Russell's Transcendental Argument in 'An Essay on the Foundations of Geometry' - page 2

Written by A.C. Grayling.

It is also worth noting at this juncture the argument Moore gives in NJ for the realist commitment he opposes to what he takes to be Russell's point here. It is remarkably weak. Writing of 'the nature of a proposition or judgment' Moore says, 'A proposition is composed not of words, nor yet of thoughts, but of concepts. Concepts are possible objects of thought; but that is no definition of them. It merely states that they may come into relation with a thinker; and in order that they may do anything, they must already be something. It is indifferent to their nature whether anybody thinks them or not' (NJ 179). 'Concepts form a genus per se, irreducible to anything else' (NJ 178-9). This Platonism about concepts itself seems to be enthymematically premissed, among other things, on the view that relations of acts of judging to their contents are external. But nothing is offered in support of that claim, and no other reason–of the better kinds needed–is offered for hypostasising concepts and judgments. Since Moore goes on to assert that the world is made of concepts, thus realistically conceived, some such argument is surely called for. (In Frege one at least has a strong motivation for assigning Thoughts to a Platonistically-conceived Third Realm, namely that the publicity constraints on sense require that it have a greater degree of objectivity than mappings across the psychological states of language-users can yield. I argue elsewhere that Frege's requirement, backed as it is in this coherent way, is overstrong.)

Russell's transcendental argument has it, as we saw, that a form of externality is necessary for the possibility of experience, because the givens of experience are complexes, that is, have parts which must be external to each other. Moore's first point is directed at this proposition. He quotes Russell's claim that necessity always involves a ground, and says, 'But this ground must itself either be simply categorical, or else it must itself be necessary and require a further ground'. In the former case we are actually trying to deduce an a priori proposition from one that, as categorical, is merely empirical; in the latter, which Mr Russell seems in the end inclined to accept, we must either allow an infinite regress of necessary propositions, and thus never reach the absolutely a priori, or else we must accept the view that knowledge is circular, and shall in the end return to the proposition from which we started as empirical, as being itself the ground of necessity of the a priori, and therefore itself as much a priori as the latter. Mr Russell seems actually to accept this latter view (pp 57-60)–a view which renders his logical criterion nugatory, since it asserts that 'that which is presupposed in the empirical equally and in the same sense presupposes the empirical.'

Moore is here arguing in effect that, leaving aside a regress of conditions which never terminates in an absolute a priori, however one otherwise tries to state the case the starting point is the nature of experience and the conclusion concerns what is required for it to be thus and so; and therefore 'to show that a "form of externality" is necessary for the possibility of experience, can only mean to show that it is presupposed in our actual experience' (CN 399). And Moore immediately sees that 'this can never prove that no experience would be possible without such a form, unless we assume that our actual experience is necessary, i.e. that no other experience is possible' (ibid).

It no more occurred to Moore to consider whether this ancillary requirement can be met than, evidently, it occurred to Russell, who like most Kantians might have simply accepted that Kant was right to allow the possibility of other forms of experience–for example he allowed that animals might have forms of spatial experience quite different from ours. For this reason Moore read Russell as having to find another way out for the argument, and therefore attributes to him what he calls the 'subterfuge' (ibid) of the presuppositional argument. Moore thus views Russell's point about presuppositions as an ad hoc step taken to avoid a difficulty, not as a characterisation of the argument to the necessity of a form of externality itself. As we saw, in this he is wrong. For it is of the essence of a transcendental argument that, in identifying what is necessarily presupposed to something which is the case, it tells us something else about what is the case as its condition. Instead of being a criticism of Russell's strategy, Moore's anatomisation describes it.

