Circuit-level modeling of prediction error computation of multi-dimensional features in voluntary actions

Yishuang Huang^1*Yiling Li²

• ¹Zhejiang University, Hangzhou, China
• ²China Academy of Engineering Physics, Mianyang, China

Predictive processing, a generative framework that explains how the brain minimizes differences between its internal predictions and actual sensory inputs, has been proposed to account for various brain functions in perception, cognition and behavior. In voluntary actions, predictive processing is thought to suppress self-generated sensory outcomes.Although recent studies have extensively investigated and modeled sensory mismatch signals (prediction errors in sensory modalities), mechanistic insights into the neural computation of predictive processing in voluntary actions are still largely lacking. In the present study, we developed a computational model consisting of two-compartment excitatory pyramidal cells (PCs) and three major types of inhibitory interneurons with biologically realistic connectivity. Our model reveals that top-down predictions can selectively suppress pyramidal cells with matching feature selectivity through experience-dependent inhibitory plasticity. This suppression relies on the response selectivity of inhibitory interneurons and the balance between excitation and inhibition across multiple pathways. We also expanded our model to handle predictions with two independent features, common in real-world actions. By combining biological connectivity data with computational modeling, our study provides insights into the neural circuits and computations underlying the active suppression of sensory responses in voluntary actions. This work contributes to the understanding of how the brain generates and processes predictions to guide behavior.