Shifts in Estimated Preferred Directions During Simulated BMI Experiments With No Adaptation.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Neuroscience

Background:

• Brain-machine interfaces (BMIs) show shifts in estimated preferred direction (EPD) of motor units after transitioning to brain control.
• The underlying cause of these EPD shifts, specifically whether they reflect neural adaptation, remains unclear.

Purpose of the Study:

• To investigate whether observed EPD shifts in cortical motor units during BMI use are due to neural adaptation or other factors.
• To analyze the conditions influencing EPD shifts and provide tools for better understanding BMI experiments.

Main Methods:

• Utilized simulations based on optimal state estimation and feedback control principles.
• Modeled cortical motor neurons encoding estimated state and control vectors.
• Analyzed different BMI filters (linear, Kalman, re-calibrated Kalman) and theoretical conditions for reducing EPD shifts.

Main Results:

• Simulations successfully reproduced apparent EPD shifts observed in BMI experiments, even without assuming neural adaptation.
• Identified experimental conditions, such as correlated velocities and tuning weights, as potential causes for EPD shifts.
• Demonstrated that EPD with respect to actual velocity may not reflect the neuron's true preferred direction (PD) under specific tuning conditions.

Conclusions:

• Observed EPD shifts in BMI experiments may arise from experimental conditions rather than neural adaptation.
• Understanding the relationship between neural tuning, estimated states, and control signals is crucial for interpreting EPD.
• The study provides a theoretical framework and simulation tools to better comprehend EPD phenomena in BMIs.