An Update on Recent Advances in Targeted Memory Reactivation During Sleep PDF

npj | science of learning Review article Published in partnership with The University of Queensland...

npj | science of learning Review article Published in partnership with The University of Queensland https://doi.org/10.1038/s41539-024-00244-8 An update on recent advances in targeted memory reactivation during sleep Check for updates Julia Carbone1,2 & Susanne Diekelmann 1,3 Targeted Memory Reactivation (TMR) is a noninvasive tool to manipulate memory consolidation during sleep. TMR builds on the brain’s natural processes of memory reactivation during sleep and aims to facilitate or bias these processes in a certain direction. The basis of this technique is the association of learning content with sensory cues, such as odors or sounds, that are presented during 1234567890():,; 1234567890():,; subsequent sleep to promote memory reactivation. Research on TMR has drastically increased over the last decade with rapid developments. The aim of the present review is to highlight the most recent advances of this research. We focus on effects of TMR on the strengthening of memories in the declarative, procedural and emotional memory domain as well as on ways in which TMR can be used to promote forgetting. We then discuss advanced technical approaches to determine the optimal timing of TMR within the ongoing oscillatory activity of the sleeping brain as well as the specificity of TMR for certain memory contents. We further highlight the specific effects of TMR during REM sleep and in influencing dream content. Finally, we discuss recent evidence for potential applications of TMR for mental health, educational purposes and in the home setting. In conclusion, the last years of research have provided substantial advances in TMR that can guide future endeavors in research and application. Sleep is known to support the consolidation, i.e. the strengthening and While memory reactivation occurs spontaneously in the sleeping integration, of newly acquired memories1. Memory consolidation during brain, it can also be noninvasively reinforced with Targeted Memory sleep is assumed to rely on the covert reactivation of newly encoded memory Reactivation (TMR)6,7. TMR is a well-established technique to selectively traces. This process involves the reprocessing and redistribution of the newly stimulate specific memories to be reactivated during sleep using sensory acquired information, together with its integration in the network of pre- cues linked to prior learning (Fig. 1). In TMR studies, specific cues like odors existing memories. The hippocampus and neocortex are two brain struc- or sounds are presented during learning to become associated with the tures that work in tandem for memory reactivation to occur2,3. New learning material. The same cues, or a subset of these cues, are then re- memories are assumed to be initially bound by the hippocampus for tem- presented during subsequent sleep. Thereby, TMR can be used to directly porary storage. During subsequent sleep, memory traces are reactivated and manipulate memory reactivation and consolidation during sleep. redistributed for more permanent storage in the neocortex. During a night Our goal with this review is to present a brief and accesible overview of of sleep, the human brain undergoes different sleep stages that alternate in a the most recent advances on the research of TMR, thereby giving an outlook cyclic manner. Human sleep is composed of rapid eye movement (REM) on the potential of this tool for the research on basic mechanisms of memory sleep and non-REM (NREM) sleep, which includes light sleep (stages N1 formation as well as for applications in various areas. This review article and N2) and deep sleep, also known as slow wave sleep (SWS, stage N3). builds on a previously published book chapter8 and includes the most recent Unique electrophysiological patterns characterize NREM sleep, most pro- TMR studies from the last 3-4 years following the publication of this chapter. minently neocortical slow oscillations (SO, 0.5–1 Hz), thalamocortical Readers looking for other more systematic reviews and meta-analyses on spindles (9–15 Hz), and hippocampal ripples (80–200 Hz). It is assumed TMR are referred to Lewis and Bendor6 and Hu et al.7. that the precise temporal coordination between these oscillations form the In the following, we will first discuss recent advances in the use of TMR basis for memory reactivation4,5. for different types of memories, such as declarative, procedural and 1 Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, 72076 Tübingen, Germany. 2Graduate Training Centre of Neuroscience, International Max Planck Research School, 72076 Tübingen, Germany. 3Department of Psychiatry and Psychotherapy, University Hospital Tübingen, 72070 Tübingen, Germany. e-mail: [email protected] npj Science of Learning | (2024)9:31 1 https://doi.org/10.1038/s41539-024-00244-8 Review article Fig. 1 | Targeted memory reactivation. In Targeted Memory Reactivation (TMR), learning content is associated with sensory cues, such as sounds, during encoding. During subsequent sleep, the same cues (e.g., sounds) are presented again to facilitate or bias memory consolidation in a certain direction. The effects of TMR on memory can be observed during retrieval (without presentation of the cue). emotional memories. Original TMR studies usually focused on the effect of satellite32. TMR was also effective in enhancing spatial memory, where TMR on strenghening these types of memories9–11. More recent studies participants had to associate words to either the right or left visual field of the explored other aspects of memory via TMR, such as the persistence of TMR screen33. More applied studies proved TMR to be successful, for example, in effects over longer time periods12,13, the role of interference between cues14, learning of a microeconomics class while listening to classical music34, as and the possibility of inducing memory forgetting15,16. What is more, some well as learning English vocabulary26 or a history lesson35 in the presence of a studies combined TMR with other methodologial approches such as virtual specific odor. Thus, TMR enhances memory performance after one night of reality settings17,18, functional magnetic resonamce imaging (fMRI) and sleep in various types of declarative memory, although some recent studies intracranial recordings to investigate specific brain areas activated during did not replicate this effect36. TMR12,13,19,20. Additionally, recent studies combined TMR with closed-loop stimulation to test the best time point for cue delivery in the ongoing Procedural memories oscillatory activity21,22. While most studies in the field applied TMR during Procedural memories consist of the development of perceptual and motor NREM sleep, we will also discuss recent research specifically looking into the skills through extensive practice. Previous