Neural circuit mechanisms for predicting sensory consequences of behavior

31 March 2017

Nate Sawtell
Department of Neuroscience
Columbia University Medical Center


Perception of the environment requires developing models of and differentiating between externally and self-generated sensations. I will discuss joint theoretical and experimental efforts that provide a detailed account of how specific neural circuitry and plasticity mechanisms support a well-defined and ethologically relevant neural computation of this type. Past work has suggested that cerebellum-like circuitry in weakly electric mormyrid fish serves to predict and cancel self-generated input related to a stereotyped motor behavior – the electric organ discharge. I will discuss how such predictions, termed negative images, are generated at the circuit level: including how brief, highly stereotyped motor command signals are transformed into longer-lasting and more diverse temporal patterns in granule cells, the critical role for spike timing-dependent plasticity in sculpting negative images, and how the temporal structure of granule cell responses appears matched to the self-generated sensory input and also explains unexpected features of negative images. I will also discuss the behavioral significance of these mechanisms for allowing the fish to detect minute electrical signals associated with prey and also how these mechanisms may be extended to understand the more complex and ubiquitous problem of predicting the sensory consequences of movements. Finally, I will discuss our recent efforts to extend insights from electric fish to understand the possible roles of similar, though independently evolved, cerebellum-like circuits in the mammalian auditory system in filtering out self-generated sounds.

current theory lunch schedule