Portfolio · Biological Architectures
Phototaxis Without Rules
Light on one eye. No `if left > right` anywhere in the loop. The bilateral differential flows through the ganglion chain as a lateral bias; the body turns toward the light because the geometry of the differential is the decision.
Phototaxis Without Rules
A lateral differential in the field produces a lateral turn in the body — no branching, no "if light is on the left" anywhere in the code.
The Question
How does an animal turn toward a stimulus without ever asking which side the stimulus is on?
The Answer, Short Version
The field carries the differential. One eye sees more light, that side's ganglia run stronger, that side's legs push harder, the body pivots. The question "which side?" is never asked because it is not a question — it is a gradient through a shared medium. The turn is produced by the geometry of the asymmetry itself.
The Setup
The tardigrade has two eyes, one each side, each feeding one side of the ganglion chain. The eyes are the only lateral input in the whole system. A shared field sits beneath the ganglia and carries drift across the whole chain. The leg activation on any side of the body is the resolution of whatever signal reaches that side plus whatever the field has carried from elsewhere.
There is no if light_left > light_right line anywhere. No sensor
comparison, no decision branch, no side-selection code. The only
lateralisation in the system is structural: the left eye feeds the
left ganglion path, the right eye feeds the right ganglion path, and
the field connects both.
The test: apply five lighting scenarios and log what the legs do.
What Emerged
| Scenario | Left legs | Right legs | Body response |
|---|---|---|---|
| Light LEFT only | 1.08 | 0.56 | pivot, head to light |
| Light RIGHT only | 0.56 | 1.08 | pivot, head to light |
| Equal light | 0.83 | 0.83 | STRAIGHT |
| Strong LEFT | 1.03 | 0.62 | pivot, head to light |
| Strong RIGHT | 0.62 | 1.03 | pivot, head to light |
The sign of the asymmetry matches the sign of the stimulus in every case. Equal light produces equal legs — the symmetry is preserved, not approximated. Strong and weak stimuli on the same side produce the same direction but different magnitudes, which is what you want from a gradient-follower: bearing is set by the sign of the differential, urgency by its strength.
Positive phototaxis — attraction to light — emerges cleanly from this geometry.
The Detail Worth Pointing At
Look at the brain output for left-light versus right-light, from the same run:
Light on LEFT — Brain: 0.2564 Ganglia: 0.290, 0.286, 0.329, 0.261
Light on RIGHT — Brain: 0.2564 Ganglia: 0.290, 0.286, 0.329, 0.261
The brain is identical. The ganglia are identical. Only the leg output differs.
This means the brain is processing "something bright is here" symmetrically — it does not encode which side. The laterality lives in the ganglion-to-leg mapping, where left ganglia feed left legs and right feed right. The differential arrives at the output layer as structural asymmetry and becomes an asymmetric push.
This is a clean architectural split between what (brain) and where (ganglion-leg coupling). Neither layer has any if/else. Both just resolve their local geometry.
What This Proves
A sensorimotor behaviour that standard reinforcement learning would treat as a policy to be learned — when stimulus left, turn left; when stimulus right, turn right — is produced here as a structural consequence of a body with two eyes, a ganglion chain, and a shared field. No rule. No training. No reward signal. The policy is a side effect of the geometry.
The same mechanism that produces the gait — field excitation resolving to coordinated output — produces the turn. Asymmetric excitation resolves to asymmetric output. One primitive; two behaviours.
What This Does Not Prove
This is not a model of how real tardigrade phototaxis works. Real animals have adaptation, context dependence, and a range of responses depending on light wavelength, intensity, and prior state. None of that is here.
Nor is the attraction direction architectural. Positive phototaxis is one sign convention; negative phototaxis — retreat from light — is the opposite sign and an equally natural resolution of the same geometry. The architectural claim is about how the turn is produced, not about why this particular version is attractive rather than aversive. That is a parameter choice, not a theorem.
Raychell Langan · NEXICOG Ltd · Hampshire, UK