PROVIDING ARTIFICIAL TACTILE FEEDBACK AT DIFFERENT LATENCIES IN A BRAIN-MACHINE INTERFACE

Brain-Machine Interfaces (BMIs) aim to improve patient autonomy. Beyond restoring movement, fine control of prosthetic devices requires restoring tactile sensory feedback. While BMIs with artificial somatosensory inputs have recently been used in patients, few studies have explored the spatio-temporal constraints of feedback integration.
This project examines how temporal latency between motor commands and sensory feedback affects control. We developed an ultra-fast bidirectional BMI using chronic recordings from whisker-related primary motor cortex (wM1) and 2D patterned optogenetic stimulation of whisker primary somatosensory cortex (wS1) in mice.
We designed a behavioral task where single wM1 neuron spikes controlled the rotation of a virtual bar. A photostimulation pattern on wS1 provided feedback about the prosthesis angle during a reaching task. Our incremental algorithm enabled fine control, with well-guided trajectories achieved using a 50-ms feedback latency. Altering this latency to 5 or 500 ms disrupted the animals' ability to move and stabilize the prosthesis, suggesting a critical time window for S1-M1 interaction.
We also explored the sensations evoked by optogenetic wS1 stimulation. After the mice mastered the BMI task with cortical stimulation, we replaced it with physical stimulation using a moving bar on the whisker array. This peripheral input targeted body regions corresponding to the previously stimulated cortical sites. The mice retained performance without relearning. Thus, S1 optostimulation appears to evoke perceptions similar to real tactile inputs, highlighting its biomimetic potential.