Goal-Directed BCI Feedback Using Cortical Microstimulation

Abstract

Paralyzed individuals would benefit from brain-computer interface (BCI) systems that restore not just motor function but also tactile and proprioceptive feed- back. Such feedback has been shown to be critical to motor performance. Intracor- tical microstimulation (ICMS) has often been employed to provide artificial sensory feedback. However, it remains a question of how best to encode the multidimen- sional nature of this information (e.g. location, intensity, frequency of tactile signals). This project explored encoding goal-directed error signals as a way to simplify the feedback. We used a behavioral paradigm with rats in which ICMS was used as a tunable error signal to direct the subjects to unseen goal locations. We found that with relatively little training, the rats performance in the task with ICMS feed- back was statistically as good as with natural sensory feedback. The results provide a demonstration that multidimensional sensory feedback can be mapped to single goal-related encoded signal in certain behavioral contexts to decrease the cognitive burden associated with interpreting multiple ICMS-evoked percepts.