Reading the Neural Code: What do Spikes Mean for Behavior?

The present study reveals the existence of an intrinsic spatial code within neuronal spikes that predicts behavior. As rats learnt a T-maze procedural task, simultaneous changes in temporal occurrence of spikes and spike directivity are evidenced in “expert” neurons. While the number of spikes between the tone delivery and the beginning of turn phase reduced with learning, the generated spikes between these two events acquired behavioral meaning that is of highest value for action selection. Spike directivity is thus a hidden feature that reveals the semantics of each spike and in the current experiment, predicts the correct turn that the animal would subsequently make to obtain reward. Semantic representation of behavior can then be revealed as modulations in spike directivity during the time. This predictability of observed behavior based on subtle changes in spike directivity represents an important step towards reading and understanding the underlying neural code.