Interpretable mechanisms of decision-making in Lenia-based sensorimotor agents

Develop a human-interpretable mechanistic account of how decision-making arises in the distributed cellular collective that constitutes the Lenia-based sensorimotor agents, specifying how local update rules and morphology produce macro-scale directional choices.

Background

The authors demonstrate sensorimotor behaviors emerging from local interactions in Lenia, where agents change direction and recover from perturbations through morphological computation. While the phenomena are observable, the internal mechanisms remain opaque.

They explicitly acknowledge the current lack of a human-interpretable explanation of how such distributed decision-making is implemented by the learned rules and morphology, indicating a need for mechanistic understanding.