Purpose. Spatial antagonism among cones has long been observed, and is believed to be caused by negative feedback from horizontal cells (Baylor et al , 1967). There is ample evidence that horizontal cells connect laterally through gap junctions and are decoupled by light. We explored the functional effects of horizontal cell coupling by measuring the spatial antagonism among cones at varying luminances. Methods. The method depends on a psychophysical model (Chen et al, Vision Res., 1993) that uses a nonlinearity in the retina to partition the visual system into two linear spatial filters. The spatial antagonism among cones is represented by a surround mechanism in the first filter, which causes a sensitivity loss at low spatial frequencies. The ratio of the contrast sensitivity function of the entire visual system (CSF) and the second filter yields the shape of the surround of the first filter, as long as the total loss of sensitivity is equal to the cascaded attenuation by the two filters. Two high frequency gratings of slightly different orientations were formed by optical interference and superimposed on the observer's retina so that the retinal nonlinearity produced a distortion product visible to the observer. The transfer function of the first filter was measured by fixing the frequency of the two primary gratings while varying their distortion product; the second filter was measured by the obverse process. The CSF of the entire system was measured with a grating of the same orientation as the distortion product. The effects of eye movements were avoided by 2 msec presentations on a steady background of 0, 10, or 100 td. Results. At all background intensities, the CSF of the system was equal to the product of the transfer functions of the two filters. Increasing the luminance decreased the spread of the surround mechanism of the first filter and increased its amplitude. Conclusions. Light diminishes the spatial extent of antagonism in the first filter. This we attribute to decoupling of horizontal cells.