Examining brain areas to identify generalizable neural features to improve brain-computer interface precision and reliability

The National Institute of Neurological Disorders and Stroke awarded the University of Kansas Medical Center Research Institute $580,517 on September 3, 2026, to examine neural features that improve brain-computer interface precision and reliability under the Extramural Research Programs in the Neurosciences and Neurological Disorders program (CFDA 93.853).

The recipient will conduct three experiments to improve BCI decoders that translate neural signals into reaching and grasping when task conditions change. Aim 1 investigates how combinations of wrist orientation and hand shape affect grasping, testing whether reaching and grasping are object-centric with grasp determined first and reach adjusted accordingly. Aim 2 will evaluate BCI decoders designed to enhance movement timing and speed profiles to resemble natural motion, testing whether condition-invariant neural activity improves movement timing and whether the primary motor cortex best provides these signals. Aim 3 will examine neural tuning to novel reach directions similar to previously learned ones, testing whether increasing the number of neural dimensions used improves decoding accuracy compared to traditional linear decoders. Recordings will be taken from the primary motor cortex, premotor cortex, anterior intraparietal area, and parietal reach region. Analyses will range from single-neuron firing rates to advanced neural population models. The experiments will identify BCI-relevant neural features preserved across tasks and clarify how population-level activity supports consistent reaching and grasping under varying conditions.

Work is performed in Kansas City, Kansas, with a period of performance from September 3, 2026, through May 31, 2031.

Abstract

Everyday life requires the ability to grasp a wide variety of objects. Brain-computer interfaces (bcis) enable individuals with paralysis to reach and grasp by using brain signals to control robotic limbs or stimulate their own muscles. However, users often struggle to perform novel movements that are similar, but not identical, to previously learned ones. The proposed experiments aim to improve BCI decoders that translate neural signals into reaching and grasping when task conditions change. We will conduct three experiments to address this goal. Aim 1 will investigate how combinations of wrist orientation and hand shape affect grasping, testing hypothesis H1-that reaching and grasping are object-centric, with the grasp determined first and the reach adjusted accordingly. Aim 2 will evaluate BCI decoders designed to enhance movement timing and speed profiles to resemble natural motion, testing hypothesis H2-that condition-invariant neural activity improves movement timing and that the primary motor cortex best provides these signals. Aim 3 will examine neural tuning to novel reach directions that are similar to previously learned ones, testing hypothesis H3-that increasing the number of neural dimensions used improves decoding accuracy compared to traditional linear decoders. Recordings will be taken from the primary motor cortex, premotor cortex, anterior intraparietal area, and parietal reach region. Analyses will range from single-neuron firing rates to advanced neural population models. These experiments will identify BCI-relevant neural features that are preserved across tasks and clarify how population-level activity supports consistent reaching and grasping under varying conditions. Improving our understanding of neural movement encoding will lead to more effective bcis and robotic systems that better replicate natural grasping, ultimately enhancing rehabilitation outcomes for individuals with motor deficits due to brain injury.