Neurons in primary visual cortex encode naturalistic visual information using multiple temporal scales

Natural stimuli have a rich temporal structure, yet it is still unclear whether the encoding of such stimuli employs more than one response time scale. To investigate this issue, we analyzed the activity of neurons recorded in primary visual cortex of anesthetized macaques during presentation of naturalistic movies, and we quantified the amount of information carried by neural codes operating at different temporal scales. We divided the recording time into stimulus windows of tens of ms, and computed the information carried by the neural response about which stimulus window was being shown. Within each time window, responses were quantified using either the spike count, or using binary spike patterns, defined by the absence/presence of spikes within short time bins (t). We found that temporal spike patterns with precision t of 8 or 16 ms provided more stimulus information than spike counts, but this information gain did not increase further when reducing the bins size t. This suggests a response precision of single neurons on the scale of 8ms. In addition, we found that the joint knowledge of spike patterns and the phase of low frequency LFPs at which these patterns occurred - computed as in (Kayser et al. Neuron 2009) - carried more information than either code considered by itself. This suggests that the information carried by slow LFP fluctuations complements that carried by spike patterns. In summary, we found evidence for complementary response time scales for the encoding of naturalistic stimuli in visual cortex. Informative codes range from spike timing precision at about 10 ms resolution to the much coarser phase of firing with respect to low frequency fluctuations. These findings indicate that sensory cortices may enhance their information capacity by multiplexing complementary information at different time scales