When does a developing system begin to become itself?
Flow Lenia begins here as a seeded field that is still reorganizing; over hundreds of updates, some fields settle into persistent, organic-looking bodies, while others fragment or take another route through shape space. We disturbed that process at different moments and measured how the available futures changed, because the first signs of an organism may appear in what a state can still become before they are obvious in its shape.
Each Flow Lenia run begins as a seeded, continuous multichannel field that is still reorganizing, and although our first experiments asked whether a disturbed run would recover the organization and trajectory of its matched undisturbed counterpart, the clearer pattern was already visible in the different futures opened by nearby interventions.
Synthesisfeature
We were asking
When does a seeded field begin to behave as an individual whose history changes both its reachable futures and the meaning of an intervention?
This experiment found
In the prospective cohort, nearby futures reorganized before a persistent body was obvious; across separate experiments, responses became more history-specific and the system grew harder to rewrite, even though its descendants did not converge on one final shape.
That sent us to
Can the early response address predict how a completely new genotype will react, and can an intervention move that address rather than merely moving the body?
Follow the experiments
The experiments that opened the next questions
These are the strongest places to enter after the synthesis, kept in editorial order so that each answer leads naturally into the question that followed it.
A whole-over-parts predictive measurement rose before an independently defined seed transition, so the field became more informative as a coordinated whole before it sustained a visibly non-square body.
We were asking
Does coordination across the whole field change before a seeded Flow Lenia run becomes visibly organism-like?
This experiment found
In a prospective cohort, the pre-transition rise was positive in fifteen of sixteen family means, which makes this a reproducible temporal signal rather than a single attractive specimen.
That sent us to
Does the same early coordinate forecast how tightly the state constrains later transitions, and can a direct intervention move that constraint?
At q8, when the seeded fields still look modest and similar, different causal histories already fan outward into distinct bodies and formation times, and the measured future map is at least as broad as it is at q32.
We were asking
Is the map from history to future assembled gradually as the body forms, or is it already present in the earliest state we measured?
This experiment found
The map was already broad at q8 and consolidated slightly through q24, while visible geometry changed later; causal multiplicity therefore precedes the body that eventually displays it.
That sent us to
What does spatial organization add if the early material already has a large action world?
Exact block rearrangements preserved genotype, channel masses, occupied material, and almost the entire outside field, while changing how clearly nearby actions separated and how explosively distant futures spread.
We were asking
If causal plasticity is already present at q8, what does the field's spatial organization contribute to development?
This experiment found
Organization made local actions more legible and development faster and more reliable, while scrambling opened a broader but less disciplined collection of later futures.
That sent us to
Can the lost constraint be rebuilt continuously, and does development reach an age at which the original repair no longer belongs to the state?
The same eight inventory-preserving pushes produced broad transition-time futures in some pre-form states and tightly canalized timing futures in others, even though their microscopic field outcomes remained diverse.
We were asking
Before a visible transition, has the state already committed to when that transition will occur?
This experiment found
Higher whole-over-parts organization forecast a narrower transition-time repertoire across fresh states, while the full microscopic future fan did not collapse into one form.
That sent us to
Can an early push move a sibling state toward that committed regime and reduce the futures still available to it?
A controller that tested five pulse sizes at each passage usually chose one strong early intervention and then restraint, and the resulting difference between alternative causal histories persisted for eight passages after control ended.
We were asking
Can feedback hold a developing system in a state where more causal histories remain distinguishable, rather than merely disturbing it repeatedly?
This experiment found
Early state-aware control kept the causal gate open, while blindly applying the full pulse at every passage drove the comparison in the other direction.
That sent us to
Does the open state give the same later actions more genuinely different futures, and what bodies do those futures eventually produce?
After an early action separated open and closed causal states, the action's special advantage faded but the two branches continued to diverge into bodies with different size, rigidity, and topological activity.
We were asking
Does an early difference in causal openness remain an abstract score, or does it become part of the later body?
This experiment found
The hidden split outlived the selected intervention and progressively appeared in morphology, while a least-divergent action-pair control stayed much closer.
That sent us to
Which later actions reveal the extra futures of the open state, and can feedback reverse the body's resulting bias?
As commitment arrived, later founder interventions acquired more stable spatial consequences across four different prior-action histories, so recent history had less power to rewrite the relationships among future outcomes.
We were asking
Does development merely reduce sensitivity, or does it stabilize what a later action means?
This experiment found
After the commitment turn, spatial outcome geometry became more consistent across prior histories at the population level, even though the effect was not universal in every organism.
That sent us to
Does that stabilization follow the body's developmental age, the transplanted hidden state, or a distributed interaction between both?
Four plans were frozen before their outcomes: the fine-shape confirmation did not reproduce its earlier estimate, the same push lost leverage with age, sibling futures widened rather than converged, and the population of close look-alikes collapsed.
We were asking
When a developing Flow Lenia state becomes harder to redirect, are its possible forms also converging toward one narrow destination?
This experiment found
No; taken together, the four experiments show commitment as increasing rigidity and individuation, because histories become easier to tell apart while the fan of sibling outcomes keeps opening.
That sent us to
Will the same separation between rigidity, individuation, and basin width survive in fresh genotypes and under interventions chosen by a different observer?
A faint preferred direction was already measurable at q8, and by q12 its control over later body geometry had nearly doubled even though the applied hidden-state dose was smaller.
We were asking
How early does a seeded field acquire a privileged direction in which an invisible composition change alters its later body?
This experiment found
Directional leverage increased sharply between q8 and q12, before development settled into the later body, while sideways controls remained weaker.
That sent us to
What information is being reorganized while ordinary predictability falls and directional control rises?
At q40, every visible pixel and the global channel balance were held fixed while only the hidden allocation between two material channels was rewired, and the later body changed.
We were asking
Can two states with the same visible field have different developmental futures because their hidden composition is arranged differently?
This experiment found
Yes; an exact invisible rewrite shifted later radius and trajectory, showing that visible morphology is an incomplete causal description of the state.
That sent us to
Can each organism's own hidden direction predict which invisible rewrite will expand or compact its later body?
Eight equal first actions created sibling states from one q32 ancestor, after which the most- and least-committed siblings received the same complete second action repertoire.
We were asking
Can one early intervention move a state toward commitment and change how many microscopic futures remain accessible?
This experiment found
Across thirty ancestors, the high-coordinate sibling had a narrower later causal repertoire, with seven of eight organism groups pointing in the same direction.
That sent us to
Does repeating the same action strengthen commitment, or does its direction rotate as the organism changes?
