The timing of sleep has been shown to influence sleep's effect on memory, inasmuch as a shorter interval between sleep and learning of declarative tasks enhances the effect of sleep on memory (Gais et al., 2006; Talamini et al., 2008; Payne et al., 2012b). Comparing retention intervals that differed with respect to the timing of sleep, Gais et al. (2006) found that sleep that follows within 3 h after encoding is more beneficial to the consolidation of English-German vocabulary than sleep following after 10 h. Thus, to make the best use of sleep's memory improving effect, sleep should follow encoding within a few hours or, alternatively, the contents that were learned during the day could be rehearsed shortly before bedtime. With regard to the amount of sleep that is necessary to boost memory, several studies have shown that declarative memory even benefits from rather short periods of sleep. 3-h episodes of sleep during the first half of the night are sufficient to prompt the beneficial effect of sleep on declarative memory (Yaroush et al., 1971; Barrett and Ekstrand, 1972; Fowler et al., 1973; Plihal and Born, 1997; Tucker and Fishbein, 2009) and even naps can be used to improve memory, albeit, in a more dose-dependent manner (Lahl et al., 2008). Although some studies found memory improvements of very short naps of only a few minutes (Lahl et al., 2008), longer periods of 60–90 min of sleep provide better outcomes (Diekelmann et al., 2012). This offers the opportunity to time short naps after intensive periods of learning to benefit consolidation and subsequent learning. However, even though short amounts of sleep can help the initial consolidation of new memories, naps cannot replace a good night's sleep but may rather be introduced in addition to nocturnal sleep. Some types of memories even require a whole night of sleep, including the cyclic occurrence of different sleep stages, to become optimally consolidated (Diekelmann and Born, 2010).

Memory consolidation during sleep is selective in preferentially facilitating those memories that are in some way relevant for the future. Wilhelm et al. (2011) showed that merely the information that memory contents will have to be recalled on the next day can decide its access to sleep-dependent consolidation. In this study participants learned a declarative word pair association task in the evening. After encoding, only one of two groups was informed that they would have to retrieve the word pairs the next morning. While the informed group showed the expected improvement in retention performance after sleep, the uninformed group did not and was on par with an informed group that did not sleep during the retention interval. This effect is confirmed by sleep's enhancement of prospective memories of tasks to be performed in the future (Scullin and McDaniel, 2010; Diekelmann et al., 2013a,b). Similar findings were observed when subjects anticipated a monetary reward for performing well in a procedural finger sequence tapping task (Fischer and Born, 2009). Emotionality of the learning content is a further feature that signals relevance for the individual and leads to stronger effects of sleep-dependent memory consolidation (Wagner et al., 2001; Sterpenich et al., 2007, 2009; Payne et al., 2008, 2012a; Groch et al., 2013, 2014). However, there seem to be conflicting mechanisms of REM sleep for consolidating emotional content and SWS for declarative content in some tasks (Payne et al., 2008; Groch et al., 2014). To optimize the beneficial effect of sleep on memory, the relevance of the learning material should be very evident to the learner and might be increased by additional incentives.

Sleep also benefits the learning of declarative tasks such as word pair associations and object locations in children (Wilhelm et al., 2008). A recent study in preschoolers shows that learning new words from storybooks that are read to them is increased in children that are allowed to nap after hearing the story (Williams and Horst, 2014). Even infants of 6–16 months learn new word meanings or novel actions better if they nap during the retention interval (Friedrich et al., 2015; Seehagen et al., 2015). For the extraction of explicit knowledge, sleep might even be more beneficial in children than in adults. Wilhelm and colleagues trained adults and 8–11-year-old children to press a specific sequence of buttons on a button-box (Wilhelm et al., 2013). None of the participants were aware of the underlying sequence of button presses during training before sleep. While both children and adults gained more explicit awareness of the sequence after sleep compared to a wake interval, children clearly outperformed the adults after sleep with almost all children being able to explicitly generate the entire 8-element sequence. Interestingly, this beneficial effect of sleep on knowledge may even extend into highly demanding education such as medical school (Ahrberg et al., 2012; Genzel et al., 2013). In a questionnaire study, sleep behavior (especially during the pre-exam period), together with the results in the pre-clinical board exam, turned out as the strongest predictor for the final grade in the medical exam (Genzel et al., 2013). Improving sleep in children and students might therefore be a promising target to improve school performance (Ribeiro and Stickgold, 2014).