studies have tested the effects of effects of TMR during REM sleep and its influence on dream content23. TMR on this kind of memory after a night of sleep with mixed results7. More Finally, we will discuss different applications of TMR for mental health recent studies have looked at the effects of TMR on the persistence of treatments24,25, as well as in educational26 and home settings27,28. procedural memories for the long-term. Rakowska and colleagues12,13 used a In order to improve readability of this review, we have classified the motor memory task that consisted of a bimanual (involving right and left recent studies into different sections. Importantly, some studies are hands) serial reaction time task (SRTT) with 12-item sequences being paired described in more than one section. Table 1 provides an overview of all with high or low pitched tones. One of the sequences was reactivated during studies included in this review with the respective sections they are men- subsequent NREM sleep, while the other was not. Interestingly, while no tioned in. effect of TMR was observed after 24 h, subjects showed better performance for the cued compared to the uncued sequence after 10 days, with this effect Types of memory vanishing after another 6–8 weeks12. In a follow-up study, longitudinal Declarative memory structural and functional MRI showed that activity in the dorsal precuneus Declarative memory refers to memories for facts and events that can be was related to the short-term effects of TMR (over 24 h), whereas sensor- explicitly retrieved29. The effectiveness of TMR has been investigated across imotor regions support the long-term effects over 20 days13. In both studies, various paradigms for declarative memory, demonstrating its versatile spindle density was further found to be higher during the cued compared to application and potential as a memory-enhancing technique. Numerous the un-cued sleep period and higher over the left versus right motor areas for studies have explored tasks involving word pairs, spatial navigation, asso- the cue period. These findings suggest that TMR for procedural memory ciative learning, and other types of memory tasks9,10,30. In many TMR stu- triggers a process that unfolds over the course of several weeks, possibly dies, this learning material is paired with distinct sounds or tones (e.g., the related to spindle activity. picture of a cat with the sound meow), with the sounds or tones then being Another approach uses virtual reality to explore procedural memory presented again during subsequent sleep. In this way, single learning items reactivation during sleep. Picard and colleagues37 applied a virtual envir- can be paired with specific sounds, allowing for a more targeted reactivation onment in which participants were required to fly and accumulate points by of single items during sleep. Other studies include a learning session in the traversing and avoiding floating items. A pleasant melody of four tones was presence of an odor (e.g., the scent of roses)31. When this odor is presented played during the game and used afterwards for either TMR during a again during subsequent sleep, it serves as a more general context cue that morning nap or a resting period. Performance overall improved when TMR reactivates the entire learning session. was applied during REM sleep but not during SWS. Yet, spindle density was A widely used memory task for TMR is the object-location task, where found to be associated with higher performance improvements when TMR subjects have to learn the location of an image coupled with a specific sound. was administered during SWS. Schechtman and colleagues14 showed that in this paradigm the beneficial Another interesting approach uses somatosensory TMR by stimulating effect of TMR is independent of the number of images that are associated the effectors that were involved in a motor task, such as the fingertips38. with a specific sound. Pairing a sound associated with either just one image, Participants learned two sequences of a serial reaction time task (SRTT) in two images or six different images resulted in comparable memory benefits the MRI scanner, and later had a stimulation session during quiet rest where for cued versus un-cued items. Another recent study found that memory for their fingertips received electrical stimulation alternating between the an object-location task can also be effectively enhanced with a device to sequence and random stimulation. After 24 h including a night of sleep at perform TMR at home28. Other than the object-location task, TMR was home, participants showed faster responses for the reactivated compared to found to enhance memory for pairs of semantically related words (e.g., the non-reactivated sequence. MRI data revealed that regions recruited DIET–CREAM)16, when the cue words were re-presented during sleep (e.g., during initial motor learning were reactivated during the TMR session and DIET), as well as for associations between verbs and images22. Another study hippocampo-cortical regions were modulated by the reactivation process. designed “satellite” objects containing similar and different features, However, it remains to be elucidated whether somatosensory TMR during showing that TMR improved memory for features unique to each individual sleep is equally effective for motor learning. npj Science of Learning | (2024)9:31 2 https://doi.org/10.1038/s41539-024-00244-8 Review article Table 1 | Recent TMR studies included in this review Sections Reference Schechtman et al., Commun. Biol. (2021) Declarative memories, Specificity of TMR 14 Whitmore et al., J. Sleep Res. (2022) Declarative memories, TMR for forgetting 28 Schechtman et al., Sci. Rep. (2021) Declarative memories, TMR for forgetting, Optimal timing for TMR 16 Ngo & Staresina, PNAS (2022) Declarative memories, TMR for forgetting 22 Siefert et al., bioRxiv (2023) Declarative memories, Specificity of TMR 32 Bar et al., Curr. Biol. (2020) Declarative memories, Specificity of TMR, Education 33 Gao et al., Neurobiol Learn Mem. (2020) Declarative memories, Specificity of TMR 34 Neumann et al., Sci. Rep. (2020) Declarative memories, Education 26 Vidal et al., Sci. Rep. (2022) Declarative memories, Education 35 Rakowska et al., Neuroimage (2021) Procedural memories, Optimal timing for TMR 12 Rakowska et al., bioRxiv (2022) Procedural memories, Optimal timing for TMR 13 Picard-Deland et al., Learn. Mem. (2021) Procedural memories, TMR during NREM and REM sleep, TMR influencing dream content, Home 37 applications Nicolas et al., J. Sleep Res. (2023) Procedural memories 39 Veldman et al., Cereb. Cortex (2023) Procedural memories 38 Pereira et al., Neuroimage (2022) Emotional memories 19 Legendre et al., bioRxiv (2022) Emotional memories 20 Xia et al., Curr. Biol. (2023) Emotional memories 41 Hutchison et al., Commun. Biol. (2021) Emotional memories, TMR during NREM and REM sleep 42 Yuksel et al., bioRxiv (2023) Emotional memories, TMR