Future-blind Flow Lenia experiment · 24 August 2026
The organism learns what actions mean.
As developmental commitment arrives, later founder interventions acquire more stable spatial meanings. Their effects depend less on what happened immediately before.
This is not just a creature becoming harder to perturb. It is a reorganization of its causal action space: recent history has less power to rewrite the relationships among possible futures.
01 / The question
Can an intervention acquire a meaning?
The same later action can lead somewhere different because the organism was changed by an earlier action. We asked whether that context dependence changes when development commits to a form.
A
First controlled action
Choose one of four founders. This establishes the immediate causal history.
·
Living background
One ordinary matched founder-supply event occurs between controlled actions.
B
Second controlled action
Choose one of the same four founders. This is the action whose meaning we map.
↻
Commitment timing
Repeat A→B across the developmental turn: before→turn and turn→after.
12
Future passages
Continue the ecology and read the spatial and founder-composition futures.
Every ordered pair, not selected examples
B0B1B2B3
A0
0→0
0→1
0→2
0→3
A11→0
1→1
1→2
1→3
A22→0
2→1
2→2
2→3
A33→0
3→1
3→2
3→3
Within each first-action row, measure all six distances among the four second-action futures.
If the six-edge map changes across rows, recent history has rewritten what the second actions mean.
Ask whether that rewrite is larger before the turn or after the turn, using the same families and matched stochastic backgrounds.
02 / The result
Commitment stabilizes the causal grammar.
The earlier action rewrote the geometry of later outcomes more strongly before commitment. After commitment, later actions behaved more like stable operations on the organism.
−11.1%
context-dependent rewrite after commitment
Before → turn
0.02636Turn → after
0.02343Resolved direction. Family bootstrap, 20,000 draws: before-minus-after =
+0.002935, 95% interval [+0.000493, +0.005409].03 / Dynamics
The difference grows into the future.
The two timing windows are nearly identical at the intervention itself. Their causal grammars separate over subsequent passages, as the interventions propagate through the living pattern.
Before → turnTurn → after
The gap is not an instantaneous scoring artifact. It expands as the two interventions are metabolized by the evolving ecology, reaching its maximum around passages 10–11 and remaining present at passage 12.
04 / Decomposition
Spatial action semantics canalize first.
The resolved grammar effect comes with a resolved narrowing of the second-action spatial repertoire. Founder composition points the same way on average, but remains much more heterogeneous.
Causal-geometry rewrite
How much recent history changes the six relationships among the four possible later-action futures.
Δ +0.00294 · 95% CI +0.00049 to +0.00541
Second-action spatial breadth
The four later actions still matter, but after commitment they span a narrower field-level repertoire.
Δ +0.00421 · 95% CI +0.00031 to +0.00837
First-action spatial memory
The direct spatial memory of the earlier action also shrinks on average, but its interval still touches zero.
Δ +0.00446 · 95% CI −0.00043 to +0.00928
Founder-composition grammar
Founder proportions show the same mean direction, but not a resolved population-wide stabilization.
Δ +0.00248 · 95% CI −0.00194 to +0.00702
05 / Families
Not universal. Still population-level.
Twelve of nineteen organism families stabilized; seven moved the other way. The resolved mean is not unanimity—it is a reproducible developmental tendency with real biological heterogeneity.
06 / Meaning
Developmental commitment is causal canalization.
A committed Flow Lenia organism is not simply a visually stable state. It is a state in which possible interventions have begun to acquire more reliable operational meanings.
01
Before commitment
The organism is causally plastic. What a later founder action does depends more strongly on which action immediately preceded it.
02
Across the turn
The spatial repertoire narrows and the relational map among action outcomes becomes less rewritable by recent history.
03
After commitment
Actions behave more like stable operations on an organized whole—not perfectly, and not identically across every family, but detectably.
Organism formation may be the process by which a living field develops a stable causal vocabulary.
The result connects developmental commitment to causal emergence in a concrete intervention space: the organism-level state increasingly constrains how local founder inputs propagate. The spatial grammar resolves before the compositional grammar does, suggesting that coherent form can canalize action semantics without freezing every internal degree of freedom.
07 / Next swings
Now turn the result into a developmental instrument.
The next experiments should locate, move, and transfer the stabilization—not merely repeat the same comparison.
08 / Source boundary
What was actually established.
Directly supported
Across a frozen 19-family panel, spatial action-outcome geometry was less context-rewritable after the previously identified commitment turn. Second-action spatial breadth narrowed in parallel. The outcome was not used to select families or timing.
Still open
This does not by itself establish autonomous replication or a formal Hoel-style causal-emergence quantity. It establishes a direct causal signature of developmental organization and gives us a sharper object to test next.
Living ecology
The two controlled actions were embedded in continuing matched founder supply. The experiment therefore tested intervention meaning inside an active ecology, not an isolated two-move toy.
Analysis
Primary outcome: mean passages 5–12. Spatial field distance: one minus anchored similarity. Causal-grammar rewrite: pairwise RMS difference among four six-edge second-action geometries. Uncertainty: family bootstrap, 20,000 draws.