To: Kaz K.
From: Geoffrey Klempner
Subject: Hume and Kant on causality - an undue scepticism?
Date: 27th February 2009 11:16

Dear Kaz,

Thank you for your two emails of 19 February with your first positioning essay towards the ISFP Fellowship Award, and the revised version of your essay, entitled, 'Hume and Kant on Causality - an undue scepticism.'

I very much like the topic as well as your focus on that topic: broadly, the question is how we can accommodate notions such as causality within an empiricist, naturalist framework, avoiding the extremes of Humean scepticism as well as the metaphysical danger of a Kantian two-world solution, which 'saves' the concept of causality for the empirical/ phenomenal world but then gives it up in the face of the unknowability of the noumenal world.

I notice that you have Bhaskar ('A Realist Theory of Science') as well as Rorty ('Philosophy and the Mirror of Nature') in your bibliography. These are right on target so far as your topic is concerned. A work which you might not have come across which is equally relevant is John Macmurray 'The Self as Agent' (as well as the second volume of his Gifford lectures, 'Persons in Relation'). Macmurray's aim is to replace the 'I think' of Descartes with the fundamental principle, 'I do', using this to lever the concepts of knowledge, causality, person etc. by means of arguments similar to Kant's 'transcendental deduction'. Macmurray broadcasted for the BBC, and is very much a 'communicator'. The down side of this is that his arguments lack the kind of rigour one looks for in academic philosophy. This resulted in his work being marginalised. Macquarrie recognizes his importance as the 'British Existentialist' (a not altogether accurate description).

Obviously, you haven't gone into too much detail. However, from what you have said, I can see some important questions relating both to Hume and to Kant. Assuming that you are right, the question is where exactly Hume and Kant make the fatal error, and how this comes about.

Hume's view of causality has been misunderstood. He has two, quite distinct, accounts of causality, the first concerning the truth conditions of causal statements, and the second giving an account of the mental processes whereby the concept of causality arises in the first place.

The truth conditions of 'A caused B' may be stated (at a first approximation) as, 'For all x, if A(x) then ceteris paribus B(x)'. Hume doesn't spend too much time on the question of how a causal law may be stated, but it is crucial and, arguably, the weakest element in his position. It is impossible, in principle, to completely unpack the ceteris paribus clause. The stone caused the window to break, agreed, and the truth of that statement depends upon the truth of a universal generalization that given a stone of sufficient weight and velocity and a window of sufficient fragility, etc. etc. a breakage will occur. But what exactly is 'sufficient'?

Such a generalization, referring as it does to all places and all times, can never be conclusively verified. But is that scepticism? It is not clear that it is.

Elizabeth Anscombe in 'Causality and Determination' mounts a fierce attack on Hume's position here, arguing that we can be more certain of a causal connection than we are of any generalization, and therefore that the truth conditions of causal statements cannot be given by any universal generalization, but I find her position bizarre. You might not.

Hume in the Treatise (in a section on rules for judging causes) in fact goes to considerable lengths to show how we are able to distinguish causal connections from accidental conjunctions. He doesn't see a problem, and I'm not sure I do either. This is basically Carl Hempel's view. Causal explanation consists in deduction from a law (the so-called 'DN model'). There is a lot more to say (Davidson on causes and events, Mackie on 'INUS' (=insufficient part of an unnecessary but sufficient) conditions. But all this is consistent with the view that the truth of a causal statement *is* ultimately a truth about a causal law.

On the other side, however, is the question exactly what is our 'concept' of a cause, how we acquire the concept in the first place. Here, Hume is unconvincing. Mere observation of regular connections between our experiences would never give rise to the concept of a cause if it were not for the fact that we are agents, capable of interfering with the course of nature. A 'cause' is identified as something you can change or control out of the many factors which all play a part in a given event.

Moving to Kant, if you read Strawson's 'Bounds of Sense' you will see that Strawson thinks that Kant's project can be saved, provided we just get rid of the stuff about noumena. Kant's view is not simply (as you seem to imply) that human beings have an in-built capacity to judge causes, or an in-built a priori concept of cause -- although this is part of his theory. He gives a 'transcendental argument' for the view that the concepts of cause, space and spatio-temporal continuant are necessary conditions for the possibility of experience as such: the hypothesis of a subject merely noticing patterns of subjective experiences is incoherent because the identity of the subject over time cannot be defined in the absence of an objective component, i.e. how things are in the spatial world which the subject perceives.

All well and good; but I think Strawson is wrong. (I had a memorable term of 1-1 supervision at Oxford by Strawson in Trinity Term 1977 where this issue was much debated!). Unless you start with the 'I do', the necessity of the self as agent and not merely a passive observer, all you get is a coherent dream world, and an ersatz concept of 'cause'. Unfortunately, Macmurray doesn't provide enough argument. In my Oxford B.Phil thesis I enlisted Wittgenstein's private language argument to fill the gap. The consequence of this is a peculiar (or maybe not so peculiar) reading according to which 'publicity' in itself is insufficient to establish meaning rules. The language users must be agents (hence, 'form of life'?). The illustration I gave for this was the hypothesis of a race of intelligent trees, who spend their time discussing the weather and the comings and goings of the various animals in the forest. On my view, this hypothesis would be incoherent. (When I say trees I mean totally immobile living creatures, not Tolkien's 'ents'.)

These are just some considerations to think about. As I said, we are broadly in agreement, but there is a lot of room for different lines of argument as well as some uncertainty about exactly what it is about Hume's or Kant's views that we should be attacking.

My view would be that you have your hands full looking at the concept of 'cause', without needing to bring in the other things you mentioned, like other minds etc. But let's see.

All the best,

Geoffrey