2D Motion Aftereffect — Neuron Adaptation Model

Grating

Drift speed 1.5×

Bar width 10

Adaptation

Adaptation duration 8s

Fatigue rate 1.0×

Neuron pools (live)

adapted-direction pool opposite pool

Phase

READY

Adaptation

Net signal

0.00

Perceived dir.

—

How it works

This is the flat, translating-grating companion to the 3D rotating-spiral aftereffect. It runs an explicit two-pool direction-selective neuron model instead of just freezing a shader — the illusion here is computed, not implied.

Adapting pool — while the grating drifts, the pool tuned to that direction is driven hard and its gain fatigues: da/dt = (input − a) / τ_adapt.
Opposite pool — receives no drive during adaptation, so it stays fresh near baseline.
Test phase — the grating is frozen (genuinely static pixels). Both pools now only fire spontaneously, scaled by their remaining gain (1 − a). The fatigued pool fires less, the fresh pool fires more, so the net signal points opposite to the adapted direction — the opponent-motion theory of the waterfall illusion.
Adaptation and recovery are both live ODE integrations you can watch decay in the firing-rate chart, not a canned animation.

Press Start adapting, then fix your gaze on the centre dot.

2D Motion Aftereffect — Direction-Selective Neuron Adaptation