Experiments

V20: Protocell Agency

Period: 2026-02-18. Substrate: Discrete grid world with evolved GRU agents (~3400 params each).

The Necessity Chain: Membrane → free energy gradient → world model → self-model → affect geometry. Each step necessary, not contingent.

Why leave Lenia: showed the wall is about agency, not signal routing. GRU agents on a discrete grid have bounded local sensory fields (5x5), genuine motor actions (move, consume, emit), and observations shaped by their own actions.

Metric	Seed 42	Seed 123	Seed 7
$\rho_{\text{sync}}$ (max)	0.230	0.232	0.212
$\mathcal{C}_{\text{wm}}$ (final)	0.122	0.126	0.152
$\text{SM}_{\text{sal}}$ (final)	1.22	0.94	1.50
RSA (final)	-0.017	-0.005	0.031

Wall broken (3/3 seeds). $\rho_{\text{sync}} = 0.21$ from cycle 0. The wall is architectural — action-observation loops are present from initialization. 70x Lenia's 0.003. Self-model emergent in 2/3 seeds ( $\text{SM}_{\text{sal}} > 1.0$ ).

Source code

Study record — canonical metadata, result path, status, seeds, and key finding.

— Discrete grid world + GRU agents
— Evolution loop with offspring activation
— Chain test measurements
— GPU runner