Dentate spikes and learning: disrupting hippocampal function during memory consolidation can improve pattern separation.

Hippocampal dentate spikes (DSs) are short-duration, large-amplitude fluctuations in hilar local field potentials and take place while resting and sleeping. During DSs, dentate gyrus granule cells increase firing while CA1 pyramidal cells decrease firing. Recent findings suggest DSs play a significant role in memory consolidation after training on a hippocampus-dependent, nonspatial associative learning task. Here, we aimed to find out whether DSs are important in other types of hippocampus-dependent learning tasks as well. To this end, we trained adult male Sprague-Dawley rats in a spatial reference memory task, a fixed interval task, and a pattern separation task. During a rest period immediately after each training session, we either let neural activity to take place as usual, timed electrical stimulation of the ventral hippocampal commissure (vHC) to immediately follow DSs, or applied the vHC stimulation during a random neural state. We found no effect of vHC stimulation on performance in the spatial reference memory task or in the fixed interval task. Surprisingly, vHC stimulation, especially contingent on DSs, improved performance in the pattern separation task. In conclusion, the behavioral relevance of hippocampal processing and DSs seems to depend on the task at hand. It could be that in an intact brain, offline memory consolidation by default involves associating neural representations of temporally separate but related events. In some cases this might be beneficial for adaptive behavior in the future (associative learning), while in other cases it might not (pattern separation). NEW & NOTEWORTHY The behavioral relevance of dentate spikes seems to depend on the learning task at hand. We suggest that dentate spikes are related to associating neural representations of temporally separate but related events within the dentate gyrus. In some cases this might be beneficial for adaptive behavior in the future (associative learning), while in other cases it might not (pattern separation).