Project summary Our lives unfold over time, weaving rich, dynamic, and multisensory information into a continuous experience. However, we remember this as a series of discrete events. For example, the memory of a two-hour movie consists of a few memorable moments tied to the main story. During encoding, we segment deviant events and associate relevant events. During retrieval, we utilize the temporal association among encoded events to search for specific memory information. The hippocampus (HPC), prefrontal cortex (PFC) and substantia nigra (SN) are thought to support cognitive computations (i.e., conceptual prediction, prediction error detection, temporal association) that are critical for the encoding and retrieval of episodic memory. However, how these three regions work together to facilitate the construction of episodic memory in humans remains unclear. The proposed study aims to address this by identifying neural dynamics in the HPC-PFC-SN network and revealing circuit-level neural mechanisms of event segmentation and its relationship with human episodic memory. The central hypothesis is that event segmentation, which is influential in episodic memory formation and retrieval, emerges from the difference between the HPC perceptual predictions and received sensory inputs, which is tracked by dopaminergic neurons in the SN to update event models stored in the PFC. To test this, I will record both single neuron activity and local field potential signals in the HPC-PFC-SN network while patients, who have depth electrodes implanted for clinical purposes, encode, and retrieve the memory of semi-realistic experience created by well-controlled video clips. I will also build a computational model that can rigorously reproduce the observed behavioral and neural signatures. The trained model will be used as a proxy of the HPC-PFC-SN network to study the causal link between this tripartite network and memory behaviors by simulating computational “lesions”, which will provide insightful guidance for real electrical stimulation. To achieve the proposed goals, I will pursue training mentored by a group of experts in different fields, including the intraoperative recordings (Dr. Ziv Williams and Dr. Adam Mamelak), analyses of inter-regional neural dynamics and electrical stimulation (Dr. Ueli Rutishauser), and computational modeling (Dr. Gabriel Kreiman). The comprehensive analytic approaches spinning from behavior measurements, invasive neural recordings from both microscopic and mesoscopic levels, computational modeling and electrical stimulation will provide valuable opportunities to strengthen our understanding of human episodic memory system. The expected outcomes of this proposal will potentially advance the development of therapeutic interventions for memory-related disorders.

Project Narrative Transforming rich and continuous experience into concise and temporally associated event sequences (i.e., event segmentation), a fundamental process for episodic memory, relies on the coordination between the hippocampus, prefrontal cortex, and substantia nigra. This proposal aims to reveal the underlying circuit-level neural dynamics using direct neural recordings (at both the single-cell and population levels), computational modeling, and electrical stimulation. The expected outcomes of this study will provide valuable insights on event segmentation and its influence on episodic memory in humans, which will shed light on developing neural prosthetics to help patients with memory-related disorders.