Abstract
Previous experiments have suggested that afterimages originate from an inverse multiplicative sensitivity adjustment in the long (L), medium (M), and short (S) wavelength cones of the retina. Our experiments with traditional CRT monitors indicate significant shortfalls from reciprocity in the regulation of sensitivity, but current experiments employ a high intensity Maxwellian View projector (BIGMAX) to provide more complete adaptation and reciprocity in afterimage saturation. We propose a color afterimage model according to which the afterimage can be matched by inverting the log excitation profile of the inducing stimulus for the L, M, and S cones, and we provide data supporting this model over an alternative additive model. Methods: Stimuli were displayed on a BIGMAX display capable of producing image intensities up to 7000 cd/m^2. In the initial experiment, stimulus consisted of a central fixation point on an equal energy grey background while either an equiluminant color- or intensity-defined blob was displayed 4° to the right of fixation. Upon adaptation, subjects matched the color and intensity of a response blob to the afterimage seen after a momentary return to background (200 milliseconds) at the location of the adapted blob. The second experiment compared the multiplicative adaptation model with an alternative additive model. Subjects adapted to and matched luminance blobs as in the first experiment, but instead of removing the adapted blob, the intensity of the entire image was briefly (200 milliseconds) increased in either a multiplicative or additive fashion. Results fell close to reciprocity predictions in the LMS color space for the luminance-defined and most color-defined blobs. In the second experiment, no afterimage was seen when the stimulus was multiplicatively changed, while additive changes did result in obvious afterimages. Predictions based on the multiplicative model accounted for the additively formed afterimages with only minor deviations.