Different state-dependence of population codes across cortex

Akhil Bandi,Caroline A. Runyan
DOI: https://doi.org/10.1101/2024.05.23.595581
2024-05-26
Abstract:During perceptual decision-making, behavioral performance varies with changes in internal states such as arousal, motivation, and strategy. Yet it is unknown how these internal states affect information coding across cortical regions involved in differing aspects of sensory perception and decision-making. We recorded neural activity from the primary auditory cortex (AC) and posterior parietal cortex (PPC) in mice performing a navigation-based sound localization task. We then modeled transitions in the behavioral strategies mice used during task performance. Mice transitioned between three latent performance states with differing decision-making strategies: an optimal state and two sub-optimal states characterized by choice bias and frequent errors. Performance states strongly influenced population activity patterns in association but not sensory cortex. Surprisingly, activity of individual PPC neurons was better explained by external inputs and behavioral variables during suboptimal behavioral performance than in the optimal performance state. Furthermore, shared variability across neurons (coupling) in PPC was strongest in the optimal state. In AC, shared variability was similarly weak across all performance states. Together, these findings indicate that neural activity in association cortex is more strongly linked to internal state than in sensory cortex.
Neuroscience
What problem does this paper attempt to address?