2013 May 30; 497(7451): 585–590. doi:10.1038/nature12160. | Mattia Rigotti1,2, Omri Barak1,*, Melissa R. Warden3,4, Xiao-Jing Wang2,5, Nathaniel D. Daw2, Earl K. Miller3, and Stefano Fusi1
The study investigates the importance of mixed selectivity in complex cognitive tasks, focusing on prefrontal cortex (PFC) neurons. Mixed selectivity, characterized by neurons encoding distributed information about multiple task-related aspects, is highly heterogeneous and difficult to interpret. The researchers analyzed neural activity in monkeys performing an object sequence memory task to understand the role of mixed selectivity in cognitive functions. They found that mixed selectivity neurons encode distributed information about all task-relevant aspects, even when single-cell selectivity to individual aspects is eliminated. This mixed selectivity offers a significant computational advantage over specialized responses, enabling a wide range of input-output functions. The dimensionality of the neural representations, which is a measure of the complexity and diversity of the representations, predicts animal behavior, collapsing in error trials. The findings suggest that the focus should shift from easily interpretable neurons to the less studied mixed selectivity neurons, which are crucial for understanding complex cognitive tasks.The study investigates the importance of mixed selectivity in complex cognitive tasks, focusing on prefrontal cortex (PFC) neurons. Mixed selectivity, characterized by neurons encoding distributed information about multiple task-related aspects, is highly heterogeneous and difficult to interpret. The researchers analyzed neural activity in monkeys performing an object sequence memory task to understand the role of mixed selectivity in cognitive functions. They found that mixed selectivity neurons encode distributed information about all task-relevant aspects, even when single-cell selectivity to individual aspects is eliminated. This mixed selectivity offers a significant computational advantage over specialized responses, enabling a wide range of input-output functions. The dimensionality of the neural representations, which is a measure of the complexity and diversity of the representations, predicts animal behavior, collapsing in error trials. The findings suggest that the focus should shift from easily interpretable neurons to the less studied mixed selectivity neurons, which are crucial for understanding complex cognitive tasks.