Engineering sleep to discover the function of slow wave activity (Commentary on Antonenko et al.)

Authors


During non-rapid eye movement sleep (NREM), the electroencephalogram (EEG) is dominated by low-frequency, high-amplitude oscillations (≈1–4 Hz ‘slow wave activity’ and < 1 Hz ‘slow oscillations’). This synchronous activity has been proposed to play a role in memory consolidation (Diekelmann & Born, 2010) and in the hypothesized process of ‘synaptic homeostasis’ during sleep (Tononi & Cirelli, 2006). Thus far, however, research on the function of slow EEG activity has been largely correlational. A new study by Antonenko et al. (2013) joins several notable exceptions to this rule (e.g. Marshall et al., 2004, 2006; Aeschbach et al., 2008; Landsness et al., 2009; Mednick et al.,2013), reporting that experimentally enhancing slow EEG activity during nap sleep improves the subsequent encoding of declarative information.

During a daytime nap, participants underwent intermittent periods of transcranial direct current stimulation (tDCS) oscillating at 0.75 Hz. Relative to a control group receiving sham stimulation, tDCS substantially increased slow EEG frequencies (0.5–4 Hz) following stimulation intervals. After the nap, participants who underwent tDCS showed enhanced performance on several declarative memory tasks (relative to controls), but not on a procedural motor-learning task.

The putative mechanism of enhanced learning following tDCS is that augmentation of slow wave activity induces plastic changes in the hippocampal system, enabling more effective subsequent encoding of hippocampus-dependent material, and having no effect on hippocampus-independent procedural learning. Specifically, Antonenko et al. (2013) speculate that boosting slow EEG activity induced synaptic downscaling (Tononi & Cirelli, 2006) in hippocampal networks, reducing synaptic strength and enabling more efficient synaptic potentiation following the nap. Although this is one possible scenario, there are certainly others. First, although increased slow EEG activity is a likely mediator of the learning enhancement, tDCS may also have had other effects in parallel. For example, in addition to increasing slow EEG activity following stimulation, tDCS also decreased beta-frequency activity (15–20 Hz) early in the nap. Other possible influences on neural excitability, sleep microarchitecture and network dynamics also cannot be ruled out, any of which could have played a role in the observed behavioral effects. A second outstanding question surrounds the apparently selective effect on hippocampus-dependent memory – although it is possible that cortical tDCS could affect hippocampal networks, the mechanisms that would allow tDCS applied to frontal cortex to selectively affect the medial temporal lobe are unclear.

Although further research will be necessary to concretely establish the mechanisms responsible, this initial study provides strong evidence supporting the hypothesis that brain activity during sleep is critical for subsequent memory encoding. The findings extend those of prior behavioral studies (e.g. Yoo et al., 2007) in several ways. First, Antonenko et al. (2013) demonstrate that direct manipulation of the sleep EEG results in subsequent performance enhancement, even in the absence of sleep architecture differences between groups – the composition of sleep stages during the nap was equivalent between stimulation and sham participants, suggesting that sleep microarchitecture is more important to the encoding effect than the composition of sleep ‘stages’ during the nap. Secondly, by establishing that post-sleep enhancement of encoding was specific to declarative learning tasks, the effects of tDCS here cannot be attributed to a general enhancement of alertness and attention. Here again, the data are most consistent with the notion that experimental augmentation of slow EEG activity directly and causally contributed to subsequent enhancement of declarative memory encoding.

In combination with other work, these observations thus suggest that the slow wave EEG of NREM sleep may serve multiple functions. Prior literature has supported the hypothesis that slow-wave activity supports hippocampal–neocortical communication facilitating consolidation of hippocampus-dependent memory (Diekelmann & Born, 2010). The present data from Antonenko et al. (2013) suggest that, at the same time, slow EEG activity prepares neural networks to continue encoding new information following sleep.

Ancillary