during NREM and REM sleep 44 Schechtman et al., Sci. Rep. (2020) TMR for forgetting 15 Whitmore et al., npj Sci. Learn. (2022) TMR for forgetting 46 Whitmore et al., Learn. Mem. (2023) TMR for forgetting 47 Göldi et al., Sci Rep (2019) Optimal timing for TMR 21 Wang et al., J. Sleep Res. (2022) Optimal timing for TMR 49 Shimizu et al., Front. Hum. Neurosci. (2018) Optimal timing for TMR 18 Abdellahi et al., Neuroimage (2023) Optimal timing for TMR 50 Antony et al., Curr. Biol. (2018) Optimal timing for TMR 51 Cairney et al., Curr. Biol. (2018) Optimal timing for TMR 52 Schechtman et al., Cell Rep. (2023) Specificity of TMR 54 Forcato et al., Commun. Biol. (2020) Specificity of TMR 55 Carbone et al., Learn. Mem. (2021) Specificity of TMR 56 Pereira et al., J. Neurosci. (2023) TMR during NREM and REM sleep 58 Abdellahi et al., Neuroimage (2022) TMR during NREM and REM sleep 59 Picard-Deland et al., J. Sleep Res. (2022) TMR during NREM and REM sleep 60 Borghese et al., Front. Psychiatry (2022) TMR during NREM and REM sleep, Mental health,Home applications 17 Abdellahi et al., Elife (2023) TMR during NREM and REM sleep 61 Wick & Rasch, Learn. Mem. (2023) TMR during NREM and REM sleep 36 Carbone et al., bioRxiv (2023) TMR during NREM and REM sleep 57 Salvesen et al., PsyArXiv (2023) TMR influencing dream content 62 Haar Horowitz et al., Cogn. (2020) TMR influencing dream content, Home applications 63 Schwartz et al., Curr. Biol. (2022) TMR influencing dream content, Home applications 23 Talamini et al., Curr. Opin. Behav. Sci. (2020) Mental health 24 van der Heijden et al., Neurosci. Biobehav. Rev. (2022) Mental health 25 Weinhold et al., J. Sleep Res. (2022) Mental health 64 Gvozdanovic et al., Hum. Brain Mapp. (2023) Mental health 65 Chen et al., Curr. Psychol. (2021) Mental health 66 Ziqing et al., bioRxiv (2024) Mental health 67 Göldi et al., npj Sci. Learn. (2019) Home applications 27 Amores et al., Front. Psychol. (2022) Home applications 68 npj Science of Learning | (2024)9:31 3 https://doi.org/10.1038/s41539-024-00244-8 Review article A recent study in older adults observed no significant effect of TMR in location of emotional pictures on a grid associated with specific sounds used improving motor memories39. Electrophysiological analyses suggested that for subsequent cueing during sleep44. Unexpectedly, reactivation of emo- responses to cueing are preserved in the ageing brain, however, the ability to tional stimuli was enhanced when cueing occurred during SWS but differentiate between relevant and irrelevant stimuli seems to be impaired. impaired during REM sleep. Thus, the effects of TMR for procedural memories may depend on certain Thus, recent studies provide behavioral and physiological evidence for characteristics of the type of task, the time of testing, the presence of certain differential effects of TMR on emotional memories. Depending on the sleep parameters as well as the brain’s ability to detect relevant stimuli. procedure, TMR may either strengthen emotional memories or reduce the affective tone. Emotional memories The effects of TMR for emotional memories and their neural correlates are TMR for forgetting still not well investigated. Non-invasive electroencephalography (EEG) has Traditionally, TMR has been used to strengthen memories and enhance limited access to deep brain regions implicated in emotional memories such memory performance. However, recent studies have begun to explore the as the amygdala and orbitofrontal cortex (OFC). Two recent studies com- intriguing possibility of using TMR to selectively weaken memories during bined TMR with MRI and intracranial recordings, respectively, to better sleep. This line of research is particularly relevant for clinical applications understand emotional memory reactivation during sleep. and the treatment of disorders associated with traumatic memories. The Pereira and colleagues trained participants on an arousal rating task, findings on emotional memory updating and the reduction of affective tone where they were asked to rate picture-sound pairs from neutral to more by TMR, as reported above, suggest that TMR could indeed be used to negative19. The same picture-sound pairs were then applied in a modified facilitate targeted forgetting. version of an object-location task. In that task, it was previously found that Simon and collaborators introduced the idea of using TMR to SWS duration and spindles predict faster memory judgments for negative, promote forgetting in a study, in which participants were first trained to but not neutral, picture locations after TMR40. Pereira and colleagues associate a tone to the act of forgetting, and subsequently learned object- obtained fMRI recordings during testing after a night of sleep with TMR19, sound-location pairings45. During SWS, object-sounds were presented showing that TMR influences changes in brain activity in the amygdala and paired to the forget-tone. One week later, participants recalled sig- OFC. In particular, TMR modulated OFC activity in a valence-specific nificantly fewer objects that were paired with the forget tone. More recent manner: cueing neutral items increased OFC activity, while cueing negative studies by Schechtman and colleagues continued exploring the possibi- items decreased it. Moreover, the effect of cueing on amygdala activation lity of inducing forgetting with TMR. In one study, they designed a task in was greater when more time was spent in REM sleep. Regarding TMR, which participants were instructed to memorize the location of some surprisingly there was no cueing effect, neither for the negative nor neutral images while the locations of other images should be forgotten stimuli. intentionally15. Whether or not a specific image should be remembered In a study by Legendre and colleagues, intracranial recordings were or forgotten was signaled by different sound cues. During a subsequent obtained from epileptic patients in brain regions implicated in emotional afternoon nap, sound cues used to signal forgetting were presented again. memories20. Patients were shown emotionally relevant (i.e., humorous) vs. After sleep, memory performance was reduced for the to-be-forgotten emotionally neutral pictures, paired with different tones. When the tones images associated with the presented cues relative to those cues not were presented again during a subsequent nap, participants showed better presented during sleep. In a subsequent study, participants learned pairs memory for humorous pictures than neutral ones. Intracranial recordings of semantically related words (e.g., DIET–CREAM) and were then revealed that the tones associated to humorous pictures enhanced SO and exposed to cue words (e.g., DIET)16. In a Think-no-think paradigm, spindle activity in the OFC, paralleled with an increase in theta connectivity participants were instructed to either recall (“think”) or suppress between the hippocampus and the OFC. Additionally, humorous