Centering biases in heterochromatic brightness matching

When the method of constant stimuli is used to measure heterochromatic brightness matches, the resulting matches can be strongly biased toward the center of the range of test luminances used [Teller, D. Y., Pereverzeva, M., & Civan, A. L. (2003). Adult brightness vs. luminance as models of infant photometry: variability, biasability, and spectral characteristics for two age groups favor the luminance model. Journal of Vision, 3, 333–346]. In the present paper, we investigate the source of this centering bias. The stimuli were 2° red squares presented in a gray surround. In the main experiments, two ranges of stimulus luminance were presented in separate physical locations on a video monitor, but with test trials interleaved in time. Subjects either fixated a fixation cross (fixation condition), creating different retinotopic locations for the two luminance ranges, or foveated each stimulus as it appeared (foveation condition), creating identical retinotopic locations for both ranges. In the fixation condition, the two different stimulus sets resulted in a simultaneous centering bias—two different brightness matches at two different retinotopic locations at the same time. This effect was essentially eliminated in the foveation condition. A dichoptic foveation condition also revealed no centering bias. The results suggest that under the conditions tested, the centering bias is caused by a process located at a post-retinal but still retinotopically organized level of the visual system, rather than by either a retinal process or a more central, spatiotopically organized one.