Revealing unexpected complex encoding but simple decoding mechanisms in motor cortex via separating behaviorally relevant neural signals

Abstract

Understanding how motor cortex encodes and decodes behaviors is a fundamental goal of neuroscience, which faces significant challenges because taskrelevant neural responses are entangled with considerable task-irrelevant nuisance signals. Due to the entanglement, it is difficult to determine to extract which part and how much neural activity to study neural mechanisms. To cope with this puzzle, current studies usually extract a subset of neural responses that satisfy empirical hypotheses from noisy signals for analysis, which may lead to a one-sided understanding of neural mechanisms. Here we proposed a new perspective — extracting complete task-relevant signals — to explore neural mechanisms and offered a distillation framework to distill signals. Analyzing data in three monkeys performing different reaching tasks, distilled task-relevant signals reveal unexpected complex encoding but simple decoding mechanisms in motor cortex.