Plans for the next 3 months: start of 3rd (and hopefully last) year of my PhD.
1. Understanding how evolutionary techniques explore CTRNN parameter space and generating practical..
There are a number of projects that Randy and I have been discussing, and that I know Inman is interested in, too. The ideas that we are thinking about are very much related to the area that I tried to explore during my master’s dissertation. In fact, the whole neutral networks while exploring CTRNN space is a subset of the bigger picture that we are thinking about.
My motivation for thinking about a new big area of research to undertake is making plans for doing a PostDoc in an area of my interest – as oppose to take up somebody else’s project – which defeats the whole purpose and excitement of research.
So, the plan is begin thinking about writing a grant proposing to research these ideas in collaboration with Randy and Inman.
1b. Start looking around for other opportunities for doing an interesting post doc somewhere not too far from London. It would be rather interesting to be able to get experience in working on animal learning. Something around the lines of Hillel Chiel’s lab. There’s some of this going on at Sussex. But I fear that I have fallen into the ‘artificial life’ stereotype there and I wouldn’t be taken seriously if I were to try to switch.
2. Work on the sensori-motor learning experiments. The aim is to produce a ‘brief’ proceedings paper for the IEEE symposium on artificial life to be held in Hawaii 🙂
3. Finish writing the Associative Learning on a Continuum paper and submit to journal.
4. Work on the embodied associative learning experiments. Aim to produce a second journal paper, oriented more towards biology – the c. elegans community if possible. This should comprise a second step in complexity in the associative learning experiments. First step: from discrete stimuli to stimuli on a continuum, leading to changes in the internal dynamics, from FSM to CSM. Second step: from stimuli on a continuum to embodied/active stimuli. How will the internal dynamics of the evolved agents change? I see this as the path towards my PhD dissertation.
5. Follow Randy’s dynamical systems course with Strogatz’ book – do the homeworks and everything.
6. Become more involved in the teaching of Inman’s artificial life course.
7. Write about what learning is. ! . well, write about why I think people (e.g. JK, MB, ED to name but a few) keep thinking that the agents that I have evolved are not learning. There are several different concerns. One is to do with mechanistic learning being equated with behavioral learning. I make explicit that I am studying only the latter. Therefore that concern is not too relevant. The more important one is the concern that it is all potentially ‘in-built’ in the dynamics of the agent. There is the common feeling that learning is more ‘open’, more unexpected, not just the discovery of new regions of dynamics. This is a more difficult concern to deal with. Particularly hard to deal with when it is human learning that is being considered. I think it relates to the question of autonomy. Are we completely capable of learning new – never previously experienced (in evolutionary time-scales) experiences? Or are we evolved to learn a certain set of things – finite set – and anything else outside of that set break us? Are we completely free to change the underlying rules of our internal dynamics into any dynamics possible? Or are we very much constrain by a set of rules and set of dynamics? Can it be the case, that humans have develop such a large set of possible dynamics, that we are fooled into thinking that it is infinitely large – that there are no boundaries to what we can learn? I’ll have to keep developing this later on.