Why a ripple and a spindle tend to arrive together
A sharp-wave ripple is a brief, high-frequency burst in the hippocampus, roughly 80 to 150 Hz, thought to replay patterns of activity from recent experience. A sleep spindle is a slower burst, roughly 7 to 15 Hz, generated in the cortex and thalamus. During non-rapid-eye-movement sleep, these two rhythms are not independent: in mice, ripples in the hippocampus reliably coincide with cortical "up-states" and spindles in a network centered on the retrosplenial cortex [5].
The proposed logic is that the ripple carries the hippocampal message while the spindle makes the cortex receptive to it. Recordings from epilepsy patients support this picture: hippocampal ripples were accompanied by widespread increases in high-frequency activity, a marker of local neuronal firing, and the strength of that modulation depended on how tightly the ripple was coupled to cortical slow waves or spindles [1]. In other words, the spindle does not create the memory trace; it appears to help decide which cortical populations are switched on when the ripple arrives [1].
What the coordination actually looks like
The coupling is not a simple one-way command. In the mouse retrosplenial cortex, transient cortical activations and spindles predicted slow gamma oscillations in the hippocampus during ripples, suggesting a two-way exchange rather than the cortex merely listening [5]. The human recordings point the same way: the timing of cortical slow-wave phase and the size of the ripple together predicted how strongly frontal and temporal cortex communicated during a single ripple [1].
This matters for the question of encoding. If a ripple reactivates a memory pattern, the spindle-linked window may be what allows that pattern to be written into cortical networks rather than staying confined to the hippocampus. The human data are consistent with a selection mechanism in which ripple–slow-wave synchronization governs the transient engagement of distributed neuronal populations [1]. That is a proposed function, not a demonstrated one, and the recordings come from patients undergoing intracranial monitoring, which shapes what can be measured.
Does the pairing predict what you remember later?
Here the evidence becomes more indirect. The strongest causal result concerns ripples, not spindles. In mice, a subset of large ripples was linked to memory reactivation in both hippocampus and prefrontal cortex, and these large ripples became more frequent during sleep after new learning. Artificially boosting ripples during sleep enhanced reactivation and improved later memory retrieval, and it also increased hippocampus–prefrontal coordination during waking [3]. A review of that work judged the experimental evidence to strongly support both claims, while noting that the link between reactivation and memory performance was presented as two separate results rather than tested directly [4].
Notably, that ripple-boosting study did not examine spindles at all [3][4]. So it establishes that ripples can causally support retention, but it does not show that spindles are the ingredient that makes the difference. The spindle side of the story rests on correlational coupling in humans and mice [1][5], which is compatible with spindles gating reactivation but does not prove that co-occurrence predicts how much is retained.
When the pairing goes wrong, memory may suffer
One line of evidence comes from epilepsy. Hippocampal interictal epileptic discharges, described as a presumed epileptic exaggeration of sharp-wave ripples, are known to accompany memory problems. In 21 patients with foramen ovale electrodes, scalp spindles that coincided with these discharges were altered: they lasted about 126 milliseconds longer, had higher amplitude by roughly 3.4 microvolts, and shifted toward the higher end of the 13–15 Hz range. Spindles that appeared to be triggered by the discharges, rather than overlapping them, looked normal [2].
The authors propose that this alteration is a marker of a pathological process in which the discharges directly affect spindle generation, and could be a mechanism by which they disrupt memory [2]. That is a hypothesis about a disease state, not a demonstration that ordinary spindle–ripple coupling controls retention in healthy sleep. It does, however, fit the broader pattern: the timing between hippocampal events and cortical spindles seems to matter, and disturbing it is associated with worse memory function [2].
Sources used in this answer
Coupling between slow waves and sharp-wave ripples engages distributed neural activity during sleep in humans
In intracranial recordings from human hippocampus, amygdala, and cortex, hippocampal sharp-wave ripples were accompanied by widespread high-frequency activity, and the strength of that modulation was predicted by coupling between ripples and cortical slow waves or spindles.
Prolongation of cortical sleep spindles during hippocampal interictal epileptiform discharges in epilepsy patients
In 21 epilepsy patients, scalp spindles overlapping hippocampal interictal epileptic discharges were longer, higher in amplitude, and shifted to higher frequencies, whereas spindles apparently triggered by the discharges looked normal.
Large sharp-wave ripples promote hippocampo-cortical memory reactivation and consolidation during sleep
In mice, large sharp-wave ripples were linked to memory reactivation in hippocampus and prefrontal cortex, and optogenetically boosting ripples during sleep enhanced reactivation and improved later memory retrieval.
PREreview of "Large sharp-wave ripples promote hippocampo-cortical memory reactivation and consolidation"
A review of the ripple-boosting study judged its evidence strong for large ripples promoting reactivation and memory, while noting that the link between reactivation and memory performance was not directly tested.
Hippocampal ripples coincide with “up-state” and spindles in retrosplenial cortex
In mice, hippocampal sharp-wave ripples reliably coincided with cortical up-states and spindles in a retrosplenial network, and retrosplenial activations and spindles predicted slow gamma oscillations in CA1 during ripples.