pictures (“no‑think”) the associated word. The findings confirmed that sup- induced a similar time course of theta power than neutral pictures in all pression impaired retention of the corresponding target word (e.g., conditions except for the sleep group at testing, suggesting that TMR during CREAM) when tested immediately after a 90‑min nap. However, TMR the nap changed the memory traces of emotional pictures. with one of two sounds conveying suppression instructions during sleep In a different line of research, TMR was applied for updating emotional did not enhance suppression‑induced forgetting. memories, in attempt to test whether this technique can be used to modify Interestingly, another series of studies explored the effect of TMR on unwanted memories, such as traumatic experiences41. Subjects learned memory when TMR cues disrupted sleep, suggesting that this procedure can associations between spoken pseudowords (e.g., GuXu) and negative images lead to selective forgetting28,46,47. In one of these studies, Whitmore and (e.g., a monster). During subsequent sleep, the aversive pseudowords were colleagues trained subjects on object-sound associations and subsequently presented again together with either positive cues (e.g., cheering) or neutral manipulated the intensity of the sound cues during sleep. They observed cues. In the next morning, cues were rated as less aversive when they were that reactivation with sound cues that induced sleep arousals selectively paired with positive words during sleep, indicating that TMR changed the weakened the associated memories47. Thus, the possibility of forgetting with affective tone of unpleasant memories and made them less unpleasant. EEG TMR should be investigated further in future studies. analyses showed that theta power predicted affective updating, with larger theta power for positive words during cueing being associated with greater Optimal timing for TMR affective updating. Considering mixed findings on the effectiveness of TMR, a more recent In an attempt to reduce emotional responses during sleep, Hutchison approach suggests that TMR effects depend on the time point of cue and colleagues42 performed TMR during REM sleep and SWS. Based on the application in relation to the ongoing brain oscillatory activity. This kind of sleep to forget, sleep to remember (SFSR) hypothesis, proposed by Walker manipulation is usually referred to as closed-loop or real-time TMR. Closed- and collaborators43, it was assumed that the dissipation of emotional charge loop stimulation during sleep was originally developed to enhance endo- relies on memory reactivation during REM sleep. Participants rated emo- genous sleep rhythms (i.e., SO) by presenting sound clicks at a specific phase tionally negative and neutral pairs of pictures and sounds for arousal both of the SO. The original study by Ngo and colleagues found that sound before and after a night of sleep, with half of the negative and half of the presentation in the up-state (or down-to-up transition) of the SO enhances neutral sounds being played again either during REM sleep or SWS. As the ongoing SO amplitude, increases associated spindle activity, induces expected, TMR during REM sleep but not during SWS resulted in a stronger another subsequent SO and improves sleep-dependent memory decrease of arousal ratings for negative pictures, which was driven by the consolidation48. These findings suggest that the SO up-state might represent largest decreases for the most negative stimuli. In another recent study on a particularly sensitive time window for the occurrence and induction of emotional memory, Yuksel and colleagues had participants learn the memory reactivation. npj Science of Learning | (2024)9:31 4 https://doi.org/10.1038/s41539-024-00244-8 Review article By combining closed-loop stimulation with TMR, recent studies stronger spindle power during the subsequent SO cycle following up-state explored the effects of precise cue presentation in the ongoing SO. Göldi and cueing compared to down-state cueing22. Up-state cueing was even found to colleagues examined this question in a study where participants learned a elicit reinstatement of target representations during the SO after sound Dutch-German vocabulary task21. They were then presented again with a offset, suggesting that up-state cueing induced reactivation of the associated selection of the Dutch words during subsequent NREM sleep, with the memory representations. words being either presented during SO up-states, down-states or were not Two other studies specifically targeted the up-to-down vs. down-to-up cued at all. Interestingly, only memory cues presented in the SO up-states transition of SO for TMR application. Shimizu and colleagues applied TMR improved subsequent recall performance, whereas down-state cueing did time-locked to the down-to-up-state transition of SO after participants not result in a clear benefit when compared to the un-cued control condi- navigated different routes in a virtual reality task in the presence of different tion. However, up-state cueing did not lead to superior performance when sounds18. When sounds were presented in the down-to-up transition of compared with down-state cueing directly. This finding is in line with ongoing SOs, average navigation time improved on the first day after another study by Wang and colleagues49, in which participants learned pairs learning. Moreover, fast spindle activity was increased in the down-to-up of words, with the first syllable of each pair being subsequently presented transitions compared with un-cued sleep. The second study by Abdellahi either in the up-state or down-state of endogenous SO. No differences were and colleagues directly compared the presentation of sounds associated with found in memory performance between up-state and down-state cueing. a procedural memory task (12-item SRTT as described above) during dif- Another recent study by Ngo and Staresina used a larger set of stimuli, ferent phases of the ongoing SO50. Using a machine learning pipeline to including associations between 120 verbs and a set of 6 repeating images predict the optimal time for TMR stimulation, they found that cues, which with associated sounds (e.g., car and engine starting)22. The 6 images and fall within the up-going transition of the SO, are more likely to elicit a sounds were classified into 3 categories: objects, scenes and body parts. classifiable reactivation. During NREM sleep, two sounds corresponding to two different categories Additionally, recent findings suggest that optimal timing for TMR also were presented either during SO up-states or down-states, with a novel depends on the timing of cueing in relation to sleep spindles. Antony and sound being used as control during both up- and down-states. Using this collaborators showed in a first step that there is a refractory period between protocol, behavioral results revealed reduced overnight forgetting