It’s 100 years since the birth of Alan Turing, and I’m attending the eponymous annual lecture given in his honour under the joint auspices of the British Computer Society (BCS) and the Institution for Engineering and Technology (IET). The lecturer, Prof. Ray Dolan of UCL, intends to review cognitive neuroscience as “hidden legacy” of Turing’s definition, and investigation, of “computable numbers”.
As my wife is a Counsellor with an interest in the working of the brain, this may be of interest beyond the full-house IT constituency currently gathered.
Well, the lecture theatre was crowded so it would have been antisocial to blog as we went. Just a short retrospective, then, and I’ll link to the video replay when it’s available.
Judging from the comments prefacing audience questions after the lecture, I may be in a minority: but to me this was an opportunity missed. We certainly heard an erudite lecture on brain function. And at the start we had a short treatise on Bayesian logic. This deals with the unravelling of uncertain data.
If I understand it correctly, you start with observations which, with a degree of uncertainty, represent the state of what’s being investigated. In the case of Turing’s work at Bletchley Park, they had observations of intercepted coded messages, and were attempting to infer the settings on the coding machine that had created them. In neurobiology, they observe the areas of brain activity with the intention of figuring how the brain works.
This seems like a fruitful parallel, though one of the experiments which Prof Dolan described turned out to have a non-Bayesian interpretation. There’s certainly a computational problem here, described as “Start with a theory of how certain parameters give rise to the data you observe, and attempt to go from the data to the parameters”. I can identify with that: as a spectroscopist, I used least-squares analysis to infer the component absorptions in both Mossbauer and infrared spectra of minerals. But by this time the parallels with Turing had been lost, apart from occasional references back to the original scene setting.
Which was a pity, since I guess most of the people at the lecture were IT people, not neurobiologists. I’d have appreciated a lecture following both threads at once. I still wonder if it’s possible to pursue the parallel, so that Turing’s work and this undoubtedly fascinating field could be explored, as it were, in a “twisted pair” of threads.
But make your own mind up. Follow the link below to the IET’s page for the lecture, and click the link to “Play webcast”.
• From cryptanalysis to cognitive neuroscience – a hidden legacy of Alan Turing: the BCS/IET Turing Lecture, IET 20 Feb 2012