But it does show that the condition identified as necessary for spatial experience is relative unless the ancillary requirement be met that no alternative experience is possible. At first sight this seems to pose too tall an order. But subsequent debate in philosophy has provided an interesting if controversial means of showing that there might indeed be a way to satisfy the requirement. (A polite form of execratory howl usually greet this assertion, chiefly because of now orthodox assumptions about possibility. When possibility is understood as an epistemic notion–as conceivability, in other words–the task is recognisably much more manageable. But the argument I'm about to borrow does not require a demonstration that most philosophers until the time of Kant were right to construe the modalities–all of them–as epistemic notions.) The argument, furnished by Davidson, is his celebrated case for saying that the idea of alternative conceptual schemes is incoherent. It has more texture than I am going to bring out here , but in any case I only need its bones for present purposes. The argument goes as follows.

The idea of 'other ways of having experience' can be generalised into the idea of alternative conceptual schemes–alternative, that is, to our own, for arbitrarily constrained 'us' (it does not matter how restrictive one is about who 'we' are in talk of 'our' conceptual scheme, for the argument goes through whatever one says about this)–and conceptual schemes can be provided with criteria of identity by identifying them with languages or sets of intertranslatable languages. Then, in turn, questions about the possibility of the existence of conceptual schemes other than our own, perhaps existing undetectably from the point of view of our own, can be framed in terms of translatability. In this idiom, the question of whether there can be other or alternative schemes comes down to the question of whether there can be languages that we cannot translate. But since we can have no grounds for treating as a language anything that cannot be translated into our own language–since, in effect, the criterion of languagehood is 'translatability into a familiar idiom'–and since intertranslatability defines membership of the same scheme, the conception of distinct schemes–in particular, of mutually inaccessible schemes–is seen to be incoherent. But to say that there can be no such thing as an alternative scheme is, transposing back into the idiom of ways of experiencing, to say that ours can be–still taking our modalities seriously–the only such way. The argument is a strongly anti-relativist one; which is appropriate, because Moore's rebuttal of Russell's project takes precisely the form of a sceptical counter-claim to the effect that forms of experience are relative. As is usual, no argument is offered for the claim: its sole ground seems to be the now orthodox view of possibility mentioned.

One characteristic reaction to this application of the Davidson argument–here cavalierly seeming to keep the 'Ptolemaic counter-revolution' going–is to say that no contradiction infects the idea of something which is recognisably an untranslatable language. The usual example offered is the Minoan linear script. A response to this example is to point out that the inaccessibility here at issue is not to a language but to a script embodying a fragment of what we suppose to be a language, and moreover one not in use, which places a contingent barrier to translation which by itself is not relevant to the question of its languagehood. But another, more general, response is to contrast the notion of an untranslatABLE language with that of an untranslatED language. Contingent barriers to translation of one established language into another (such as the current but, I surmise, reducible inability of most of us here to translate Magyar into English) do not, obviously, put them beyond the pale of languagehood; which tells us much, on reflection, about what would. Rather, the accessibility point rests on these general thoughts: that the concept of differences between languages, or schemes, or ways of having experience, essentially trades upon there being enough access between supposed alternatives for the differences to be apparent. There has to be a background of shared assumptions and beliefs giving rise to the degree of mutual comprehensibility which alone makes differences recognisable. And this is not a point about mere cultural relativities–about what might be called anthropological divergences, differences of opinion and high-level social practices–but about the most basic levels of cognitive activity turning on individuation, reference, property-attribution and assent. In the fuller treatment of these points referred to above, I argue that this shared basis has to be rather rich and fine-grained, to the extent that it rules out indeterminacy of reference . As applied to the question of language, one can connect points about shared background beliefs and holdings-true as a condition of getting translation started, with points about the recognisability of devices in the language for reference, predication and assent and dissent: so a much more articulated grasp of an alien tongue is required even by Quine's philosophical anthropologist before he can fairly get going. Alternatively put, the requirement for starting a translation manual is that there should already be one available.

