BCI Study Shows Neural Activity Scales With Human Likeness, Not Mirror Neurons

Researchers at Brown University, in collaboration with the Mass General Brigham Center for Neurotechnology and Neurorecovery and the VA Center for Neurorestoration and Neurotechnology, have published new findings in the Proceedings of the National Academy of Sciences that refine the understanding of observational motor activity traditionally associated with mirror neurons. The study, conducted as part of the BrainGate clinical trial, utilized intracortical brain-computer interface systems to record single-neuron activity in the motor cortex of two participants with tetraplegia. The objective was to determine how visual feedback and stimulus anthropomorphism influence neural responses during action observation, a factor critical for optimizing brain-controlled assistive devices. During controlled sessions, participants observed animations of various gripping effectors, ranging from a biological human hand and an anthropomorphic robotic hand to a three-pronged mechanical claw and a simple cube. Data revealed that observational neural activity was strongest when viewing the human hand and decreased proportionally as the stimuli became less humanlike. This graded response occurred at both the network level, with more neurons firing for realistic representations, and the individual neuron level, which exhibited faster firing rates. Notably, when participants actively attempted to perform the grips themselves, neural activity became uniform across all effectors, indicating that motor execution signals override observational cues during voluntary control. A secondary experimental series using dot-pattern animations further illuminated the cognitive dimensions of this neural phenomenon. High-fidelity dot patterns that clearly resembled hands elicited robust observational responses, while sparse, amorphous patterns generated weaker signals. However, the study captured a significant cognitive shift when one participant successfully recognized an abstract dot pattern as a hand. Following this realization, neural activity increased across all levels of anthropomorphic abstraction, demonstrating that top-down contextual awareness and conscious recognition can amplify bottom-up visual processing. These results indicate that observational motor activity is not governed by dedicated mirror neuron cells, but rather by a flexible neural network modulated by visual fidelity and cognitive context. The findings carry direct implications for the development of next-generation implantable brain-computer interfaces. By clarifying how visual feedback and perceived human-likeness shape motor cortex activity, engineers can design more intuitive assistive robotics and computer cursors that align with natural neural processing patterns. Lead author Jacob Gusman emphasized that understanding these visual feedback mechanisms will enhance patient control over external devices. Co-lead investigator Leigh Hochberg noted that while the primary mission of the BrainGate trial is to restore function to individuals with paralysis, such fundamental neuroscience discoveries are essential for advancing the field. The research establishes a clearer framework for integrating cognitive and visual parameters into BCI algorithms, ultimately accelerating the delivery of reliable neurorestorative technologies for patients with severe motor impairments.