philosophy is for everyone
and not just philosophers

philosophers should know lots
of things besides philosophy

Philosophical Connections

Electronic Philosopher

Feature Articles

University of London BA

Philosophy Lovers Gallery

PhiloSophos Home

International Society for Philosophers

Descartes' proof of God, and Hume's bundle theory


To: Vincent L.
From: Geoffrey Klempner
Subject: Descartes' proof of God, and Hume's bundle theory
Date: 19 April 2006 15:07

Dear Vincent,

Thank you for your email of 13 April, with your University of London essay, 'Descartes attempt to prove the existence of God in Meditation III is not merely a failure, it is a philosophically uninteresting failure.' Discuss, and your email of 18 April, with your UoL essay, 'Why does Hume think he is only a "bundle or collection of different perceptions"? Is his view sustainable?'

Existence of God

I'm glad that you think that Descartes' argument from perfection is philosophically uninteresting, because that gives me more to write about - because we disagree.

This essay is well laid out. You give a clear exposition in five steps of the argument from perfection. Then you give your reasons why the argument does not persuade (and moreover does so in an uninteresting way). Let's look at each of these in turn.

'In essence, it seems as though Descartes' argument is that whatever X may be, the cause of an idea of an X is an actual X. What then should we say of a unicorn?'

This is a good example, but I don't think that the objection holds. In fact, this would be a perfect opportunity for Descartes to illustrate what in his view is so special about the idea of perfection.

I have an idea of a unicorn. This is something that requires explanation. However, as Descartes indeed suggests in Meditation 1 there is a perfectly good explanation for concepts of this sort. ('Nevertheless it must be admitted at least that the objects which appear to us in sleep are, as it were, painted representations which could not have been formed unless in the likeness of realities; and, therefore, that those general objects, at all events, namely, eyes, a head, hands, and an entire body, are not simply imaginary, but really existent.')

In the case of unicorns, I have horses before, and horns before, and white things, and (in case you didn't know) I know what a virgin is. (Unicorns have the unique ability to tell whether a girl is a virgin. When a unicorn sees a virgin, it lays down at her feet.)

There is, however, an argument deriving from the work of Saul Kripke ('Naming and Necessity') that it is impossible to have the idea of a 'unicorn' in the sense of a creature that might possibly exist somewhere (discounting the virgin testing attribute). If we found a creature that looked like a unicorn it wouldn't be a *unicorn* but merely a 'creature that looks like a unicorn'. This is because 'unicorn', like 'horse' functions as a natural kind term, and as Kripke and Putnam argue, the meanings of natural kind terms depend on the actual existence of the kind to which we have given the name. - This looks like a replay of Descartes argument about God, and is certainly not without philosophical interest.

Think of the most outlandish thing you might possibly imagine. I have an idea of travelling in a time machine into the past and accidentally killing my great great grandfather. Well, do I? To have an idea is more than merely thinking you have the idea (more on this below). It must be a coherent idea. Arguably, the time travel scenario I have described is not coherent.

So Descartes thought is that ideas must come from somewhere, and moreover we have the capacity to tell, through reason and reflection, whether our ideas are coherent.

Let's look at the idea of 'cause and effect'. You say that this idea, 'does not seem to have any basis whatsoever'. Hume famously criticised the 'a priori' notion of cause, but his 'principles for judging causes and effects' explain why we make judgements of this kind (whose meaning Hume explains in terms of unrestricted universal generalization). Hume, finding in himself an 'idea' of X, would indeed consider it to be a perfectly well framed question where this idea comes from in a causal sense, in his terms, how it is formed from ideas and impressions through the process of association of ideas.

Hume, of course, would disagree on the question of whether the idea Descartes identified was indeed the idea that Descartes claimed it to be.

Is it possible for an imperfect being to have an idea of perfection? A good way to frame this question is in terms of the concept of infinity. Human beings are finite, yet we seem to be able to talk about, e.g. the infinite series of integers. (There is an excellent book on this theme, A.W. Moore 'The Infinite'.) If we conclude that it is possible to frame the concept of an infinite series, how is it possible? That looks like it could be an interesting question in the philosophy of mathematics.

The continental philosopher Emmanuel Levinas (in 'Totality and Infinity') gives central prominence to Descartes' idea of perfection, arguing that its true meaning can be understood in terms of my notion - which seems paradoxically impossible to explain - of the 'other person'. Establishing the legitimacy of this idea would be nothing less than a solution to the problem of solipsism.

Hume's bundle theory

It takes you over two pages to get to your answer to this question. As a general rule, in an exam should concentrate on answering the question and nothing but the question. If you take up too much space with preliminaries, then you obviously will have less time to answer the question. Also, there is a danger that you will lose marks from impatient examiners.

I realize that these essays partly serve the function of collecting together the results of your studies, and this is the reason why you spend part of the essay discussing Hume attack on the notion of the self as substance.

I disagree with Stroud and (I think) also with Dicker about the coherence of the bundle theory (although I am not a Humean).

Let me try to restate Hume's theory in a way which makes it appear somewhat less implausible.

First, we have to distinguish identity of a subject at a time from identity over time.

Identity at a time. Assume that there is such a thing as one idea being 'co-present' with another idea. Let's take your own example. The impression of 'woman pushing a pram' is co-present with 'car crashing into tree'. It is not co-present with the impression 'foul tasting instant coffee' (I ran out of tea bags). However, the impression of 'foul tasting instant coffee' is co-present with the impression, 'Matrix screen saver' (currently running on my laptop).

So here we have the basis for two individual bundles. Co-presence is an all-or-nothing matter. So all we have to do is take any impression, and consider all the impressions that are co-present with it. The result is a neat partition into discrete bundles with no overlap. Maybe as I look out of my window I see a woman pushing a pram. The very fact that this impression of 'woman pushing pram' is not co-present with the impression of 'car crashing into tree' logically shows that here we have two impressions, not one, belonging to two different bundles.

Next comes the tricky bit. We have to construct an identity over time.

A moment after the car crashed, you have the vivid memory of what you saw, plus some new impressions. This is how a bundle 'connects' with a bundle which existed a moment ago. If bundles were only bundles of impressions, then there would be nothing to have a bundle theory about, because there would not even be the appearance of identity over time.

So each bundle contains with in it a cinematographic impression of a series of bundles over time. There can be overlap between the cinematic impressions of two distinct bundles (we both have 'the same' memory M an event that happened at time t), but this does not make the memories part of the same bundle. My memories are co-present with 'Matrix screen saver' while yours are co-present with (say) 'Fish tank'.

Where does Hume go wrong?

As a matter of contingent fact (for Hume), these cinematographic impressions are not random but show clear patterns and lines of development. As Kant subsequently argued, this cannot be a mere matter of fact but is an a priori condition for the very possibility of experience. So arguably here Hume was wrong.

What Hume fails to take account of is that ideas and impressions form the data for a 'theory' of spatio-temporal objects in a world. If it wasn't for the possibility of forming such a theory, there could be no such thing as an 'individual'. This is how Kant substitutes his 'a priori unity of apperception' for Descartes' 'unity of substance'. Experience would be impossible if we were not able to identify a 'self' as the subject of experience. However, this 'self' is not Descartes' thinking substance but an empirical self, a subject which itself traces a path through space and time.

All the best,