The units are homeostatic spiking neurons after Falandays et al. (2021) — the update equations and so forth can be found here.
One object of study — a class of entity: self-organising collectives of minimal units, the same thing whether the units are nodes in a reservoir or agents in a swarm. It self-organises toward its own viability, not toward any task of ours; that it can be made to do a task is what we probe, from two sides.
What it can do
Cellular cognition. Design and evolve node models under a strict locality constraint, then read a capacity for behavioural complexity — responsiveness, coordination, memory — off a calibrated score. Each task is a test of a capacity, not a benchmark to win.
What makes it possible
Collective intelligence. Read the collective as a network and a multi-agent system at once — profile and sweep the dynamical qualities (criticality, order, information flow) that underwrite a capacity.
A product of the Diverse Intelligences Summer Institute 2026, Geneva NY.
