Although I haven’t found the time to explain the broad motivation of what I am doing, I will mention here what I am currently working on.
– Evolved 3 and 4 node circuits (CTRNNs) for a discrete associative learning task resembling Phattanarasri’s but with one crucial difference that allows moving seamlessly towards the continuum.
– Analysed the best and smallest successful circuit. The result concurs with Phattanasri’s analysis in that a finite state automaton can be extracted to explain the internal dynamics generating the learning behavior.
– Evolved 3, 4, 5 and 6 node circuits on the same associative learning task but on the continuum – everything else the same. There’s some analysis on the evolutionary success in this part, but it is not the main focus of my work.
– Currently analysing the underlying dynamics of the best and smallest successful circuit (5 node CTRNN).
The aim is to produce a journal publication (which should be at the heart of my PhD thesis) out of the explanation of how the 5 node circuit performs the learning behavior and how it is different from the evolved circuit for the discrete task. This is work in collaboration with Inman and Randy.