Distribution of colour-selective activity in the monkey inferior temporal cortex revealed by functional magnetic resonance imaging

In monkey visual cortex, colour information is primarily processed in the ventral pathway through areas V1, V2, and V4 into the inferior temporal (IT) cortex. Recent studies have shown that the IT cortex, which is situated at the final processing stage of the ventral visual pathway, plays an important role in colour processing. Single cell recording studies have shown that there are numerous colour-selective neurons in the IT cortex, some of which are narrowly tuned to hues and/or saturations and exhibit task-related responses during colour discrimination. In addition, lesion studies have shown that bilateral ablation of the IT cortex disrupts colour discrimination, though ablation of V4 does not. Although many colour-selective neurons have been found in the IT cortex, their distribution in this cortical region remains not fully clear, especially in the anterior part of the IT cortex. In the present study, we explored the distribution of colour-selective activity in the IT cortex using functional magnetic resonance imaging (fMRI) in alert macaque monkeys. As exploratory examinations, we have measured retinotopic organization, motion-selective activity and shape-selective activity of the visual cortex. The retinotopic organization was obtained by comparing responses to the horizontal meridian and vertical meridian stimuli and was helpful in determining the boundaries of visual areas. The motion-selective activity was obtained by contrasting response to moving random-dot stimuli with that to static random-dot stimuli and used to determine the physiological isoluminance point of colour grating. The shape-selective activity was obtained by contrasting responses to object image stimuli and scrambled image stimuli. The distribution of the shape-selective activity was later compared with that of the colour-selective activity. These exploratory experiments were required for analyzing the results in the main experiment. In the main experiments, we examined colour-selective activity in the IT cortex with two types of stimuli: an isoluminant colour grating and a multicoloured (‘Mondrian’) pattern that have been commonly employed in human fMRI and have identified colour-selective areas/regions in the fusiform gyrus, which is suggested to correspond to the monkey IT cortex. The paradigm using gratings is to compare responses to an isoluminant, colour-varying grating (e.g. a red-blue grating) with responses to a luminance-varying grating. This paradigm identified colour-selective regions in the posterior part of the human fusiform gyrus that is called V8/VO. The paradigm using Mondrian is to compare responses to a multicoloured (‘Mondrian’) pattern with responses to its achromatic counterpart. This paradigm identified colour-selective regions in the more anterior part of the fusiform gyrus, in addition to the posterior part. Earlier imaging studies of the monkey IT cortex used only the first paradigm, but the second paradigm may elicit more effectively the colour-selective activity in the anterior IT according to the findings in human fMRI. Moreover, Mondrian stimuli may be more suitable for activating higher areas, since the stimuli containing various hues could elicit responses from many neurons, each tuned to a specific hue. For that reason, we have used both grating and Mondrian stimuli to examine colour-selective activity in the IT cortex. As a result, we found that colour-selectivity is not uniformly distributed in the IT cortex, but is clustered in discrete subregions that are located in the posterior and the anterior part of the IT cortex. The colour-selective activity in the posterior IT was obtained both with the grating and Mondrian stimuli, but the positions of the activity were different dependent on the stimuli. On the other hand, the colour-selective activity in the anterior IT was obtained only with the Mondrian stimuli. We have examined whether these topographical differences of colour-selective activity depending on the stimuli were attributable to the difference in the luminance contrast between the chromatic Mondrian stimuli and colour grating stimuli: although the luminance contrast in the chromatic Mondrian stimulus matched that in the achromatic Mondrian, the colour grating stimulus contained much smaller amount of luminance contrast than the luminance grating. To examine the possible influence of the luminance contrast in the chromatic Mondrian stimuli, we employed isoluminant Mondrian stimuli that contained much less luminance contrast than the achromatic ones, and compared the distributions of colour-selective activity obtained with this isolumiant Mondrian, the Mondrian containing matched luminance contrast and the isoluminant grating. The results have shown that the topography of the colour-selective activity obtained with the isoluminant Mondrian was very similar to that obtained with the Mondrian containing matched luminance contrast whereas it was very different from that obtained with the isoluminant grating. This indicates that the difference in the distribution of the responses to the Mondrian and grating stimuli in the IT cortex cannot be attributed solely to the difference in luminance contrast. We finally compared the distribution of colour-selective activity with that of shape-selective activity in the IT cortex. We found that the colour- and shape-selective activity tends to overlap little in the anterior IT whereas the colour-selective activity in the posterior IT and early visual areas overlapped with the shape-selective activity. This finding was consistent with the results of previous electrophysiological recording experiments comparing the responses of neurons to colour and shape stimuli and suggests that colour and shape information is clustered in different modules specifically tuned to each attribute. The present study shows that colour-selective activity is clustered in discrete regions of the monkey IT cortex and that these colour-selective regions are distributed in both the anterior and posterior IT. The difference in the response properties and the retinotopy suggests that these regions may correspond to different spots of colour-selective activity reported in the human fusiform gyrus: the colour-selective activity in the posterior IT may correspond to V8/VO which is in the posterior part of the fusiform gyrus, and colour-selective activity in the anterior IT may correspond to V4α and regions in the more anterior part of the fusiform gyrus