Abstract
Purpose: Is the color of temporally asymmetric, chromatic flicker correctly predicted by linear temporal integration of intensity (the Talbot-Plateau law)? Methods: Exp 1. The point of subjective isoluminance (flicker null) for a 25 Hz, red-green (RG), square-wave flicker was determined across multiple values of the duty cycle- expressed as the proportional duration of the green phase of the flicker cycle- from 0.25 to 0.75. Exp 2. For a range of frequencies (30–80 Hz) and duty cycles (0.25, 0.5, 0.75), subjects matched the time-average color of isoluminant (based on Exp 1) RG flicker, by adjusting the RG intensities of a steady field. Results: Exp. 1 for a fixed-intensity red (or green) phase, the longer was the green (or red) phase of the flicker cycle, the higher (up to fourfold) was its intensity at the flicker null. Exp. 2 For a duty cycle of 0.25 or 0.75, the long phase of the flicker cycle dominated the flicker color, but not to the extent predicted by the linear integration. The color-matched steady field required a lower intensity, by as much as 20% at 30 Hz, than the flicker's time-average intensity of the long-phase color. This deviation was measurable out to 70 Hz, beyond the CFF. Conclusions: (1) Dependence of the flicker null on duty cycle can be explained if selective adaptation of the cone type (L or M) most sensitive to the long phase of the flicker cycle calls for a higher intensity during that phase. (2) The color matches suggest an input from an early temporal-contrast nonlinearity. If this nonlinearity demodulates the cone signal, its dc output would escape high-frequency attenuation by more central stages that limit flicker perception. At isoluminance, this dc signal would be similar for L- and M-driven cells, biasing the perceived flicker color away from the Talbot-Plateau prediction, in the direction observed.