Context-dependent proprioceptive encoding

Circuits processing proprioceptive information face two important challenges. First, proprioceptive processing must be fast: animals often only have a few hundred milliseconds to integrate information from disparate proprioceptors before relaying that information to motor circuits. Second, proprioceptive processing must be flexible: the influence of proprioceptive feedback should change with behavioral context. An effective strategy to simultaneously deal with speed and flexibility constraints may be using an efference copy to predict movement outcomes, also known as a forward model. By combining an internal representation of the expected limb state with an external measure of actual limb state, neurons can compensate for feedback delays and affect motor commands only during unexpected movements.

Does the fly leg motor control system use a forward model? If so, then we would expect that cells within the fly’s central nervous system would differentially encode externally- and self-generated movements. To test for this difference, I am first building models describing how propriocceptive neurons encode externally-generated leg movements. I am recording neural activity while moving the fly leg along a pseudo-random broadband or naturalistic trajectories, and then will construct a model that accurately predicts the cell’s activity. Next, I will record the activity of these central neurons during spontaneously generated movements by the fly, and compare this activity to what is predicted by my model.

This work is currently underway in the lab of Dr. John Tuthill at the University of Washington.