Friday 3 March 2017

Gombrich, Ashby and Seth on Consciousness

I attended the Manchester Intervarsity club last night for a discussion about “The neuroscience of consciousness”. We watched a video of a presentation by Anil Seth. I wouldn’t necessarily have paid much attention to this had it not been for the meeting - I’m glad I went along.

For all the neuro fetishism and misplaced confidence in the ability to create metrics of consciousness (which is partly what Seth is about) – stuff which makes me uneasy – Seth’s deeper theorizing draws on Ross Ashby and Ernst Gombrich. That’s useful, because Seth is a mainstream cognitive psychologist looking at cybernetics, which enables today’s cyberneticians to make references to empirical things which are going on now, rather than stuff which was happening in the 1960s. (He doesn’t mention Bill Powers “Perceptual Control Theory” – but that it perhaps the closest correlate of what he is articulating).
Seth builds on Ashby’s central idea of constraint: that perception involves cognitive processes of prediction of possibilities which are constrained by the senses. In essence, consciousness is cybernetic: we all observe “what might have happened but did not”. Seth applies the principle not only to perception of the external environment, but to the body. This isn’t a new idea. Apart from Ashby, Robert Rosen’s work in biology, and Daniel Dubois’ mathematical articulation of anticipatory systems (not forgetting Loet Leydesdorff’s unfolding of these ideas at the social level) are all fishing in the same pond.

I think this is basically right, but if I was to take issue with him, there is an implicit assumption (highlighted well by Rupert Sheldrake) that the mind is in the head. I also think Seth is unaware of the sophistication of Ashby’s thought with regard to problems like “analogy”, “induction”, “regularity”, “isomorphism” and so on. But that means there’s a discussion here. The problem with ignoring the first point is that consciousness is seen as an apolitical issue. The problem is immediately apparent in Seth’s work on measuring consciousness (John Searle also suffers from the same problem) – the possibility of “consciousness pills”, of tests for “how conscious is your child?”. You don’t have to be Aldous Huxley to work out the implications.

Gombrich’s influence on Seth is perhaps more subtle – but I think it is equally important. Seth takes from Gombrich the basic assumptions of Gestalt psychology, and the role of reflexive processes in “making reality”. But this gets more interesting. Gombrich’s work on pattern directly referenced information theory (particularly in his “A sense of order”), and by implication, the role of information redundancy. Gombrich’s social network included two other Viennese émigrés: Friedrich Hayek and Karl Popper. I find that all three were very close in their thinking, not just in their friendship.

Hayek also wrote a book about consciousness (his book of 1952 – “The Sensory Order”), and he clearly understood the cybernetic principles which Ashby articulated. Stafford Beer, on meeting him, apparently declared “At last! An economist who understands cybernetics” – only to revoke any approval of Hayek on becoming aware of his right-wing sympathies, and particularly his support for Pinochet and Thatcher. Whilst I share Beer’s disgust for this, Hayek’s work remains of the highest order and sight should not be lost of it.

But Hayek is warning for Seth. Seth’s idea of consciousness leads to fascism. Nick Land, the British philosopher who has become the intellectual voice of the Alt-Right, and who most articulately utters a form of second-order cybernetics, shows us exactly where this stuff leads.

The way out of this is to look more deeply at Ashby and Gombrich. The relationship between constraint, information and description is fundamental. Seth gives us a description of consciousness. Shakespeare does the same – but there is a difference between the two. Gombrich knew very clearly what the difference is; Ashby knew something about how different descriptions interact with one another, which is where that difference lies.

No comments: