לוח שנה משולב

Studying the neural code deals with trying to understand how information is stored and processed in the brain, searching for basic principles of this "language". The study of population codes aims to understand how neural populations collectively encode information, and to map interactions between neurons. Previous studies explored the firing rates of single cells and how they evolve with time. Other studies have shown that neural populations are correlated and explored spatial activity patterns of large groups. In this work we combine these approaches, and study population activity of large groups of neurons, and how they evolve with time.

We studied the fine temporal structure of spiking patterns of groups of up to 100 simultaneously recorded units in the prefrontal cortex of monkeys performing a visual discrimination task. We characterized the population activity using 10 ms time bins and found that population activity patterns (codebooks) were strongly shaped by spatial correlations. Further, using a novel extension of models which describe spatio-temporal population activity patterns, we show that temporal sequences of population activity patterns have strong history-dependence. Together, the large impact of spatial and temporal correlations makes the observed sequences of activity patterns many orders of magnitude more likely to appear than predicted by models that ignore these correlations and rely only on the population rates.

Surprisingly, despite these strong correlations, decoding behavior using models that were trained ignoring these correlations perform as well as decoders that were trained to capture these correlations. The difference in the role of correlations in population encoding and decoding suggests that one of the goals of the complex encoding scheme in the prefrontal cortex may be to create a code that can be read by simple downstream decoders that do not have to learn correlations.