The whole model
Uses the four spatial mass coordinates together. It can represent cross-region dependence: one lobe’s movement informing another’s.
A+B+C+D → next01
The whole moved first
The morphology rule knew nothing about information theory. The information model knew nothing about the future event. Yet in fresh specimens the integrated predictive advantage climbed during the final 16 transitions before the certified seed break.
Visual example, fresh specimen p01. The detector is deliberately conservative: it certifies eight sustained frames below the frozen rectangularity boundary. It marks commitment to a coherent non-square body, not the first faint internal texture.
02
One simple comparison
At each pre-event step, we ask how surprising the next four-quadrant mass pattern is under one joint model versus the best split into independent spatial parts.
The parts model
Chooses the most favorable bipartition, then treats the two groups independently. Whatever it predicts separately is subtracted away.
(A+B) + (C+D)Whole − parts · frozen primary+0.364 · 95% [+0.231, +0.502]
Φ-R MMI · direct companion+0.171 · 95% [+0.084, +0.255]
Whole TDMI · ordinary predictability-0.044 · 95% [-0.609, +0.449]
Whole − parts minus TDMI+0.408 · 95% [-0.099, +0.996]
Scale: −0.7 to +1.1 local predictive-information units. The primary and direct Φ-R companion both replicate upward. Ordinary whole TDMI has no consistent direction; the formal primary-minus-TDMI difference is positive on average but noisy.
03
Not one lucky creature
The effect is distributed across the fresh cohort. Fifteen of sixteen four-specimen family means point upward.
15/16positive family means · descriptive breadth, no family was selected by outcome
f17
+0.406
f18
-0.170
f19
+0.832
f20
+0.471
f21
+0.446
f22
+0.056
f23
+0.625
f24
+0.192
f25
+0.449
f26
+0.480
f27
+0.047
f28
+0.199
f29
+0.673
f30
+0.361
f31
+0.218
f32
+0.289
04
What changed in the story
We now have two prospective clues at different levels: a structural peak before future-repertoire reorganization, and an integrated predictive rise before a visible organism-like transition.
Development
Four-fold method development found a +0.507 whole-over-parts rise in 44/59 fully measurable specimens.
Fresh replication
The next 64 source specimens were frozen before replay. Among 54 with complete 32-step event windows, the mean rise was +0.364, with the whole interval above zero.
Biological breadth
Sixty-three of 64 formed the certified morphology within 900 steps. One did not and remains explicitly right-censored—never assigned an imaginary event.
The new working claimA whole-over-parts predictive configuration precedes a reproducible Flow Lenia morphogenesis event.
05
Now make it causal
The observational precursor is earned. The clean next move is to intervene before the morphology boundary rather than return to late ecological restoration.
Earlier onset
Freeze a morphology-only detector for the first coherent departure from the noisy square, then test it on the remaining untouched families with shorter pre-event windows.
Advance or suppress
At matched early states, choose small perturbations predicted to raise or lower whole-minus-parts while holding mass and raw motion close. Ask whether the seed break moves in time.
Replay the action
Replay the exact chosen perturbations open-loop. If feedback-specific steering moves both the signal and birth time, the controller and organism are making the transition together.
Scope, without deflating the find: this is a prospective precursor to a certified morphogenesis event. It does not yet establish that the metric causes the transition, and the event is organism formation—not autonomous replication.