Hippocampal gamma oscillations form complex ensembles modulated by behavior and learning

The hippocampus and the entorhinal cortex display a rich oscillatory activity, believed to support neural information processing in key cognitive functions1. In the hippocampal region CA1, a "slow gamma" rhythm (30-80 Hz) generated in CA3 would support memory retrieval whereas a "medium gamma" rhythm (60-120 Hz) generated in the entorhinal cortex would support memory encoding2,3. However, descriptions involving discrete gamma sub-bands can only partially account for the haphazard diversity of oscillatory behaviors observed in individual recordings during spatial navigation behavior. Here, we stress that transient gamma oscillatory episodes at any frequency or phase relative to the ongoing theta (4-12 Hz) rhythm can be recorded at any layer within CA1. Eventually, the commonly reported averages are dominated by a minority of very strong power events overshadowing gamma heterogeneity. Nevertheless, we show that such gamma diversity can be naturally explained by a simple mechanistic model, and that behavior-related information (position within a maze) can be decoded from most individual gamma events, despite their low power and erratic-like nature. Our results indicate that behavior specifically shapes ensembles of irregular hippocampal gamma oscillations, in a way which evolves with learning, depends on the hippocampal layer and is hard to reconcile with the hypothesis of rigid, narrowly tuned gamma sub-bands. Beyond randomness, the pervasive gamma diversity may thus reflect complexity at the "fringe-of-synchrony"4 likely functional but invisible to classic average-based analyses.