Older adults typically show a dramatic reduction in the amount of slow wave sleep, with this reduction being associated with reduced sleep-dependent memory consolidation (Backhaus et al., 2007). The importance of slow wave sleep for memory consolidation in the elderly was shown in a cross-sectional study (Mander et al., 2013b). In this study young and elderly adults learned a declarative word pair task before going to sleep. Overnight retention was predicted by slow wave activity, which in turn was associated with prefrontal gray matter volume. The authors argue that the age-related reduction in gray matter volume causes the decline of slow wave generation in the brain, which in turn compromises sleep's ability to consolidate memories. In another study from the same group, it was shown that the reduction in sleep spindles observed in the elderly is predictive of impairments of new learning (of an episodic task) and hippocampal activation after sleep (Mander et al., 2013a). Importantly, in these studies it remains unclear if sleep is causal to the observed impairments. Clarifying this question will be essential to identify therapeutic targets for improving sleep and memory in the elderly.

Many memory-related psychiatric disorders, such as post-traumatic stress disorder (PTSD), are associated with altered sleep profiles (Germain, 2013; Steiger et al., 2013), suggesting that sleep might play a role in the etiology of these disorders. Considering that behavioral therapy is effective in many anxiety disorders and relies on learning mechanisms, its success may be further improved by using sleep treatments. Particularly in phobias diminished extinction of conditioned fear responses is often a problem. Sleep benefits the generalization of the extinction of conditioned fear (Pace-Schott et al., 2009) and this finding has been applied to simulated exposure therapy showing that sleep can be used to consolidate and generalize extinction of spider fear responses (Pace-Schott et al., 2012). Recently, Kleim et al. (2013) were able to show that napping for 90 min after an exposure therapy session can reduce self-reported fear and negative cognition at a 1-week follow-up assessment in patients suffering from spider phobia. Hence, applying sleep as a non-invasive method might prove to be a highly useful and inexpensive tool to boost the efficiency of psychotherapy.

Recent research suggests that memory replay can be biased by external cues presented during sleep that had been present during prior encoding (Bendor and Wilson, 2012). In their seminal paper, Rasch et al. (2007) used odor cues to trigger memory reactivation processes during sleep in humans. Participants were asked to encode locations of picture pairs shown on a computer screen (similar to the game concentration, Figure 1 upper panel) while they smelled the scent of roses. During subsequent slow wave sleep participants were re-exposed to this odor or an odorless vehicle. Those participants who had smelled the odor during learning and during slow wave sleep showed increased recall of the card locations at retrieval compared to the vehicle condition. Memory was not improved if participants received the odor only during sleep but not during prior encoding, and odor re-exposure was also not effective during wakefulness and post-learning REM sleep. Since this report, the paradigm has been extended to show that specific auditory cues associated with picture-locations (e.g. the sound “meow” associated with the picture of a cat at a specific location), likewise foster memory reactivation if presented during slow wave sleep, increasing subsequent recall performance (Rudoy et al., 2009; van Dongen et al., 2012). Additionally, it was shown that auditory cues during sleep can enhance the retention of non-declarative tasks by using a melody for cueing a sequence of button presses in a motor skill task (Antony et al., 2012). Interestingly, this effect is highly specific, inasmuch as participants only perform better on that part of the sequence that has been cued during sleep, while performance on the un-cued part is unchanged (Schonauer et al., 2014). Finally, it was demonstrated that the strengthening effect of reactivation cues for memory is specific for the sleep state and that presentation of the same cues during wakefulness can even labilize and disturb the memory trace (Diekelmann et al., 2011). Both odor cues and tones are stimuli that are easy to apply in the home environment to improve individual memory capacities. For example, students might consider using fragrance lights or aromatic lamps during learning and subsequent sleep, or re-play previously learned spoken vocabulary during sleep. Care should be taken, however, to make sure that these stimuli do not disturb sleep per se.

It has now been attempted to translate the memory cueing paradigm to fear conditioning tasks, which opens the possibility of enhancing efficiency of exposure therapy. In one of these studies participants who had undergone olfactory contextual fear conditioning were re-exposed to the olfactory context cue during subsequent slow wave sleep. Compared to the control condition without odor during sleep, these participants showed greater extinction of contextual fear in the next morning (Hauner et al., 2013). Likewise, subjects who were fear conditioned on a neutral tone and were re-presented this tone during slow wave sleep, showed reduced fear responses to the tone after sleep (He et al., 2014). Two similar studies in rodents, however, showed a greater fear response after re-exposure of the conditioned stimulus during sleep (Rolls et al., 2013; Barnes and Wilson, 2014). Although much more research is needed to clarify these controversial findings and to identify possible boundary conditions and adverse effects (for an opinion on the contradictory nature of these results see, Diekelmann and Born, 2015), this method bears great potential for applications in clinical contexts, e.g., for the treatment of anxiety disorders.

