UCL Institute of Behavioural Neuroscience
Roberts Building 508 (<map>)
Distributed encoding of visual space and visual motion in area MT of the marmoset
The middle-temporal (MT) area of primate visual cortex is an important stage in the analysis of visual motion, and the response properties of single units in area MT are well established. We know much less about the response of populations of neurons in area MT. Here we characterised functional connectivity of neurons in area MT of marmoset monkeys, where the area lies exposed on the cortical surface. Our first analyses used standard methods to reveal the spatial- and temporal distribution of noise correlations. The amplitude of noise correlations in area MT decreased with increasing distance between pairs of neurons, and with increasing difference in their preferred direction. Analysis of correlation time-course showed that synchronous spiking is restricted to nearby neurones with highly overlapping receptive fields; noise correlations over wider areas of cortex, perhaps the whole hemifield representation, primarily reflect mechanisms with longer time courses. The tight relationship between receptive field overlap and near-synchronous activity may provide constraints on the capacity of area MT populations to represent spatial position. Our second analysis therefore explored how populations of neurones in area MT might encode spatial position. We show that populations of neurons in area MT are capable of fine spatial analysis even though the receptive fields of individual neurons are very large. Our analyses show that this population acuity relies on neurons with receptive fields that flank the position of an object. In addition we find a potential role of inter-neuronal correlations in spatial vision. These correlations are richer in structure than previous work has revealed: we show why that is the case, and how they can be exploited.
Page last modified on 25 mar 14 10:11