Symmetry

The past few weeks I’ve been looking into why creatures in the game exhibit less symmetry than real creatures do in nature. As a bit of advance warning, this is something I am still working on, so unfortunately this blog is going to end not with an answer to the question but instead with a “To be continued…”

One piece of the solution seems to be that symmetries are easier to evolve in nature than in-game. In real life, genes tell a cell what to do in part based on their chemical environment; symmetries in the newborn creature arise from symmetries in that environment itself. The simulation’s interpretation of DNA is much more abstract and doesn’t account for this, so I adjusted the mutation code to add a special branch for matching pairs of limbs.

One interesting thing is that mirroring the limbs, neurons, and torques resulted in some off-balance creatures. This is a little mysterious to me. It shouldn’t happen, in theory, but I suspect something in the internals of the physics engine is causing a problem. To address it, I adjusted the nerves so that, instead of just mirroring impulses traveling down the neurons, one limb specifically tries to duplicate the position and orientation of the other.

However, when I evolved creatures with swimming speed as their selection criteria, only 10-20% of them actually exhibited symmetry. This isn’t quite what I was expecting, but I think it does make sense. If a limb or entire subsection of the body is generating forward thrust, you would think duplicating it would help. But creatures always exist in a tension between ‘good’ drag, which they can use to propel themselves forward, and ‘bad’ drag, which slows their motion (adding another limb also increases the mass, which can further inhibit motion). For many sea-creatures, the main source of thrust is the tail-fin, while the symmetric fins that run along its sides are more for stability and steering. So I think this question is bound up with target-seeking creatures. That’s an area I’ve been meaning to focus on for a while, as the target seekers take longer to evolve and move much slower than their cousins who are just swimming as fast as possible. This will be a task for next month, but hopefully I will be able to kill two birds with one stone.

Thanks for reading!
Tom