An Ultra Sparse Code Underlies The Generation Of Neural Sequences In A Songbird

Motor sequences are encoded in many vertebrate brains by complex spatio-temporal patterns of neural activity. However, the neural circuit mechanisms underlying the generation of these premotor patterns are poorly understood. In the songbird, one prominent site of premotor activity is forebrain nucleus RA, which generates stereotyped sequences of spike bursts during song, and recapitulates these sequences during sleep [1,2]. 

Here we show that the stereotyped sequences in RA are driven from nucleus HVc, the principal premotor input to RA [3,4]. Recordings of identified HVc neurons produce bursts sparsely, at a single precise time during the RA sequence. Furthermore, these HVc neurons burst sequentially with respect to one another. 

We suggest that each time-step in the RA sequence is driven by a sub-population of RA-projecting HVc neurons that is active only at that time. Thus, as a population, the HVc neurons form an explicit representation of time in the RA sequence. Such a sparse representation, a temporal analog of the "grandmother cell" [5] concept for object recognition, eliminates the problem of temporal interference during sequence generation and learning that has been attributed to more distributed representations [6,7].