when spindles of about 3–6 s51. Using real-time spindle tracking, the authors then cueing occurred during the SO up-state compared to the down-state presented TMR cues either within or right after this refractory period and (Fig. 2a). found better memory improvement when cues were presented after the EEG analyses from the previously mentioned studies revealed mixed refractory period. Furthermore, a study by Cairney and colleagues showed findings. While Wang et al. observed no differences between up-state and that it is possible to decode the content of memory reactivation based on the down-state cueing49, Göldi et al. found that successful TMR during SO up- TMR-evoked spindle activity, providing further evidence for the role of states was related to higher theta and spindle band activity than non- spindles in memory reactivation52. For a framework of the assumed memory successful TMR, which was not evident for the down-state condition21. Ngo function of spindle oscillations, the reader is referred to Antony and et al. observed that up-state cueing resulted in a prolonged SO up-state and colleagues53. Together, these findings suggest that the timing of cue Fig. 2 | Important recent findings on TMR. a Optimal timing for TMR. Ngo & compared to uncued words (gray). Reprinted from Curr. Biol. 30, Bar et al. Local Staresina delivered sound cues at slow oscillation (SO) up-states (UP-stim) and Targeted Memory Reactivation in Human Sleep, 1435-1446.e5, Copyright (2020), down-states (DOWN-stim) compared to respective sham conditions. The upper with permission from Elsevier. c REM sleep TMR and dreaming. Picard-Deland and panel shows the mean amplitude for SO in the UP-stim (green) and DOWN-stim colleagues tested whether TMR and dream reactivations benefit whole-body pro- condition, indicating the cueing (Cue, light orange) and post-cue periods (Post-cue, cedural learning. Participants learned how to fly in a virtual reality setting in the dark orange). Lower panel: Presenting the sound cues during SO up-states resulted presence of a melody, which was subsequently played again during a nap in either in signficantly better memory performance for previously learned sound-word- NREM sleep (STIM-NREM) or REM sleep (STIM_REM) compared to a control nap image associations (UP, green) than when cues were presented in the SO down-state without TMR (CTL-nap). Performance improved particularly for those participants (DOWN, red). Reprinted from Proc. Natl. Acad. Sci. 119, Ngo & Staresina. Shaping who received TMR during REM sleep and reported dreaming of the task (Task- overnight consolidation via slow-oscillation closed-loop targeted memory reacti- dream reactivation). Reprinted from Neurobiol. Learn. Mem. 38, Picard-Deland vation, e2123428119, Copyright (2022). b Unilateral odor TMR. Bar and colleagues et al. Whole-body procedural learning benefits from targeted memory reactivation designed a TMR approach to stimulate one brain hemisphere during sleep. Beha- in REM sleep and task-related dreaming, 107460, Copyright (2021), with permission vioral results for a declarative spatial memory task show that memory was selectively from Elsevier. improved for words processed in the cued hemisphere (cued words, purple) npj Science of Learning | (2024)9:31 5 https://doi.org/10.1038/s41539-024-00244-8 Review article presentation in relation to the ongoing SO and spindle activity is critical for reminders stabilized memories. The authors propose that only incomplete TMR effects, which should be considered in future TMR studies. reminders initiate long-term memory stabilization. In a subsequent phar- macological study, they modulated the GABAergic system via Zolpidem Specificity of TMR administration to evaluate changes in sleep oscillations and their functional In research on TMR, it is still an open question to what extent cueing can relation to memory reactivation56. Using the same memory paradigm, they specifically target single memories or certain aspects of a memory and presented incomplete reminders and found that TMR with Zolpidem sig- whether cueing of one memory may interfere with the consolidation of nificantly enhanced memory performance when compared to placebo. other memories. Moreover, the underlying mechanisms behind TMR, such Furthermore, Zolpidem increased the coupling of fast spindles and theta to as the structure of the cue presented during TMR as well as the relationship slow oscillations, suggesting that GABAergic activity may be functionally of TMR with certain neuromodulators are still a matter of debate. A study by implicated in memory reactivation processes during sleep. Bar and colleagues tested this question, taking advantage of the unique ipsilateral neuroanatomy of the olfactory system to perform TMR locally33. TMR during NREM and REM sleep Participants learned a spatial memory task, i.e., associating words to either Most TMR studies have presented cues during NREM sleep, especially the right or left visual field of the screen, which elicited lateralized EEG during SWS. However, it remains a matter of debate whether different markers of one brain hemisphere. During learning, bilateral stimulation of NREM sleep stages may play different roles for TMR. Two recent studies the olfactory system with an odor was delivered through a nasal mask directly targeted this question, comparing TMR during SWS and sleep stage covering both nostrils. When the odor was administered again specifically to N236,57. Wick and Rasch implemented a vocabulary learning task that had one nostril during a subsequent nap, memory was selectively improved for already been used in previous studies from their group27,30. They observed no words processed in the cued hemisphere (Fig. 2b). Time frequency analyses differences in recall performance between words reactivated during SWS during TMR revealed spindle power and slow wave activity power increases and N236. Furthermore, they failed to find an overall memory benefit of during odor stimulation. Slow wave power (0.5–4 Hz) was lower in the cued TMR over non-reactivated cues, questioning the robustness of TMR ben- hemisphere and correlated negatively with memory improvement for cued efits. Similarly, Carbone and colleagues applied a sound-syllable-word words. Interestingly, SO-spindle coupling changed locally such that spindles memory paradigm that had been tested in previous studies55,56. Again, in the cued hemisphere peaked later and closer to the SO peak. memory performance was not significantly different between cues pre- Another study tested whether multiple memories can be reactivated sented during SWS and N257. Interestingly, Wick and Rasch found that simultaneously or whether they would interfere with each other14. Partici- words played during N2 elicited stronger characteristic oscillatory responses pants learned an object location task, where they had to memorize the when compared with SWS, whereas Carbone et al. observed higher evoked- location