of mind is that this massive question is held to turn on a simple – but neglected – logical point. Once this point is made plain we can be all we want to be. No ideological posturing will be necessary; no spurning of the obvious; no deep unifying revisionary metaphysics will have to be generated. All we need to recognise is a. that there are events and b. that events, like material objects, admit of a type-token distinction (about which more in a minute). We don’t even need a theory specific to the mental: once we get clear about our talk of events in general we will already have the necessary resources with which to explain how the mind can be both rational and natural, irreducible and physical, causal and lawless. To put it differently, once we properly grasp the distinction between events and their descriptions we will be in a position to be both materialists and mentalists.

It works like this. First, it is obvious that we talk of events as well as objects, as when we say that the bridge collapsed because of the explosion, or that Smith went to the shops because it occurred to her that she needed some milk. That is, we routinely include mental and physical events in our ontology. Secondly, and only slightly less obviously, we allow that there can be different instances of the same general type of event, as when two bridges collapse on different days, or when the same milky thought occurs to Jones. These instances are the event tokens and the universals they exemplify are the event types. You have to count types differently from tokens, since many tokens can correspond to the same type and a given token can exemplify many types. In other words, distinct particular dated events can fall under the same general description, and one and the same particular event can fall under many descriptions. Accordingly, mental events, too, admit of a type-token distinction, requiring us to distinguish particular events from the descriptions that apply to them.

And now Davidson’s master-stroke is just this: every mental-event token is identical with some physical-event token in the brain, but mental-event types are not identical with physical-event types, nor are they reducible to them. Ontologically, then, every mental particular is a physical thing (falls under a physical description), but it is not possible to reduce mental concepts or properties to physical concepts or properties. According to this position, which Davidson christened ‘anomalous monism’, every menial event falls under a physical law, but there are no laws of psychology; indeed, psychology is not really a science at all. The reason is that mental descriptions apply to things that also satisfy physical descriptions, but no systematic relation between mental and physical discourse is entailed by this identity of events. Thus we can be, as it were, natural under one description and rational under another; lawlike when described one way but lawless when described another way. We consist of a single series of events, but this series admits of quite distinct and irreducible modes of description. (Technical note: the type-token distinction as here invoked is really a special case of the use-mention distinction – we use descriptions to mention events.)

On the one hand, then, we can say that mental events are physical, that reasons cause actions, and that everything is subject to strict law; while, on the other hand, we can insist that mental concepts are irreducible, that action explanation is essentially normative, and that there are no psychophysical or psychological laws. We do not need, on this Davidsonian view, to explain the relation between mental and physical properties of a person: we ascribe these properties under appropriate criterial conditions, following distinct sorts of principle, and that is all that needs to be said. In order to solve the mind-body problem, then, we do not need to represent mental properties as physical properties in disguise, nor do we need to effect a conceptual revolution that will bridge the gap, or even to acknowledge that there are facts about the mind-body relation that we cannot grasp. We need merely to observe that one and the same event can be described in these two ways. We may also, if we like, hold that mental descriptions are supervenient on physical descriptions, so that physical twins must also be mental twins: but this is an optional extra, in no way entailed by the thesis of anomalous monism. There are indeed two very different sides to our nature, but from an ontological point of view we are undivided beings.

