UCLA

This thesis investigates changes in neural dynamics during spatial learning by recording the activity of large populations of CA1 hippocampal neurons during performance of the Morris water maze (MWM) navigation task.

Employing a generalized linear model, we found significant enhancements in the sparsity of both spatial and head-directional firing rates, indicating more refined neuronal representations after learning. Artificial neural networks successfully decoded position, head direction, and distance-to-platform from neural activity patterns, with decoding accuracy improving as animals mastered the task.

To determine if long-term potentiation (LTP), a form of synaptic plasticity important for learning, was essential for learning associated changes in hippocampal representations, we recorded CA1 hippocampal activity in GluA1\textsuperscript{C2KI} mutant mice, which have impaired CA1 LTP and learning deficits. We found that GluA1\textsuperscript{C2KI} mice exhibited impaired spatial memory consolidation during probe trials, suggesting a specific impairment in the retention of learned spatial information. Single-cell and network analyses in GluA1\textsuperscript{C2KI} mice revealed no significant increase in head-directional sparsity with learning, nor any notable reduction in decoding error for head-direction, and distance-to-platform information. These findings suggest a potential role of LTP in fine-tuning neural representations associated with task learning. These findings underscore the importance of LTP mechanisms in spatial memory and highlight its implications for cognitive disorders involving spatial navigation deficits.