Another response to these claims is that there are certain identifiable possessors of conceptual schemes which are not possessors of language, for example cats and cows, so the identification of schemes and languages fails, and with it the putative access required and therefore afforded by translation. One can leave aside the short answer from the inescapability of 'reading-in' in such cases–which might be (so say those who do not belong to a cat) what our attribution of concept-possession to languageless creatures consists in. We can leave this aside because it is obvious that one can give a strongly motivated and cogent argument for attributing concept-possession and attitudes, at least to a certain level of complexity (remembering the strictures of Frege and Wittgenstein on this subject), to at least the higher mammals, among other things on the grounds of the successful pragmatics attached to doing so: as witness Fodor's cat. The point might be put by saying that much of what can be attributed concerning the beliefs and intentions of such mammals is representable in one's own language in statements having as good a chance of being true on the evidence as those about creatures capable of making their own avowals. On this head, there is nothing second-class about third-party attribution.

These points are made, remember, in response to questions about the intrinsic merits of Moore's attack on EFG, and whether Russell was without recourse in defending the philosophical strategy there adopted. These points at least show that both Moore and Russell were too swift here. And this can be substantiated by noting that there is of course a similar but much weaker response, or set of responses, that can be made to this aspect of Moore's criticism. As Strawson has argued (in Scepticism and Naturalism and elsewhere) there is much to be gained from an investigation of what our kind of experience requires, even if it is not the only kind there can be. This either allows philosophical interest to what Moore dismisses as merely parochial, or it refuses to interpose so impermeable a membrane between psychological and philosophical considerations as was all the rage at the turn of the century. But in the presence of a stronger argument for the cogency of the general enterprise–I do not say, Kant's or Russell's in particular–it is worth contesting Moore's rejection of their style of argument more vigorously.

It might be asked what difference would have been made, outside his philosophy of mathematics, if Russell had retained some part of his early convictions about a priori knowledge. With the exception granted, one answer is: less than one might suppose, since it would not have interfered with his pluralism, his atomism or with his adoption of certain sustainable versions of non-Platonic realism; and another answer is that it might have offered resources to his epistemology from the lack of which they badly suffer. This is well illustrated by one salient consequence of dispensing with a priori constituents in knowledge, namely, Russell's reliance on the notion of acquaintance. His response to Moore's Platonism about concepts and judgments–which remained when the objects of acquaintance had become simultaneously far more various and refined–was to treat our relation to them as direct, theory-free, unmediated and conditionless. The relation is curiously thin and undefined; it comes without constraints, as if it were primitive; and in so far as it admits of being described as a mental operation, it is distinctively passive, the very opposite, one might say, of a relation in thought or experience in which some act–of perceiving or judging–plays a constitutive or partly constitutive role with respect to its objects. Now one need not be interested in specifically Kantian strategies for understanding how this works to feel the deficiency in the theory of acquaintance. One thing one can safely say is that if he had not abandoned the approach in EFG so entirely, Russell would have written differently later about knowledge and perception.

One of the really interesting features of EFG is that in it Russell is not an idealist; he as an anti-realist. (So is Kant in fact, on a certain best reading). The differences are considerable. Idealism is a metaphysical thesis which asserts that reality is fundamentally mental; it is a Mind (e.g. Bradley) or it consists of minds and their ideas (e.g. Berkeley). Anti-realism is an epistemological thesis, which asserts that the relations between thought and its objects, perception and its accusatives, experience and its targets, language and the world, or whichever of these (different) pairings one takes as the focus, is not external, as realists claim. To say that the relations in question are internal is far from saying that thought or experience creates the world or is in some other way ontologically responsible for them: that is metaphysics, and specifically idealist metaphysics. talk of the nature of the relations between mind and world is purely epistemological; it is not of course independent of considerations about what there is, but it does not by itself involve constitutive ontological claims. Now, Russell did not see the difference, so his unpreparedness to defend against Moore's attacks on idealism led him to abandon his anti-realism, a much more moderate position which he replaced, at first, with a very immoderate realism. It is the anti-realist features of Russell's thought in EFG which would have served his later epistemology well.