Degrees of freedom in evolution
Posted by Tanas Gjorgoski on May 22, 2006
The program itself, as its author says, focuses on examining how the sexual selection (i.e. the criteria of what looks attractive) when put together with natural selection (the ability to stay alive and reproduce) affects the evolution of locomotion in swimbots.
Looking at those swimming bots today, I was thinking what kind of additions would be nice to see in this application. Here are my thoughts…
The natural selection in the program is based on the ability to stay alive and reproduce (which is basically ability to move towards a selected mate). There is no predator/pray relation – in the simulation there are just herbivores (we can imagine dots which give energy as plants). What would be interesting is to see another type of swimbots – carnivores, which would live in same simulation, but which would not eat the dots, but the herbivore swimbots themselves. Having those two types whose evolution would be closely connected might serve as “boost” for the evolution. Of course the number 2 is not magical, so maybe food chain of several levels would give interesting results also. From how I understand that program works, this shouldn’t be hard to accomplish.
The program allows the user to “tweak ecology”. Among the settings there are such like “swimbot hunger threshold” and “swimbot energy % to offspring”. The user can tweak those per whole pool, but it would be interesting to see how the setting affects the evolution. How? By changing those parameters to be from general for all pool to internal parameters for every swimbot (of course they get to mutate, and be transfered through genes).
This also opens interesting question about the degrees of freedom of the evolution in the virtual simulations – the evolution in the simulations will always happen in some abstract space of possible mutations… It is closed space, and there is just given finite number of mutations, the possible animals that can develop are limited in this abstract space. I wonder in this context, if in the nature there is also similar abstract space of possible mutations, which is set once for all, by the mechanism of the mutations in the evolution, or if this mechanism in the nature itself changes… Does the life and evolution in nature transcend this kind of abstract space?