Abstract
Objects in the natural world usually have multiple features that each can vary in strength. We previously showed that neuronal responses in extrastriate area MT elicited by two overlapping, moving stimuli can be described as a weighted sum of the neuronal responses, and the weight is stronger for the stimulus that has a stronger signal strength. These results can be explained by response normalization. Here we investigate how MT neurons of fixating monkeys respond to multiple moving stimuli that have two competing features, motion coherence and luminance contrast. Visual stimuli were two random-dot patches moving simultaneously in different directions separated by 90°. The diameter of each patch was 3°. One patch moved at a high coherence (100%) with a low contrast (36%) (Hcoh/Lctr), whereas the other moved at a low coherence (60%) with a high contrast (77%) (Lcoh/Hctr). The two patches were placed within the receptive fields (RFs) of MT neurons, and either overlapped, or were separated with at least 1° gap in-between. We varied the vector-averaged direction of the two patches to characterize the direction tuning. We found that although the MT response elicited by Hcoh/Lctr patch alone was significantly stronger than that elicited by Lcoh/Hctr patch, MT response elicited by two patches was strongly biased to the weaker Lcoh/Hctr patch response when the two patches overlapped. When the two patches were separated, MT response showed a drastic change and became slightly biased to the stronger HCoh/Lctr patch response. When overlapping, the two patches compete within the RFs of both V1 and MT neurons, whereas when separated, the two patches compete within the RFs of only MT, but not V1 neurons. Our results highlight the importance of normalization occurring at different stages of visual processing that differ in spatial scale and preference for stimulus feature in shaping neuronal responses.
Meeting abstract presented at VSS 2017