Population coding in the primary visual cortex

Abstract

In this thesis, the nature of the neural code employed by the primary visual cortex is examined. Recordings of single, as well as pairs of cortical cells, were analyzed as they were presented with natural movie stimuli. The responses were found to be less variable than what has been reported in literature, which meant that information could potentially be reliably encoded using the intervals between spikes. Using techniques from information theory, it was found that on average, 135% more information could be encoded by taking individual spike intervals into account compared to simply counting those intervals. Furthermore, if joint responses form pairs of neurons were used, 35% more information was encoded if the identity of those neurons were kept, compared to simply pooling their inputs together. Finally, in an attempt to analyze larger neuronal population, a novel algorithm was developed to assign recorded spikes to combinations of putative neurons.