Four pulses. Zero net input. Two histories.
Each pulse changes only hidden channel allocation. Every pulse has the same energy, preserves the current visible body, and runs for eight steps before the next pulse.
If development only integrated total dose, every zero-net history would converge. If the last pulse dominated, −−++ should end larger. Neither happens.
The first block writes the early future.
Across +36 to +128, ++−− produces +0.054 more radius than −−++, interval [+0.012, +0.093]. The equal-order orthogonal control is -0.002.
Immediately · +36
Specific beyond orthogonal: +0.059 [+0.026, +0.092].
Still present · +48
Native specificity remains positive: +0.052 [+0.019, +0.081].
Finite memory
By +192 the interval crosses zero. The primacy memory is real and temporary.
The body remembers what happened first.
++−− ends with two amplifying pulses. −−++ ends with two canceling pulses. Nevertheless ++−− remains larger. The earlier contiguous block determines the sign of the early response.
Actual +48 frames from harmonic-glider-0307. The order effect is small in any single frame and robust in the paired 64-organism family contrast.
Blocked history
A sustained first context establishes the stronger memory.
Alternating history
Rapid sign alternation largely cancels the order memory across the window.
The causal operation is not commutative. “Cancel, then amplify” is not equivalent to “amplify, then cancel,” even though their signed sums are identical.
The specific response is not the largest raw divergence.
At +36, the two native blocked histories are 0.189 Hellinger units apart. The orthogonal pair is farther apart at 0.200, yet its signed radius order effect is near zero.
Native blocked order
Full-field separation with a direction-specific radius effect.
Orthogonal blocked order
Even larger raw separation, but no matching signed developmental response.
All-cancel versus all-amplify reproduces the hidden control axis.
In this eighth fresh cohort, the four-pulse all-cancel arm is +0.140 radius units above all-amplify across the early window, interval [+0.103, +0.180]. Every one of the sixteen organism families agrees.
The privileged q12 direction is no longer a one-cohort curiosity. It survives a new population and a new temporal design.
An early hidden causal grammar.
Noncommutativity
Equal net hidden inputs produce different futures when their order changes.
Primacy
The first sustained block controls the response more than the final pulses.
Finite memory
The order trace lasts through the early developmental window, then relaxes.
This is exactly the kind of bridge we wanted: information-rich early state is not merely predictive. It interprets causes according to developmental history.
The next bold swing is to locate the boundary of primacy: move the same zero-net sequence earlier and later, and ask when causal order first becomes legible—and when the organism stops caring.