نقش خواب برای رمزگذاری حافظه هیجانی

Total sleep deprivation (TSD) has been consistently found to impair encoding of information during ensuing wakefulness, probably through suppressing NonREM (non-rapid eye movement) sleep. However, a possible contribution of missing REM sleep to this encoding impairment after TSD has so far not been systematically examined in humans, although such contribution might be suspected in particular for emotional information. Here, in two separate experiments in young healthy men, we compared effects of TSD and of selective REM sleep deprivation (REMD), relative to respective control conditions of undisturbed sleep, on the subsequent encoding of neutral and emotional pictures. The pictures were presented in conjunction with colored frames to also assess related source memory. REMD was achieved by tones presented contingently upon initial signs of REM sleep. Encoding capabilities were examined in the evening (18:00 h) after the experimental nights, by a picture recognition test right after encoding. TSD significantly decreased both the rate of correctly recognized pictures and of recalled frames associated with the pictures. The TSD effect was robust and translated into an impaired long term memory formation, as it was likewise observed on a second recognition testing one week after the encoding phase. Contrary to our expectation, REMD did not affect encoding in general, or particularly of emotional pictures. Also, REMD did not affect valence ratings of the encoded pictures. However, like TSD, REMD distinctly impaired vigilance at the time of encoding. Altogether, these findings indicate an importance of NonREM rather than REM sleep for the encoding of information that is independent of the emotionality of the materials.

The ability to encode episodic memory information deteriorates after total sleep deprivation (TSD) (Drummond et al., 2000, Polzella, 1975 and Walker and van der Helm, 2009). This effect has been attributed to the loss of non-rapid eye movement sleep (NonREM) rather than REM sleep during TSD before learning, and specifically to functions of slow waves (0.5–4 Hz) and spindles (12–16 Hz) characterizing the electroencephalogram (EEG) during NonREM sleep stage 2 and slow wave sleep (SWS) (Mander et al., 2011 and Van Der Werf et al., 2009). Thus, suppressing slow waves during sleep prior to learning impaired subsequent encoding of pictures of landscapes and buildings (Van Der Werf et al., 2009) whereas enhancing slow waves by electrical stimulation improved subsequent encoding of such stimuli (Antonenko, Diekelmann, Olsen, Born, & Molle, 2013). Also, numbers of spindles during sleep were found to be positively correlated to post-sleep encoding capabilities (face-name associations, (Mander et al., 2011)). On a conceptual level, the benefiting influence of NonREM and SWS has been linked to the <1 Hz slow oscillation as a neurophysiological correlate underlying EEG slow wave activity, in the framework of the “synaptic homeostasis hypothesis” (Tononi and Cirelli, 2003 and Tononi and Cirelli, 2014). This hypothesis assumes that synaptic strength in cortical and hippocampal neural networks globally increases with the encoding of information in wakefulness, and that with prolonged wakefulness these synaptic networks become saturated and, thus, exhibit a reduced capacity to encode further information. Sleep, in particular the slow oscillations of SWS, drive a global re-normalization of synaptic strength and, thus, allows for encoding capacities to be recovered. While the contribution of SWS to recovering encoding capabilities appears to be well-established, there are hints that REM sleep might also play a role in this process. For example, in rats, decreases in hippocampal neuronal activity across triads of NonREM–REM–NonREM periods correlated positively with the amount of theta activity during the intervening REM sleep period, suggesting that REM sleep down-scales synaptic connectivity in hippocampal networks (Born and Feld, 2012 and Grosmark et al., 2012). Also, deprivation of REM sleep compromises subsequent synaptic long-term potentiation in the rat hippocampus (Kim et al., 2005, Lopez et al., 2008 and Ravassard et al., 2015). In particular, REM sleep might affect encoding of emotional stimuli. REM sleep is known to play an important role in the processing of emotional memory (Payne et al., 2012, Wagner et al., 2001, Walker and van der Helm, 2009 and Werner et al., 2015). Humans after selective REM sleep deprivation showed an increased reactivity of emotional brain regions to threatening stimuli (Rosales-Lagarde et al., 2012). On the other hand, the ability to recognize affective facial stimuli was found to be attenuated after a REM-rich sleep (Gujar, McDonald, Nishida, & Walker, 2011). Collectively, these findings suggest an additional contribution of REM sleep to encoding particularly of emotional stimuli. Surprisingly, studies directly comparing the effects of TSD and selective REM sleep deprivation (REMD) to dissociate REM and NonREM-sleep related effects on encoding during ensuing wakefulness are scarce, and to the best of our knowledge, there is no study additionally controlling for the emotionality of the stimuli. Here, in two separate experiments we compared effects of TSD and REMD, relative to undisturbed sleep, on the subsequent encoding of neutral and emotional pictures. The pictures were presented in conjunction with colored frames to also assess related source memory. Encoding capabilities were examined by a picture recognition test right after encoding. In order to assess whether changes in encoding translate into corresponding long-term memory changes, an additional recognition test was performed one week later. We expected that TSD would generally impair encoding whereas REMD would selectively affect encoding of emotional pictures.