Description:
- Neurological disorder treatment
- Mental health treatment
- Neuroscience research
Abstract
USC researchers have developed preferential subspace identification (PSID), an innovative algorithm designed to model neural activity while prioritizing behaviorally relevant dynamics. In experiments with two monkeys performing reach and grasp tasks, PSID revealed lower-dimensional behaviorally relevant dynamics and distinct rotational patterns that predicted behavior and learned relevant dynamics for each joint and recording channel. When tested with saccades, PSID was versatile across behaviors, brain regions, and neural signal types. This groundbreaking tool uncovers crucial behaviorally relevant neural dynamics that might otherwise remain unnoticed, offering valuable insights for neuroscience research.
Benefit
- Modeling of behaviorally-relevant neural activity
- Accurate and reliable across different behaviors, brain regions, and signal types
Market Application
In neuroscience, current dynamic models of neural activity lack the ability to dissociate behaviorally relevant and irrelevant dynamics, hindering the understanding of brain functions and neurological dysfunctions. Building a novel dynamic modeling framework that prioritizes extracting neural dynamics related to specific measured behaviors is essential. This new approach will enable the study of diverse brain functions, including movement, speech, decision-making, mood, and neurological disorders.
Publications
Modeling behaviorally relevant neural dynamics enabled by preferential subspace identification, Sani et al., 2021.
Other
Stage of Development
- Experimentally validated in vivo with monkeys