An fMRI study of chromatic processing in humans

Abstract

Visual performance in the detection of luminance patterns is generally well matched to the behavior of retinal ganglion cells across the retina, after stimuli are appropriately scaled with retinal eccentricity for the size of the cortical projection (M-scaling). However, this is not the case for chromatic vision. Chromatic sensitivity of a human observer is high in foveal vision, and – especially to red-green modulation – deteriorates towards the periphery of the visual field even after M-scaling. However, midget retinal ganglion cells, which are responsible for the detection of red-green change, respond equally well to chromatic modulation in the peripheral visual field as in the fovea. It has been postulated that central mechanisms are involved in the psychophysical sensitivity loss for color discrimination in the periphery. Further, psychophysical sensitivity to chromatic modulation has been observed to decrease at high temporal frequencies (12 Hz), whereas the relevant ganglion cells (midget ganglion cells for red-green modulation, small bistratified cells for blue-yellow modulation) show robust responses to high temporal frequency chromatic modulation. It has therefore been posited that low-pass filtering of chromatic information occurs at neural loci across the visual cortex. In this thesis, I conducted three functional magnetic resonance imaging (fMRI) experiments on human subjects to investigate the cortical representation of peripheral as well as high temporal frequency chromatic information. In the first experiment, I employed retinotopic mapping methods to identify visual areas and to obtain detailed maps of visual field eccentricities. To ascertain the neural locus of peripheral color sensitivity loss, in the second experiment, I measured fMRI responses as a function of eccentricity in response to high cone-contrast chromatic, as well as luminance modulated circular grating stimuli. Furthermore, I studied the effect of spatial frequency on the eccentricity-dependent response. The experimental findings in the primary visual cortex (V1) closely resemble retinal physiology, and imply that V1 is not associated with the psychophysical sensitivity loss. In addition, I observed a high degree of interaction between spatial frequency and retinal eccentricity. However, by accounting for the primary-cortical projection sizes (cortical magnification) by spatial frequency scaling at different eccentricities, it was possible to achieve an approximately even distribution of responses across eccentricity. When extending the analysis to the extrastriate cortex, it appears that the color-selective area V4 can be regarded as a neural substrate for the psychophysical sensitivity loss to red-green colors in peripheral vision. In the third experiment, I focused on ascertaining the characteristics of retinotopic visual areas in processing high temporal frequency chromatic information. To this aim, fMRI responses to high cone-contrast chromatic and luminance grating stimuli at various temporal frequencies were measured in both the lateral geniculate nucleus (LGN; the primary visual pathway"s thalamic relay station) and cortical visual areas. Special M-scaled circular grating stimulus patterns were designed to eliminate the confounding effect of spatial frequency across visual eccentricity. The fMRI results provide clear evidence that high-temporal-frequency-chromatic information crosses LGN. On arrival in V1, however, blue-yellow information is subjected to low-pass filtering. This finding implies that a loss of psychophysical sensitivity to high temporal frequency blue-yellow information has a neural substrate as early as V1. There was no filtering of high temporal frequency red-green information in V1. Moreover, the data suggest that ventral and dorsal visual areas have distinct specialization for temporal frequency-dependent chromatic information. The ventral areas present with low-pass tuning characteristics, whereas the dorsal areas reveal robust responses to high temporal frequencies. In comparison to the other visual areas, responses in medial temporal area (MT) to luminance modulation show a strong amplification. Furthermore, MT responses increase with increasing temporal frequency, which is in line with MT’s established role in luminance-mediated motion processing. On the basis of the temporal frequency characteristic processing of luminance and chromatic information, it is proposed that visual areas can be hierarchically organized in clusters. A cluster containing the dorsal areas V3d & V3a and, in combination with area MT, constitutes a functional network for the coding of high temporal frequency information. In contrast, another cluster comprising the ventral areas VP & V4 constitutes a functional network for processing low temporal frequency chromatic information. Hence, it might provide a neural substrate for the psychophysical sensitivity loss to high temporal frequency chromatic information. These findings provide neurophysiological evidence for two behaviorally defined processing streams for motion that differ, mostly, in their temporal characteristics.