Accessing the diversity of hippocampal ripples

Thesis

The brain is capable of retaining prior knowledge while continuously incorporating new information. How neuronal populations support this dual function — updating internal representations without disrupting existing ones — remains a fundamental biological question. Hippocampal ripples have been identified as key events supporting offline processing and memory consolidation. These short-lived, highly synchronised patterns reactivate waking experiences during sleep and rest. Yet, it remains unclear whether, at the level of individual ripple events, the associated laminar currents, neuronal firing patterns, and reactivation content are consistent across events or flexibly tuned on a ripple-by-ripple basis.

In this thesis, I addressed this question by profiling individual ripple events using laminar current source density recordings in the mouse hippocampus during sleep and rest. I identified two ripple profiles — Rad^sink and LM^sink ripples — characterised by current sinks in stratum radiatum and stratum lacunosum-moleculare, respectively. The distinct laminar signatures of these ripple types suggest they may arise from separate hippocampal circuit mechanisms. Each profile was associated with characteristic spectral features and LFP waveforms. To further support a circuit-level distinction, I showed that the two ripple types differentially recruited CA1 and CA3 neurons, giving rise to distinct motifs of millisecond-timescale coactivity and different topological organisations of population activity. I then investigated how current diversity mapped onto functional diversity in reactivation content. Rad^sink ripples integrated recent motifs of waking coactivity, combining superficial and deep CA1 principal cells into denser, higher-dimensional patterns that exhibited stable reactivation throughout sleep. In contrast, LM^sink ripples contained core motifs of prior coactivity, engaging primarily deep cells into sparser, lower-dimensional patterns that gradually drifted over time, updating pre-existing content with recent experience.

Together, these findings suggest that ripple-by-ripple diversity enables parallel reactivation channels that support both the integration of recent experience and the gradual updating of prior representations.

Authors: Castelli, M

• DOI: 10.5287/ora-5jm7ebznq
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: ripples; memory; hippocampus; systems neuroscience
• Subjects: Systems Neuroscience