Neural symmetry

Based on the experiments demonstrated on 23rd, I want to now ask precisely how the underlying CPG neural architecture and body plan morphology emerge with and without the hardcoded turning mechanism. It is expected that even when the turning mechanism is hardcoded, the underlying CPG circuit will become tuned as a matter of facilitation. First let us look at the morphology of the agent:

animatSchematic
As shown, the agent features 15 body segments. The smaller circles indicate actuator locations (motor neurons). Now what we can do is evolve (i) the length of each body segment and (ii), the architecture of the underlying nervous system. The weight values can then be derived from the euclidean distance between neuron pairs so that we can directly couple computational properties to spatial characteristics.

Given the circuit pointed to in my last post, we can intuitively imagine a symmetry in weight values. And since weight values are tied to Euclidean information, we can then expect an emergence of symmetry within the neural architecture. Indeed this looks promising given the following architecture to have emerged in a recent experiment (very tentative).
exampleSymmetry
As we can see from the partitioning made by the central dashed line, with the exception of the very last part of the neural circuit, a rudimentary level of bilateral symmetry has emerged within the neural circuit.