of an image coupled with a specific sound. Images were classified to response potential amplitudes for cues presented during SWS, and a higher either small (1 item), medium (2 item) or large categories (6 items) (e.g., 6 density of SO and SO-spindle complexes during SWS compared to N2. different cat pictures in different locations were associated to the same meow The effects of TMR during REM sleep have rarely been investigated. A sound). For those categories that were reactivated during subsequent sleep, few recent studies directly focused on TMR during REM sleep, when memory performance was improved independent of the number of items compared to NREM sleep stages. The study by Hutchison and colleagues (as per category. EEG analyses showed spindle and theta power increases upon mentioned in the section on emotional memory) compared the effects of reactivation, with the highest increase for sounds related to six items. Thus, it TMR during REM sleep and SWS for the reduction of affective tone in an seems to be possible to reactivate multiple memories independently at the emotional memory paradigm42. Considering that REM sleep has been same time, suggesting that reactivation occurs in a simultaneous and pro- proposed to play a specific role in the consolidation of emotional memories miscuous manner. by reducing the affective tone of these memories43, the authors hypothesized Another study tested the question whether memory reactivation drives that TMR during REM sleep, but not SWS, would result in reduced arousal memory transformation32. Siefert and colleagues applied an object category ratings for negative pictures. As expected, cueing the negative pictures learning paradigm, where participants learned three categories of “satellite” during REM sleep reduced the arousal, while cueing during SWS did not. objects before taking a 90 min nap. Each satellite had unique features as well This effect was driven by the most negative stimuli. as shared features with other members of its category. Moreover, each object Another study focused on the role of TMR for rule abstraction, in had a specific name that was used for subsequent TMR. Interestingly, TMR particular during REM sleep and SWS58. The experiment consisted of improved memory for features unique to each individual satellite, while it pairing abstraction problems with sounds to be used for TMR during the impaired memory for features shared across satellites in the same category. different sleep stages. The results revealed an improvement on abstraction The question whether reactivation of some memories may also affect problems for REM cues, but not for SWS cues, at the one week follow-up other related memories, was recently tested in a study on TMR and context test. The authors propose that a sequence of plasticity events is initiated reinstatement54. Participants were asked to invent idiosyncratic stories for during REM sleep that requires more time to be completed. Focusing on different contextually bound sets, each set containing one image of a place emotional memory consolidation, Yuksel and colleagues compared TMR and four objects. Subsequently, the objects were presented in different during SWS and REM sleep44. Contrary to the authors expectations, reac- positions on a 2D grid associated to an object-related sound, while parti- tivation of emotional stimuli was enhanced during SWS but impaired cipants were asked questions related to the objects present in their story to during REM sleep. keep the encoding context salient. Afterwards, participants had a nap and Another study provided evidence for wake-like memory reactivation some of the sounds corresponding to a contextual set were presented via during REM sleep after TMR59. Machine learning classifiers were used to TMR. The results suggest that memory recall of non-cued objects was identify the reinstatement of wake-like memory patterns during REM sleep affected by contextually linked cued objects. The authors propose that after participants were trained on a procedural SRTT task. TMR during individual memory reactivation during sleep may impact consolidation of REM sleep indeed elicited detectable reactivation, which was mediated by memories sharing an associated encoding context. high theta activity and was partly temporally compressed and partly dilated Another study tested the effect of TMR when either complete or compared to the wake experience. Although TMR during REM sleep did not incomplete reminders are presented55. Participants learned the association directly improve SRTT performance, the amount of reactivation observed between sounds and words, and where then presented with either the sound during REM sleep predicted overnight improvement on the SRTT. +word (i.e., complete reminders) or the sound+first syllable of the word Picard and colleagues published two subsequent studies, showing that (i.e., incomplete reminders). When participants slept for 40-min, both TMR during REM sleep benefits procedural skill memory consolidation37,60. reminders were equally effective in stabilizing memories. However, when Participants completed a virtual reality (VR) flying task prior to and fol- the sleep period was extended to 8 hours, only incomplete but not complete lowing a morning nap, during which tones related to the VR task were npj Science of Learning | (2024)9:31 6 https://doi.org/10.1038/s41539-024-00244-8 Review article replayed in either SWS, REM sleep or wakefulness. The findings indicate PTSD. In the following, we will only highlight the most recent studies not that VR performance benefits most from TMR when applied during included in the aforementioned reviews. REM sleep. A clinical study included stable and medicated schizophrenia patients Another applied TMR study delivered cues during REM sleep to treat in a double-blind sham-controlled TMR design64. Memory performance social anxiety disorder17. The findings indicate an association between the was assessed by a verbal (word pairs) and non-verbal (complex figure) number of auditory cues during REM sleep and a parasympathetic mea- memory task. TMR evoked an electrophysiological response similar to that surement of anxiety level (for more details see section on mental health). in healthy participants, increasing SO and spindle coupling during stimu- Thus, TMR during REM sleep may be effective for specific types of memory, lation. However, there was no memory performance improvement mostly emotional memories and certain types of procedural memories. A after sleep. recent study by Abdellahi and colleagues even provides evidence that TMR A study by Borghese and colleagues focused on TMR to treat social in REM sleep can elicit detectable reactivation, showing similarities with anxiety17. Participants with social anxiety had two exposure therapy sessions reactivation during NREM sleep stages61. where they had to give a public talk in front of a jury in a virtual reality setting. At the end of the session, half