Simon Evnine’s introductory book does a creditable job of bringing all this usefully together, enabling the student to grasp how the various parts of Davidson’s philosophy cohere. He has a sure grip on the metaphysical and logical bases of Davidson’s distinctive approach, particularly as regards causation, laws and ontology, and he puts the essential points in such a way that only the most determined could miss them. His book should take a lot of the pain out of learning and teaching Davidson. He is also well aware, on the critical side, of the tensions that lurk within his subject’s hybrid picture of human mentality. The heart of the trouble, as he notes, lies in maintaining both a causal and a normative account of the nature of propositional content. Davidson invokes causation at three critical junctures: to relate reasons (qua reasons) to actions; to confer content on beliefs, by identifying the object of a belief with its environmental cause; and to account for cases of irrationality, where the notion of a mental cause that is not a reason is brought into play. These three causal theses correspond to three threats to the other component of Davidson’s overall conception. First, if reasons (qua reasons) are causally relevant to action, then there must after all be psychological laws of some sort, just as there are laws in other special sciences: and this means that there is no inherent conflict between normativeness and lawfulness. Second, if belief content is fixed by environmental impingements, then it is hard to see how it could also be holistically determined by principles of rationality that essentially advert to the agent’s other beliefs and desires – any more than the identity of the impinging objects is so determined. Third, if we are allowed, in cases of irrationality, to make ascriptions of content that violate conditions of rational justification, so that lack of rationality does not undermine the possession of content, then it becomes unclear why we cannot push this separation further, until the point at which the agent is preponderantly irrational. The trouble with causation, as the cement of the mind, is that it has the wrong properties to sustain Davidson’s hermeneutic-holistic-normative picture of mentality. Once causation is allowed to flow through the mind’s channels it threatens to flush out the kind of anomalism Davidson wishes to combine with it. If this threat cannot be convincingly repelled, then we shall be forced into one or other of two kinds of extreme position: either a position like Jerry Fodor’s in which causality reigns and rationality be hanged, or a position like Daniel Dennett’s in which rationality is prized but the idea of inner causes is fed to the dogs. Certainly, Davidson needs to say more about why he is not forced in either direction. To be sure, it would be nice to be able to combine both viewpoints: but mere conjunctive affirmation is not enough to bring this off.

Davidson’s philosophy of language is intimately connected with his philosophy of mind, but it raises questions of its own. Evnine also does a good job with this more technical aspect of his subject, which is more than I can say for Bjorn Ramberg’s ill-expressed attempt to convert his reader to the Davidsonian faith. This could not be used as an introductory text, despite its title, because of its failure to explain technicalities and its general sloppiness; nor does it contain material of sufficient originality to be of interest to those already acquainted with the literature of Davidson. It is exactly the kind of book he doesn’t need: an exercise in undisciplined banner-waving. What would have been more helpful is a clear tracing out of the several strands that link Davidson’s work on semantics with his epistemology, and ultimately with his view of how the mind contrives to confront reality. For here we can discern an instructive evolution, in which an initially technical problem leads to a questioning of the entire empiricist tradition.

I remarked earlier that Davidson seeks to dissolve the traditional opposition between reverence for formal systems in developing semantic theories and respect for the structures actually present in natural languages. By pairing vernacular sentences with suitable formal counterparts, and providing a theory of the latter, Davidson proposes indirectly to give a theory of the former. Hence his claim that Tarskian truth theories – in which the predicate ‘true-in-L’ is rigorously defined for a formal language on the basis for a finite set of axioms – provide the basis for a theory of meaning for natural languages such as English. Davidson is able to suggest this thanks to his considerable success in ingeniously translating logically recalcitrant idioms of natural language, such as adverbs and indirect discourse, into formulas of a standard logical language. Logical form is what we need if we are to provide a theory for natural language at all, rather than being a rival to it.

Details aside, however, there is the question whether this entire approach can capture the full meaning of sentences, given that it operates with the apparently much weaker notion expressed by ‘is true if and only if’. Davidson’s response to this persistent problem has been to enrich the set of notions used to capture meaning to include those that would feature in an account of how we would empirically verify that a given truth theory successfully interprets the speech of a particular community; specifically, to employ such psychological notions as belief and desire. The question has become: what are the right assumptions to make about the mind of a speaker who holds a sentence true in certain environmental conditions? And here Davidson applies a principle that has figured increasingly in his philosophy – the so-called principle of charity.

In order to arrive at an attribution of belief to the speaker we are to assume that his beliefs are true, so that we can use the surrounding facts to determine a content for his belief. If he holds ‘it’s raining’ true when it is in fact raining, then we should take it that he believes that it’s raining and so interpret his sentence to mean that it’s raining, instead of assuming, uncharitably, that he has made a mistake and thinks it’s not raining when it is. This sounds like sensible enough practical advice to the would-be interpreter, but it raises the question of how we can be so confident that people regularly believe what is true. Isn’t it at least conceivable that a community could speak an interpretable language and yet be massively deluded about the extra-linguistic facts? Couldn’t I be a speaker of English and yet be a brain in a vat, as Cartesian sceptics have long assumed to be possible? Davidson’s shocking answer is that actually I could not: he thinks that it is a necessary conceptual truth about belief that one’s beliefs are mainly true. Beliefs must not only be rationally consistent with each other in order to be possessed at all: they must also veridically represent how the world is. But this yields a startling result: scepticism must be incoherent, since it tries to envisage situations in which people have beliefs but get everything wrong. Davidson justifies this result by insisting that the content of belief is fixed by its actual cause, and not by any epistemic intermediary such as experience. He is thus led to reject the empiricist dichotomy of scheme and content, of concepts and the given. That is, in sum: in order to make up for the logical extensionality of ‘is true if and only if’ Davidson is led first to interpretation, then to charity, thence to a rejection of scepticism, and finally to an abandonment of the third dogma of empiricism. Where the empiricists took meaning to be possible only if it stems from sensory experience, Davidson takes the theory of meaning to be possible only if experience plays no role in fixing meaning. Meaning, for him, results from a direct collision, or collusion, between belief and fact.

This is stirring stuff, but it is far from obvious that it is correct. What powers the whole argument is the initial claim that interpretation can only get going if we make a charitable assumption about belief, since mere holding-true is mute as to what is believed. But isn’t there something Davidson is forgetting? Agreed, mere assent to sentences will never by itself decide between different hypotheses about what is believed, so that something else must take up the slack if we are to interpret at all. But we are not compelled to leap to a fixed policy of charity, since we can always appeal to the speaker’s non-linguistic behaviour to narrow the options down. Suppose our speaker assents to ‘it’s raining’ in broad sunshine: we might be inclined to suppose that he can’t really believe it’s raining – so we charitably assign to him the belief that it’s sunny, reinterpreting his words accordingly. Of course his assent is not all we have to go on: we might observe him scampering under a tree, swearing, dabbing at his face with a hanky – giving all the signs of a man who is convinced it’s pouring down. Well, that would be evidence that he actually believes it’s raining, in plain contravention of the facts. And we might then go on to assemble further evidence that he is suffering from delusory perceptions, perhaps caused by malnourishment or whatever. None of this would be conclusive – he might be trying to deceive us into thinking he believes it’s raining when he knows it isn’t – but then no empirical evidence for anything is ever conclusive.

The point is that we are not, as interpreters, stuck merely with inscrutable assent, so that we have to go by the charitable assumption or not go at all: there is other behaviour to appeal to. I take this to be a Wittgensteinian point: mere ostension is always multiply interpretable, and the only way to give it specific content is to bring in an extensive range of behaviour and ‘forms of life’. Furthermore, since it is certainly coherent to keep a subject’s behaviour fixed while varying his environment, we have here a basis for interpreting his speech that does not presuppose that his beliefs fit the facts. Beliefs are not just caused by things outside us: they are also that on which we act, so that how someone acts gives purchase in deciding what he believes. The upshot is that Davidson’s anti-sceptical argument does not go through: there can be true and warranted attributions of predominantly false beliefs. The good news, so far as Davidson’s overall scheme is concerned, is that it becomes possible to accept his semantics without embracing his epistemology.

Davidson’s work combines rigour with imagination, caution with boldness. He shows what analytic philosophy can be like at its best. In tackling head on some the most profound and perplexing questions he has opened up new areas of inquiry, and it is impossible not to learn from thinking through his ideas – even when one disagrees with them. There is a well known genre of philosophical joke – ‘X’s proof that p’ – that parodies a given philosopher’s characteristic style of argument. Davidson’s proof that p goes: ‘Consider the bold conjecture that p. Therefore p.’ (That, for a disciple of Davidson, goes: ‘Davidson has considered the bold conjecture that p. Therefore p.’) Of course, this is as unfair as it is meant to be (it is actually a good deal milder than other examples of the genre): but it does signal one very commendable feature of Davidson’s work – its courage. Davidson wants to answer the big questions, and he is not afraid to muster whatever degree of boldness is requisite to the task. What is amazing is that he has done this while remaining as scrupulously analytical as even the most inhibited of thinkers. Read ‘In Defence of Convention T’.