To: Mark S.
From: Geoffrey Klempner
Subject: Hume on the 'fiction' of personal identity
Date: 6th April 2010 12:08

Dear Mark,

Thank you for your email of 30 March, with your essay for the University of London Modern Philosophy: Descartes et. al. BA module, in response to the question, ''The identity which we ascribe to the mind of man is only a fictitious one' (Treatise, I. iv. 6). Did Hume deny the unity of mind, or simply advocate a new view of what it is?

In my response to your last essay on Hume's two definitions of 'cause', I placed fairly heavy emphasis on Hume's naturalism -- his vision of a 'theory of human nature' which located the mind in the world of nature, as something whose behaviour is explained by laws. That this is only half the story becomes apparent both in Hume's discussion of 'on scepticism with regard to the senses' and in his account of the 'fiction' of personal identity. In brief, Hume is unable to reconcile the naturalistic view with his methodological solipsism.

It is interesting that you only mention the possibility that 'ownership' of perceptions depends on association with a 'particular body'. This is actually A.J. Ayer's view, rather than Wittgenstein's. It's a glaringly obvious solution to the problem of the unity of a bundle of perceptions at a time, and also the identity of that bundle over time. But it is a solution which is unavailable to Hume, for the reasons which you give.

The unity of a bundle of perceptions is a breeze, if we follow Hume's assumption of 'mental transparency' as you term it (despite what you quote Garret and Pears saying). We simply define a relation, 'co-presence' which is symmetrical, reflexive and transitive. All the perceptions in the universe are automatically divided into bundles, where A and B belong to bundle X if and only if A is co-present with B. No bundles can overlap (because then they would be the same bundle). Problem solved.

Identity over time is more tricky, because all we have are present perceptions. The bundle of GK's perceptions that existed a minute, or a year ago, is present in the current bundle as a memory impression. Note that this is not a story about overlapping bundles. We have already seen that bundles can't overlap.

An immediate consequence of this definition, however, is that it becomes perfectly conceivable that you and I should undergo the experience of changing bodies. I wake up in Australia and you wake up in the UK. Methodological solipsism makes this possible.

Again, as we've discussed, there is a solution which preserves a modified version of solipsism/ mental transparency, namely Kant's. What you quote as the 'mind-world separation problem' is actually the solution: as the early Wittgenstein expresses it in the Tractatus: 'I am my world'. Experiences are necessarily interpreted in terms of a theory according to which they result from the movement of a subject (which can be 'embodied' or a disembodied point of view) through a spatial world. The finesse is that we are, in effect, solving a large number of simultaneous equations: any given sequence of experiences can be explained in more than one way, we can invent any number of possible worlds which would explain their occurrence. But as the sequence gets longer, the number of alternatives decreases.

The unity of the self -- both synchronic and diachronic is, on this view, a necessary feature of the theory. It doesn't need to be explained or defined further. It is simply a necessary condition for the possibility of experience.

The only problem with this solution from Hume's point of view is that it doesn't require the assumption of naturalism. The mind does not have to be part of nature (because, e.g. it can be a mere disembodied point of view -- although Kant never actually says this explicitly). The theory is consistent with naturalism, but it is also consistent with Kant's theory of the phenomenal and noumenal worlds.

Again, with regard to the 'inadequacy problem', the difficulty for Hume is that it looks like he is forced to identify what there is with what we can know, on the assumption of methodological solipsism. Locke and his notion of observable properties flowing from a real essence would be a solution, but one which Hume's principles will not allow him to embrace.

It is easy to conclude that Hume was right when he composed his Appendix, that he realized that his theory of the self was inadequate. (I liked what you said about the anxious patient.) But I think he was being over-scrupulous. His theory isn't that bad. The alternatives (like Kant, or Ayer, or the later Wittgenstein) have their own cost. The beauty of Hume's theory of the self is in its purity and simplicity.

All the best,

Geoffrey