of the participants (TMR group) TMR influencing dream content received positive feedback paired with a sound, which was subsequently A recent research area combines TMR methods with dream content studies presented again for one week during sleep at home. Although TMR did not to investigate the impact of external stimuli on dreams (for a systematic affect subjective measures of anxiety when compared to a control group, the recent review on this topic refer to62). This line of research may eventually number of auditory cues during REM sleep as well as REM sleep duration provide insights into the question whether and how memory reactivation was associated with a parasympathetic measurement of anxiety level (i.e., during sleep is related to ongoing dreaming. root mean square of successive differences between normal heartbeats), Two studies by Picard and colleagues (see above) combined TMR with suggesting that more TMR cues and longer duration of REM sleep reduced a VR task in which participants were trained to fly in a virtual environment. the physiological anxiety reactions. The authors then explored procedural memory benefits and their relation to Another study examined the effects of sleep and TMR on traumatic dream content37,60. REM sleep dream content related to the VR task (e.g., memory development65. Participants were presented with an odor while flying, accelerating) was found to be associated with better performance on watching a trauma film. The same odor was then presented again during the VR task (Fig. 2c). Moreover, TMR had a delayed effect on dream content subsequent SWS to facilitate memory integration. The findings show that that depends on the sleep stage. Participants dreamed more about the task sleep per se reduced the number of intrusive traumatic memories compared 1–2 days later when TMR was applied in REM sleep and 5–6 days later when to wakefulness, however, there was no additional benefit of TMR. it was applied in SWS, compared to participants with no TMR. Another line of research investigated TMR in relation to positive self- A group of researchers has recently developed a wearable electronic images, considering that the self-concept can affect mental health in various device for dream incubation called ‘Dormio’63. This device makes use of the ways. Chen and colleagues combined a self-esteem training with audio cues, basic idea of TMR to present auditory information during a hypnagogic which were then re-presented during a subsequent 90-min nap66. Results period, with the purpose of incorporating the auditory information into the showed that TMR increased individuals’ implicit self-esteem level, which dream content, a process called ‘targeted dream incubation’ (TDI). This tool was still maintained one week later. Ziqing and colleagues tested whether can easily be applied in a home-setting but does not include EEG recordings. TMR can enhance positive self-evaluative memories67. Participants first had In a pilot experiment with this device, participants received instructions to to rate positive and negative personality traits in relation to themselves. They think of a tree, whereas the control condition was asked to observe their then performed a Go/NoGo task, in which they were presented with positive thoughts. Subsequently, every time the device detected that the participant personality traits (Go traits, e.g., brave, kind) that required them to press a was falling asleep, it asked them to report verbally what was going through button as quickly as possible. When half of the positive traits were presented their mind and recorded their response. It was found that 67% of the again during NREM sleep in a subsequent nap, the recall of positive traits awakenings in the experimental group incorporated the word ‘tree’. that were strongly memorized before sleep was increased, suggesting that Finally, a recent study applied TMR to influence the content of pre-TMR self-evaluative memory strength modulated TMR benefits. The nightmares23. In an imagery rehearsal therapy (IRT) session, participants strongest memories thereby elicited significantly larger sigma power were instructed to consciously change the negative storyline of their current changes. Thus, TMR may be applicable in changing memories that are nightmares, which took place in the presence of a sound. Participants directly or indirectly linked with mental health. performed this IRT session in the evening at home and the sound was then re-presented during subsequent REM sleep. After two weeks of IRT with Education TMR participants reported less frequent nightmares and more positive In the last few years, several studies tested the applicability of TMR in a dream emotions than the control group. The decrease of nightmares was school setting to improve learning in students. Neumann and colleagues sustained after three months, suggesting that TMR may also be applicable to applied TMR with an odor in the class room of a 6th grade26. Students were treat certain mental disorders. instructed to study German-English vocabulary at home at their desk in the presence of an incense stick and to put the stick next to their bed at night Applications while sleeping. This procedure resulted in superior performance in a sub- In the last section we will introduce some of the most recent TMR appli- sequent vocabulary test when compared to students who did not receive the cations. First, we will review a group of studies that explores TMR as a odor TMR. Effect sizes in the TMR groups were quite large (d = 0.6–1.2), possible tool to treat mental disorders or to improve mental health. Next, we indicating that TMR does not only work in a lab environment but also in a will describe recent studies using TMR for educational purposes. Finally, we regular school setting. The authors also showed that TMR works effectively present studies introducing the possibility of using TMR at home outside of when presented continuously during the entire night rather than only laboratory settings. during certain sleep stages. Using a similar study design, Vidal and colleagues applied TMR in Mental health secondary school students who had a history lesson in the presence of an Two recent review articles have discussed potential applications of TMR for odor dispenser35. When students were presented with the same odor during traumatic memories25 and affective disorders24. The first review explored the the night at home, compared to a different odor in the control condition, potential benefits of TMR for the treatment of Post-Traumatic Stress Dis- memory performance of the history class was significantly improved. In order (PTSD), while the second one proposes TMR as a sleep-based treat- another study, undergraduate students had a microeconomics class while ment for maladaptive emotional memories such as in phobias, addictions or listening to classical music34. During subsequent SWS, students were re- npj Science of Learning | (2024)9:31 7 