Intuitively, memory is conceived as a collection of static images that we accumulate as we experience the world. But actually, memories are constantly changing through our life, shaped by our ongoing experiences. Assimilating new knowledge without corrupting pre-existing memories is then a critical brain function. However, learning and memory interact: prior knowledge can proactively influence learning, and new information can retroactively modify memories of past events. The hippocampus is a brain region essential for learning and memory, but the network-level operations that underlie the continuous integration of new experiences into memory, segregating them as discrete traces while enabling their interaction, are unknown. Here I show a network mechanism by which two distinct memories interact. Hippocampal CA1 neuron ensembles were monitored in mice as they explored a familiar environment before and after forming a new place-reward memory in a different environment. By employing a network science representation of the co-firing relationships among principal cells, I first found that new associative learning modifies the topology of the cells’ co-firing patterns representing the unrelated familiar environment. I fur- ther observed that these neuronal co-firing graphs evolved along three functional axes: the first segregated novelty; the second distinguished individual novel be- havioural experiences; while the third revealed cross-memory interaction. Finally, I found that during this process, high activity principal cells rapidly formed the core representation of each memory; whereas low activity principal cells gradually joined co-activation motifs throughout individual experiences, enabling cross-memory in- teractions. These findings reveal an organizational principle of brain networks where high and low activity cells are differentially recruited into coactivity motifs as build- ing blocks for the flexible integration and interaction of memories.
Finally, I employ a set of manifold learning and related approaches to explore and characterise the complex neural population dynamics within CA1 that underlie sim- ple exploration.