Visual adaptation is highly localized in the cat's retina.

1. The aim of this study was to determine how the spatial pattern of steady light in a visual stimulus affects the state of adaptation of the retina. 2. Impulse rate was recorded from single X and Y ganglion cells in the cat's retina. The luminance of a narrow bar of light centred over the receptive field was modulated sinusoidally in time about a steady background, and a cell's contrast gain was measured as the ratio of impulse rate modulation to bar contrast. 3. The contrast gain of a cell was set by the background, a fixed luminance level about which luminance varied in the form of a grating; grating luminance varied sinusoidally with distance but did not vary in time. When the spatial frequency of the grating was low, contrast gain was increased by a grating with a trough centred over the receptive field, and decreased by a peak-centred grating. 4. As the spatial frequency of the grating increased, its effect on contrast gain disappeared. For cells around 10 deg from the central area, this change occurred at spatial frequencies close to 1 cycle deg-1. 5. For each cell the effect on contrast gain of the background's spatial frequency was compared with the spatial frequency response to a time-varying grating. It was found that the summation area for adapting light in both X and Y cells is very close in size to an X cell centre mechanism, and that the summation area for adapting light in Y cells is therefore considerably smaller than a Y cell centre. 6. From this and other evidence it was shown that sub-areas of the Y cell centre mechanism can be independently adapted. 7. A background grating with a trough centred over the receptive field raised contrast gain more at mid-range spatial frequencies than at low frequencies, producing a hump in the contrast gain versus frequency curve. A peak-centred grating reduced contrast gain more at mid-range frequencies than at low, producing a dip. 8. The dip in the contrast gain versus frequency curve for a peak-centred grating was always greater than the hump for a trough-centred grating. 9. These humps and dips were interpreted in terms of a model containing two antagonistic pathways. One pathway had a smaller summation area for adapting light than the other.(ABSTRACT TRUNCATED AT 400 WORDS)