Volitional generation of reproducible energy-efficient temporal patterns

Abstract

One of the extraordinary characteristics of the biological brain is its low energy expense to implement a variety of biological functions and intelligence compared to the modern artificial intelligence (AI). Spike-based energy-efficient temporal codes have long been suggested as the contributor for the brain to run with a low energy expense. Despite this code having been largely reported in the sensory cortex, whether this code can be implemented in other brain areas to serve broader functions and how the brain learns to generate it have remained unaddressed. In this study, we designed a novel brain-machine interface (BMI) paradigm, by learning which two macaques could volitionally generate reproducible energy-efficient temporal patterns in the primary motor cortex (M1). Moreover, most neurons that were not directly assigned for controlling the BMI did not boost their excitability, demonstrating an overall energy-efficiency manner in performing the task. Over the course of learning, we found that the firing rates and temporal precision of selected neurons co-evolved to generate the energy-efficient temporal patterns, suggesting a cohesive rather than dissociable processing underlie the refinement of energy-efficient temporal patterns.