Change arrangement. Change nothing else.
For each organism we captured the exact two-channel q8 state. Then we moved channel-coupled blocks inside the same central 32×32 window.
Same cell vectors
Same channel masses
Same world
At least 98.61% of every channel’s q8 mass lay inside that window; state outside it stayed byte-identical. The three shuffles were also comparably far from intact in raw Hellinger distance (0.377, 0.384, 0.389). What changed downstream therefore tracks preserved spatial scale—not simply a larger perturbation.
Representative source s11, rendered from the exact q8 state patches with one shared intensity scale.
Organization made nearby actions more distinct—and less self-contradictory.
Matched-history separation at release
Larger means the two score-matched histories arrive at more distinct local states.
History-tangent agreement
1 = aligned, 0 = unrelated, −1 = opposite. Cell shuffling makes the action surface nearly reverse itself.
The 8×8 and 4×4 controls had matching quality as good as intact, yet both already lost local separation and tangent coherence. The cell-level condition was harder to score-match, so it is the strongest ablation, not the only support for the result.
Destroying q8 organization delays the visible transition.
Untouched transition time
Every organism still crossed the frozen morphology detector. The transition simply moved later.
Action-route timing range
Within each state: latest minus earliest certified transition across the four matched action histories.
| Spatial state | Mean untouched transition | Mean route range | Mean action shift from own clock |
|---|---|---|---|
| ● Intact | 50.1 | 7.8 | 4.5 |
| ● 8×8 tiles | 62.9 | 12.8 | 7.3 |
| ● 4×4 tiles | 72.6 | 17.4 | 11.4 |
| ● Cells | 80.6 | 19.7 | 10.4 |
Relative to intact, the mean untouched delay was +12.85 [+3.67, +24.03] passages for 8×8 tiles, +22.51 [+10.44, +35.95] for 4×4 tiles, and +30.59 [+15.15, +48.46] for cells. All 39 untouched states still transitioned.
The same representative source s11 at passage 128 under untouched continuation from each q8 state.
Disorder opened more futures. It did not create a better map.
q900 matched-history distance
Distant outcomes diverge more after any spatial shuffle. Raw possibility expands as organization is removed.
q900 tangent agreement
Most local geometric ordering washes out by q900; cell-level destruction leaves the clearest residual loss.
Representative s11, high-hold continuation at q900. These are examples, not the source of the quantitative result.
Causal plasticity comes first. Organization canalizes it.
More organized q8 state
- stronger local separation between matched histories
- less reversed / twisted action geometry
- earlier visible transition
- less action-dependent timing variation
- narrower distant future repertoire
More disordered q8 state
- weaker local distinctions
- nearly opposing local tangents
- delayed organization
- more timing susceptibility
- broader q900 divergence
This is a causal result about arrangement. The same material behaves differently when its spatial relations are destroyed. Early organization acts like a developmental constraint: it turns an already-rich possibility space into a narrower, more coherent trajectory toward form.
That is directly relevant to causal emergence. It suggests that the interesting precursor may not be a scalar “more information” peak. It may be the onset of canalization: the moment distributed structure begins to make interventions have stable, coordinated meanings for the future of the whole.
Can we rebuild the constraint?
The strongest next intervention is a rescue curve: begin from the cell-shuffled q8 state, restore increasing fractions of the original coarse spatial arrangement, and ask whether tangent coherence and developmental timing snap back before visible morphology does.
If a partial spatial repair restores the action geometry first, we have a manipulable order parameter for developmental commitment—not merely an ablation effect. That is where we go next.