A new paper in PNAS from recent CNC PhD graduate Alison Duffy and former CNC undergraduate fellow Elliot Abe (currently a PhD student at U Oregon) provides insight into how songbirds robustly maintain their precise songs in the face of environmental and neural changes. CNC faculty members David Perkel and Adrienne Fairhall also contributed.
Courtship song in birds is a well-characterized model of a learned and maintained behavior. Juvenile birds learn from adults in a trial-and-error manner. Adult birds are able to precisely replicate their songs but show shifts in response to environmental factors. Neurons in birds are also known to undergo continuous and seasonal replacement. Using a computational model of bird song learning, Duffy et al. examined how a static learned behavior responds to instability of neuronal population, and if this instability provides any advantage.
Duffy et al. used of several different approaches to perturb activity in the cells that govern birds’ song system, mimicking the effects of cell death and replacement and variable cell participation. Their results found that variations in cell activity slightly degrade song quality, but allowed the system to more rapidly adapt to changes in muscle activity and withstand cell loss in the downstream neural network. Thus, the ongoing instability in the neurons controlling song execution allows for a quicker adaptation to changes in environment, increasing robustness and maintenance of the precise song. They conclude that variability in neural systems provides a constructive advantage in maintaining skilled behavior.