Slow oscillations are the brain oscillations that have been most closely linked to sleep-dependent memory processing. Intensifying these slow oscillations with electrical or auditory stimulation has been shown to enhance learning and memory consolidation (Marshall et al., 2006; Ngo et al., 2013b). In the ground-breaking study by Marshall and colleagues, participants learned a word pair association task in the evening. During subsequent sleep they were subjected to trans-cranial electric stimulation (TES) at the same frequency as natural slow oscillations (~0.75 Hz). This stimulation did not only intensify slow oscillations as well as spindle activity, but also increased the efficacy of sleep to consolidate the previously learned word pairs (Marshall et al., 2006). Using the same stimulation at the same frequency (~0.75 Hz) during wakefulness increases theta activity and improves encoding of new learning material, while consolidation remains unchanged (Kirov et al., 2009). Finally, applying this slow oscillation stimulation during a daytime nap improves the encoding of new information afterward (Antonenko et al., 2013), suggesting that slow wave activity during prior sleep frees space for new learning after sleep. Even though some consumer devices for TES have become available (e.g., http://www.foc.us/, http://thebrainstimulator.net), the possibility to affect sleep's slow oscillations by auditory stimulation may offer a more elegant and less invasive approach for the bedroom (Ngo et al., 2013a). Ngo et al. timed short (50 ms) bursts of white noise on the slow oscillation up-states, which increased the amount of consecutive slow oscillations as well as spindle activity coupled to the slow oscillations. When compared to a sham condition, the stimulation improved sleep-dependent consolidation of declarative word pair memory (Ngo et al., 2013b). Such an auditory stimulation of slow oscillations could be applied in the home environment to boost memory performance with relatively little effort. Very recently it has been shown that slow wave sleep can even be enhanced by hypnosis (Cordi et al., 2014, 2015). This method may offer the possibility of enhancing sleep without recurring to invasive and potentially sleep-disturbing techniques.

Many neurotransmitters and hormones that are known to be implicated in learning and memory show characteristic changes across the sleep-wake cycle. It therefore stands to reason to apply certain drugs targeting specific neurotransmitter systems during sleep to enhance memory. GABA A modulators, for example, have long been known to induce sleep and have thus been used extensively to treat sleep disorders. Yet, these drugs may not induce natural sleep but rather light sleep with low amounts of slow wave sleep and REM sleep (Lancel, 1999). The GABA re-uptake inhibitor tiagabine, on the other hand, increases the occurrence of slow wave sleep, but unfortunately decreases sleep spindles and even deteriorates consolidation of a procedural motor skill task (Feld et al., 2013b). The GABA A positive modulator zolpidem has been shown to have a more beneficial profile, as it increases the amount of sleep spindles and improves memory consolidation of word pairs (Kaestner et al., 2013; Mednick et al., 2013). However, other authors do not find this effect or even observed a detrimental effect of zolpidem on declarative memory consolidation during sleep (Melendez et al., 2005; Hall-Porter et al., 2014). Recently it was shown that the NMDA receptor co-agonist d-cycloserine can enhance the consolidation of word pair associations if given before a retention interval containing sleep (Feld et al., 2013a). D-cycloserine has also been discussed as a cognitive enhancer to improve efficacy of cognitive behavior therapy to attenuate anxiety disorders (Otto et al., 2007; Hofmann et al., 2012), hence the application of d-cycloserine during retention sleep after successful exposure sessions may increase the therapeutic value of this drug. Other pharmacological treatments have likewise been shown to affect memory formation during sleep more or less beneficially, such as the noradrenaline reuptake inhibitor reboxetine (Rasch et al., 2009; Gais et al., 2011), the cytokine interleukin-6 (Benedict et al., 2009), and the dopamine D2-like receptor agonist pramipexole (Feld et al., 2014). Generally, drugs that increase sleep and memory are potentially interesting for clinical applications, but should presently not be considered for non-therapeutic sleep and memory enhancement, as the potential adverse effects remain completely unclear.