A causal dissection of the hidden developmental field

The code is in
the relationships.

The hidden pattern can be moved or rotated and still turn the future. Scramble its internal spatial phase and the direction disappears. Development reads organization—not coordinates, bulk amount, or perturbation strength alone.

2,496new causal futures
16scale, phase, pose, and region operators
0.0visible starting-field change
3.63×10⁻⁹maximum global-balance residual
01 / The finding

Rigid motion preserves meaning. Phase destruction does not.

-0.50+0.00+0.25+0.50intact patternintact pattern: +0.369 [+0.254, +0.490]translated mapstranslated maps: +0.316 [+0.208, +0.412]rotated mapsrotated maps: +0.336 [+0.233, +0.463]phase-scrambledphase-scrambled: -0.113 [-0.570, +0.335]fine scales onlyfine scales only: +0.317 [+0.122, +0.488]coarse σ16coarse σ16: +0.249 [+0.141, +0.346]

RIGIDLY MOVED PATTERNS

+0.323

Pooled q128 movement toward the paired donor future · 95% CI [+0.262, +0.383].

PHASE-SCRAMBLED PATTERN

-0.113

Same global balance and Fourier magnitude, but randomized spatial relationships · [-0.570, +0.335].

Rigid maps beat phase-scrambled maps by +0.436 at q128 [-0.008, +0.906]. The interval nearly clears zero; rotations alone do clear it.

The rigid and phase-scrambled interventions are comparably far from the intact host pattern and comparably energetic. The difference is relational: translation and rotation preserve who-neighbors-whom; phase scrambling destroys it.

02 / What survives

The code is tolerant to pose and spread across scales.

same topology
translated / rotated
→
donor-directed future
+0.323 pooled
≠
same spectrum, lost phase
-0.113 pooled
-0.2+0.0+0.2+0.4intact q32: +0.158intact q64: +0.107intact q96: +0.226intact q128: +0.369rigid q32: +0.232rigid q64: +0.255rigid q96: +0.276rigid q128: +0.323phase q32: -0.059phase q64: +0.245phase q96: +0.256phase q128: -0.113q32q64q96q128intactrigidphase

Fine-only and strongly smoothed patterns both retain positive donor-directed motion. There is no single magic pixel scale. The developmental instruction is multiscale.

FINE ONLY

+0.317

[+0.122, +0.488]

SMOOTH σ2

+0.393

[+0.229, +0.547]

SMOOTH σ16

+0.249

[+0.141, +0.346]

03 / Where it lives

No single patch owns the instruction.

CORERIMcore half+0.25rim half+0.22tip half+0.21flank half+0.19intact whole-map effect +0.37

Each half-map carries roughly half to two-thirds of the intact effect. Complementary halves approximately reconstruct the whole.

RADIAL PARTITION

Core: +0.249. Rim: +0.220. Each mask contains half of the donor–host hidden-pattern energy before rebalancing.

AXIAL PARTITION

Tips: +0.209. Flanks: +0.194. The effect is distributed across complementary geometries.

This looks less like one privileged organizer pixel and more like a redundant field code: several large subregions can each push the whole organism along the same developmental direction.

04 / Two developmental grammars

Compactification is distributed. Expansion is more centralized.

EXPANSION

core +0.201

[+0.016, +0.379]. The other half-maps remain positive on average but weaker and less stable.

COMPACTIFICATION

all four positive

Core +0.297; rim +0.339; tips +0.269; flanks +0.285. Every interval is above zero.

The same body does not use one universal morphogenetic grammar. Expansion and compactification distribute causal control differently across the hidden field.

05 / The dynamical surprise

Wrongly arranged codes do not merely fail. They kick the trajectory sideways.

Phase-scrambled and shifted maps create large transient excursions—especially around q96—before settling. The hidden field is not a static label. It is an instruction injected into a nonlinear developmental flow.

That suggests a richer object than a scalar precursor: a spatial control vector whose topology, pose, amplitude, and timing jointly determine where the organism moves next.

06 / The next direct swing

Match intervention energy exactly. Then map the syntax.

EXACT DELTA-ENERGY MATCH

The spectral and rigid operators were matched in global hidden-contrast RMS, but they are larger than the intact donor–host delta. The next run will match that delta exactly so topology and strength are completely separated.

TOPOLOGY DOSE

Interpolate from intact phase to scrambled phase, and from correct pose to displaced pose. This will expose thresholds, reversals, and whether the code has an error-correcting basin.

We are no longer asking whether an early hidden state predicts an organism. We are beginning to decode the causal language that tells the organism what to become.