The influence of learning on sleep slow oscillations and associated spindles and ripples in humans and rats

Abstract

The mechanisms underlying off-line consolidation of memory during sleep are elusive. Learning of hippocampus-dependent tasks increases neocortical slow oscillation synchrony, and thalamocortical spindle and hippocampal ripple activity during subsequent non-rapid eye movement sleep. Slow oscillations representing an oscillation between global neocortical states of increased (up-state) and decreased (down-state) neuronal firing temporally group thalamic spindle and hippocampal ripple activity, which both occur preferentially during slow oscillation up-states. Here we examined whether slow oscillations also group learning-induced increases in spindle and ripple activity, thereby providing time-frames of facilitated hippocampus-to-neocortical information transfer underlying the conversion of temporary into long-term memories. Learning (word-pairs in humans, odor-reward associations in rats) increased slow oscillation up-states and, in humans, shaped the timing of down-states. Slow oscillations grouped spindle and rat ripple activity into up-states under basal conditions. Prior learning produced in humans an increase in spindle activity focused on slow oscillation up-states. In rats, learning induced a distinct increase in spindle and ripple activity that was not synchronized to up-states. Event-correlation histograms indicated an increase in spindle activity with the occurrence of ripples. This increase was prolonged after learning, suggesting a direct temporal tuning between ripples and spindles. The lack of a grouping effect of slow oscillations on learning-induced spindles and ripples in rats, together with the less pronounced effects of learning on slow oscillations, presumably reflects a weaker dependence of odor learning on thalamo-neocortical circuitry. Slow oscillations might provide an effective temporal frame for hippocampus-to-neocortical information transfer only when thalamo-neocortical systems are already critically involved during learning.