To: Tom M.
From: Geoffrey Klempner
Subject: Hume on the self
Date: 28 May 2002 09:41

Dear Tom,

Thank you for your e-mail of 17 May, with your second essay for the Philosophy of Mind program, in response to the question: 'Do you agree with the philosopher David Hume that, "I never catch myself at any time without a perception, and never observe anything but the perception"? Examine Hume’s account of the nature of self, showing the main features that distinguish it from Cartesian dualism.'

You have given a clear account of how Hume's bundle theory explains the identity of a subject, defined as a collection of mental items, at a given time. The key point is that the relation of 'presence to' defines what is known as an 'equivalence relation'. An equivalence relation is transitive (if xRy and yRz then xRz), symmetrical (if xRy then yRx) and reflexive (for all x, xRx). Given these three properties, any domain of mental items will be divided by the equivalence relation into distinct, non-overlapping classes.

Commenting on Hume's theory, you say, 'It is a simple explanation, which adheres to Ockham's razor...If mind can be explained without reference to a subjective witness, then this is preferable to a more complex model.'

So we have two arguments for Hume's theory: 1) the self apart from its perceptions cannot be perceived 2) by Ockham's razor such a self is redundant anyway.

This sounds very much like an argument that might be given against the theory that car engines misfire because of gremlins: 1) No-one has ever seen a gremlin 2) it is possible to explain engine misfiring without positing gremlins.

Can it really be as simple as that?

One niggling doubt concerns our understanding of the relation, 'x is present to y'. Logically, the definition is impeccable. We know exactly what is entailed by any statement of that form. So, for example, given a situation where a lump of cheese is 'present' to a pair of scissors, if the moon is present to the lump if cheese then it is also present to the scissors. Now you know!

If you reply, 'But my notion of presence only applies to mental items, not to physical things!' then you still owe an explanation of what it is about being a mental item that gives rise to the 'presence' relation.

Hume would agree that mental items or 'perceptions' such as itch, or a memory of Scarborough, or the decision to have hamburgers for lunch are 'substances', like the cheese, the scissors or the Moon. You could take everything in the universe away and leave just the moon, or just the pair of scissors, or the cheese. Similarly, you could take everything away and leave just an itch. An itch can be present to other mental items, but logically it need not be. That is a consequence of defining the 'self' as an equivalence relation between mental items. But do we really understand what it would mean to say that, in some possible world, the only thing that exists, under the category, 'perceptions', is a single, lonely itch?

The second problem concerns the 'tapestry theory' of identity over time. On Hume's own theory, a memory of an itch is a distinct perception from the itch itself. Perceptions are defined not only by their capacity for 'presence' to one another but also by their existing *in the present*. So if we look at this itch, what we see is not something that persists from moment to moment, the way physical substances do, but rather:

t0 itch
t1 memory of itch (t0)
t2 memory of itch (t0) and memory of 'memory of itch' (t1)
t3 memory of itch (t0) and memory of 'memory of itch' (t1) and memory of "memory of 'memory of itch'" (t2)

...and so on.

Each line gives a list of momentary mental items, which can exist independently of the existence of lists of items on the other lines. So, just as we did with the lonely itch, we could take a moment in the life of GK or TM and posit a possible world in which the total mental contents of GK or TM at that moment exist, but no other mental items exist in the past or future.

However, it is possible to use the notion of a tapestry to construct an 'ersatz' concept of identity which would look for examples of the above series and call such examples 'a persisting self'. As you note in your essay, this allows for branching lists where using the above example, the series t0, t1, t2, t3 is continued t4, t5, t6... and simultaneously continued t4', t5', t6'....

If one is unhappy about these consequences, positing a Cartesian 'soul substance' will not help solve the problem. This is shown by the series of 'spectator in the theatre of consciousness' thought experiments which you quote.

So where does that leave us?

My conclusion would be that we should reject both Hume and Descartes. The only way to do that is by ditching mind-body dualism.

All the best,

Geoffrey