Event

PhD Dissertation Defense: Agampodi Dulara De Zoysa

Thursday, July 31, 2025
1:00 p.m.
PSC 1136 (1st floor conference room, Physical Sciences Complex)
Debbie Chu
301 405 8268
dgchu@umd.edu

Title: Real-time analysis and photostimulation for collective learning in living neural networks

Committee members:
Dr. Wolfgang Losert, Chair
Dr. Pamela A. Abshire
Dr. Giuliano Scarcelli
Dr. Huang Chiao Huang
Dr. Behtash Babadi, Dean's Representative

Abstract:
The brain’s ability to process and encode information is driven by complex, dynamic interactions between neurons across diverse circuits. Traditional analysis methods often fail to capture the real-time dynamics of these interactions, especially when studying neural circuits at single-cell resolution. These conventional techniques typically provide post hoc insights, limiting our understanding of the transient and intricate nature of neural dynamics, especially during collective learning. To address these limitations, there is a critical need for real-time analysis pipelines that can resolve and modulate neural activity as it occurs, enabling the study of dynamic processes in living neural networks. This thesis introduces NeuroART, a platform for real-time neural activity analysis and photostimulation, designed to overcome these challenges. By integrating two-photon (2P) calcium imaging with adaptive optics for Zernike polynomial-based aberration correction, NeuroART allows for high spatial resolution and effective optogenetic stimulation of single neurons. Furthermore, optical vortex beams are utilized for photostimulation, shaping light to match the size of neuronal membranes and minimizing photodamage by reducing localized intensity hotspots.

Real-time optogenetic stimulation capabilities of NeuroART were utilized to study spike timing dependent plasticity (STDP) based collective learning mechanisms of neuronal groups, and one of the key findings of this work is that providing contextual inputs through background stimulation leads to rapid Hebbian learning within neuronal groups in less than two minutes. This discovery emphasizes that coordinated, context-driven activity enables rapid ensemble-level learning, offering new insights into how learning and memory emerge from neural circuit dynamics. The closed-loop, activity-guided stimulation provided by NeuroART allows for precise, real-time exploration of these mechanisms, shedding light on how neural activity of localized ensembles contributes to broader cognitive functions.

This thesis not only advances our understanding of collective learning mechanisms of living neural networks but also provides a powerful framework for future investigations into complex neural dynamics, neural plasticity, and information flow by integrating real-time neural activity analysis with precise optogenetic stimulation through the NeuroART platform.