A ganglion cell does not report how much light falls on it. It reports how the light in one small region compares with the light around it. Everything below is recorded the way a physiologist would: flash a stimulus, count the spikes, repeat.
Each stimulus is flashed for half a second, over and over. One row per flash; the shaded band is when the light is on. Rates are counted from the spikes shown.
Rates are spikes per second, counted over every flash so far. With no stimulus both cells fire about 20 per second. “After” is the 300 ms after the light goes off.
Click anywhere to flash the spot there. The electrode stays in one cell throughout; only the light moves.
Each mark is scored from the spikes one or more flashes produced at that spot.
Conditions are shuffled, as in a real experiment, and sorted afterwards.
The top row is the flash in progress. Below, the latest four flashes of each kind.
Numbers at right: spikes per second while the light is on, over every flash of that kind so far.
Two rows of cells tile this image, one ON-center and one OFF-center, each comparing its own center with its own surround.
Above, the luminance across the image. Below, each dot is one cell's spike count divided by the time counted; lines are the rates many repetitions would converge to.
The receptive field is drawn to scale against the bars.
Each row is one full cycle of the grating passing the receptive field, aligned to the light at its center.
Look at one intersection while attending to the others.
Expected sustained response of a whole sheet of these cells, warm above the maintained rate and cool below. This panel is a model prediction, not a recording.