Monday 24 March 2014

Mutual Information, Mutual Redundancy and Interdisciplinary connections

One of the advantages of working in a small University is that interdisciplinarity can become a habit in the coffee bar. I had a great experience this morning talking to a colleague who is relatively new to the university and whose disciplinary expertise is in biology. The conversation revealed some of the problems with current statistical techniques for genome analysis. "I and a Chinese colleague have started to look at Mutual Information and Entropy," he said. So up crops Shannon once more (see http://en.wikipedia.org/wiki/Information_theory) - and perhaps more importantly is the connection between the problem faced by this academic and the problems which I am spending most of my time thinking about which have no relation to his disciplinary area at all. Indeed, what happened between us is precisely indicative of mutual information!

But more interesting is the fact that the adoption of the Shannon equations present new kinds of problems which have been explored in the (again unrelated) domain of the economics of the knowledge economy - particularly through Loet Leydesdorff's work. Mutual Information, it turns out, is fine in two dimensions. But most communications in the world, including (I guess) those interactions which happen at a molecular level, are not in two dimensions. These are many-dimensional communication situations. Like education. Under these conditions, Shannon's equations deliver inconsistent results. Which led to a discussion about the value of looking at mutual redundancy rather than mutual information. So I could send him this latest paper which is appearing in the cybernetics journal Kybernetes: (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2358791). Geneticists reading a cybernetics journal - well, that's the kind of thing that ought to be happening in a University!

Actually, our discussion started with my colleague's complaint that our University ought to be doing more research, and that in research-active universities, teaching isn't so important. I challenged this view by arguing that Bolton's students provide rich opportunities to study the human experience of learning and teaching: Cambridge isn't lucky enough to have these opportunities! If we take this seriously then important cross-disciplinary fertilization can occur so that we gradually see serious scholars taking an interest in all kinds of students, their learning processes and the relation between the University and its community. It is because of my interest in these problems that my studies have taken me to Shannon, Leydesdorff and many other cybernetic theories. But then it is because of this, and because of the shared situation we all have in our institution (some of it very political and not very constructive) that new connections can be made between the study of learner experience and the latest techniques of biomedical engineering, or indeed many other applications. This too creates the conditions for mutual information and mutual redundancy.

The idea reinforces the mutuality of university life in general. However much managers might want to see staff as 'units of production', that simply isn't how it is. When you gather individuals together each of whom is committed to the search for knowledge in whatever domain, they will find rich common ground between them, from where innovations are born. But look at any single individual and you will see different weaknesses and strengths in performance. Individual performance metrics are dangerously crude. But perhaps what strikes me most is that it is in the cracks within the institution which (like many others) appears to want to box everything in and bring it under central control, new things appear which are quite unexpected. This is where the hope is - but perhaps we could do with a few more cracks in the plaster!

No comments: