Wednesday, June 22, 2011

Fichte and human agency

To: Jurgen L.
From: Geoffrey Klempner
Subject: Fichte and human agency
Date: 27 March 2003 12:11

Dear Jurgen,

I have just been struggling with your essay on Fichte (16 March). I thought that I would be able to say something intelligent about it (despite the fact that, as I warned you, I have not made any serious study of Fichte) but I can't.

This is just to warn you that what follows is going to be pretty unintelligent, or at best ignorant.

I understand Fichte to be grappling with the nature of self-consciousness, its 'reflexivity' or ability to become an object to itself. It would have helped me in my Fichtean striving to grapple with your essay and assimilate it into my world if I could have found a hook which related your exegesis to the reflexivity problem. But I seem to have missed it. So this puts me at an even greater disadvantage.

Let's start at the other end. It is possible to read the Tractarian 'the world is my world' into Kant's account of the spatio-temporal form of appearances. If we take a subject which is capable of making judgements and feed it a stream of data, there is no 'subject', no 'I' unless the data is structured so as to enable the construction of a *theory* in which percepts are located at places as well as times, while the location of the subject in relation to those places is itself part of that same theory.

Kant goes into great detail about the mental machinery which the self employs in constructing this theory (Strawson calls this aspect of the Critique the 'transcendental psychology of the faculties'). It is easy to dismiss this (as Strawson does) as irrelevant quasi-empirical speculation masquerading as metaphysics. For the essential point about the limits, or necessary form of experience, has already been established.

But there is a glaring metaphysical problem which Kant does not tackle at all. In the foregoing account, we have taken the terms 'self', 'judgement', 'theory' for granted, as if we didn't need to ask what it is to be a self, or what it is to make judgements. Kant starts at the point where a fully formed I is already making judgements, theorising, revising its theory in the light of new data, which are themselves interpreted in terms of that theory. In this picture, the I is just a point of view, possibly embodied, possibly not (the refutation of idealism only requires spatial location, not physical embodiment). How can the I ever become an object to itself? What is it to be an entity which is 'an object to itself'?

On the Kantian (or quasi-Kantian) view which I have sketched, of course no data are assimilated raw. We start with an on-going theory of a world of objects distributed in space. Perception slots new percepts into a pre-existing picture, or when, things go awry and the theory clashes with perception, both the theory and our originally unreflective interpretation of our present experience need to be revised. I am saying this as a challenge to you to explain how Fichte goes beyond that idea.

Or let's look at memory. Of course, the idea of memory as a data bank from which items can be retrieved is useless as an explanation of what it is to be a subject, to be conscious of one's self-identity. Did Fichte first point this out, or is what I have just said simply a consequence of Kant's assertion that time is the 'a priori form of inner sense?'

Action provides another angle, potentially the most illuminating way to approach Fichte. You only hint at this, but it seems that something is essentially missing from the model which I sketched above, of a possibly disembodied I, gliding around a world of objects which it 'encounters' passively as percepts which are recognized, categorised, slotted into a theory. What is missing is that the subject has the capacity for physical and not just mental agency. The subject physically grapples with objects. Its interest in objects, for example the mosquito, is formed by its physical needs and desires which determine which aspects of the world of its perception are important and which aspects to ignore.

This is an important point to make in relation to AI research. Some more enlightened AI theorists concede that if there were an intelligence program, it could only 'run' on a computer endowed with physical needs and the physical means - arms and legs, hands and mouth - to satisfy them. Why is that concession not enough? What are they still missing?

What is more important to you, to offer an interpretation of Fichte, or to make the point about the incompatibility of the AI model with our best knowledge of the structure of the human mind? If it is the former, then you haven't said enough (e.g. you all-too quickly withdraw from 'Fichtean dialectics' to the image of the lake). If the latter, then might it not be better to launch off with that and insert somewhere a section on Fichte - or even a long footnote - to show where these ideas originally come from?

I have a strong intuition that the notion of human knowledge as a 'theory' is fundamentally wrong, not just wrong in detail, in the way you might add extra bits to take account of the fact that the subject who constructs this theory also finds it quite useful. Any number of philosophers come to mind (starting with Schopenhauer, then Marx, James, Heidegger, Macmurray). Fichte, in looking at the primordial structure of the self, promises something deeper. I wish I could see what this is.

All the best,

Geoffrey