Probing Connectivity Update Rules in Biological Neural Networks in vivo

The ability to learn from experience is a defining feature of both biological and artificial intelligence. While today's artificial intelligence systems excel at many tasks, they often require massive amounts of data, struggle to continually learn new information without forgetting previous knowledge, and rely on learning rules that differ substantially from those used by the brain. This project seeks to discover the biological rules that determine when and where neural connections are updated in the living brain. By revealing how the brain selectively modifies only the appropriate connections while preserving existing knowledge, this research will advance fundamental neuroscience and provide experimentally grounded principles for developing more adaptive, efficient, and continual-learning artificial intelligence systems. The project will also train graduate students in cutting-edge neuroscience, optical imaging, computational analysis, and interdisciplinary research while broadly disseminating new methods and discoveries. This project will identify the biological principles that govern how synaptic connectivity is updated in the living brain. The central hypothesis is that efficient learning depends on connectivity update rules that determine not only how neural activity modifies synaptic strength, but also when and where those modifications are permitted. To test this hypothesis, the investigators will develop an all-optical approach that combines two-photon calcium imaging, single-cell holographic optogenetic stimulation, and functional connectivity mapping to measure and experimentally induce plasticity in awake mice with single-cell resolution. The project will determine how behavioral state, neuromodulatory signaling, and local inhibitory circuits regulate connectivity update rules, and how local network organization, spatial arrangement, and functional identity of neuronal ensembles influence which connections are modified. The resulting experimentally validated framework will establish fundamental principles governing context-dependent connectivity updates in biological neural circuits and provide biologically grounded learning rules that can be translated into computational models and evaluated in artificial neural networks, potentially informing future AI systems capable of more selective, adaptive, and continual learning. The project advances NSF priorities in Artificial Intelligence and Biotechnology. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.