https://doi.org/10.1038/s41539-024-00244-8 Review article presented with either the classical music (TMR) or a control noise, showing declarative and procedural memories. There is also increasing evidence for that TMR increased the probability of passing the test with a grade of 70 or the potential of TMR to change emotional memories. More recent TMR more. Interestingly, this benefit was only seen in female students but not in studies raised other interesting questions, such as the role of different sleep males, and the benefit was lost at a 9-month follow-up test. EEG analyses stages, the importance of considering interference between cues, or even the found that frontal theta activity was higher during SWS for the TMR group possibility of using TMR for memory forgetting. Auditory cueing studies when compared to the control condition and greater theta activity was gained popularity by opening the opportunity to consider the best time associated with better subsequent test performance as well as with less point for stimulation in the ongoing oscillatory activity (i.e., closed-loop forgetting after 9 months in females. These findings provide convincing TMR), suggesting the up-state or down-to-up state transition of the SO as evidence for beneficial effects of TMR in educational settings. well as the periods outside of the spindle refractory period to be the most promising time windows for cue delivery. Meanwhile, odor cueing remains Home applications resourcefull as it can be easily transferred to more realistic scenarios. The last In the previous paragraphs, we have already mentioned a few studies that years have seen an upsurge of research on meaningful applications of TMR, successfully applied TMR in a home setting. These studies were possible due confirming applicability in educational and clinical settings as well as to the recent development of devices for the application of TMR outside the at home. lab. Borghese et al. applied TMR at home in patients with social anxiety Despite the fast advances in this field, there are still a number of open using a wearable headband device that automatically identified sleep stages questions that need to be addressed in future research. One of the most and administered the TMR sounds during REM sleep17. They made use of interesting directions for future research will be the role and interaction of an already existing headband designed by the company Dreem (Dreem SAS, different sleep stages for memory reactivation. Particularly, the role of reac- Paris, https://dreem.com/), where sounds can be delivered via bone- tivation during REM sleep and its connection with emotional memories and conduction transducers integrated in the headband. Schwartz and collea- dreaming should be further explored. Another open question relates to gues likewise used the Dreem headband to treat patients with nightmare potential boundary conditions and critical factors determining whether TMR disorder across a 2-week period at home23. Two more studies effectively is effective or not, such as the timing of cue presentation or the number of items applied TMR with odors at home using incense sticks27 and odor associated with single cues. Finally, the field would benefit from more research dispensers35 without any EEG monitoring to present learning-associated on potential applications of TMR in various settings to foster the transfer of odors during sleep at home in school students. promising basic research findings to much-needed real-life applications. More recent studies have tested potential boundary conditions for However, TMR research also faces important limitations. Several home-applications of TMR. Göldi and Rasch performed unsupervised TMR studies have recently failed to find a significant difference between reacti- at home to improve foreign vocabulary learning27. During three consecutive vated and non-reactivated memories39,69. Future research should take nights, participants used an mp3-player to play previously learned foreign greater efforts to replicate important findings in the field. Moreover, most vocabulary during sleep, without any control of sleep stages or awakenings. published TMR studies include rather small sample sizes, challenging the Per night analyses revealed that memory benefits by TMR were significant possibility of comparing published work. Future TMR studies should only in the third night, indicating that sleep disturbances and habituation to include larger sample sizes and multi-lab studies of established paradigms, the sounds might be critical factors for the success of unsupervised TMR in a ideally with preregistration. Additionally, transferring laboratory experi- home setting. ments into real-life situations, such as educational settings, has some lim- Several groups of researchers have developed different devices for itations. For example, unsupervised TMR sessions at home may lead to sleep applications of TMR outside the lab. One previously mentioned device, i.e., arousals or awakenings (e.g., when sounds are too loud), which may hinder ‘Dormio’, was developed by Horowitz and colleagues. Dormio is a wearable memory improvement or even weaken the associated memories. These device for dream incubation designed to be used at home63. It consists of a limitations need to be considered when assessing TMR effects. hand-worn sleep tracker and an associated app used to communicate with users and record dream reports. Reporting summary Whitmore and collaborators have developed ‘Sleepstim’, a system to Further information on research design is available in the Nature Research perform TMR at home46, including a smartwatch to collect movement and Reporting Summary linked to this article. heart-rate data, plus a smartphone to emit auditory cues. A machine- learning model is used to identify periods of deep sleep to trigger TMR Received: 21 September 2023; Accepted: 4 April 2024; sounds. In two experiments using SleepStim, the spatial memory benefit of in-laboratory TMR studies was tested. Participants learned object-locations associated to a sound and half of the objects were reactivated during three References subsequent nights. The results showed that the memory benefit of TMR was 1. Rasch, B. & Born, J. About sleep’s role in memory. Physiol. Rev. 93, dependent on sound intensity and sleep stage, with greater improvements 681–766 (2013). when sound intensity was lower and when cues were delivered in SWS. 2. Diekelmann, S. & Born, J. The memory function of sleep. Nat. Rev. Another research group has developed a device for odor TMR at Neurosci. 11, 114–126 (2010). home68. ‘Ezzence’ is an olfactometer for home settings that can be controlled 3. Watson, B. O. & Buzsáki, G. Sleep, memory & brain rhythms. Daedalus with the smartphone to release odors in real-time during specific sleep 144, 67–82 (2015). stages. Feasibility of the device was tested in a study where half of the 4. Mölle, M., Eschenko, O., Gais, S., Sara, S. J. & Born, J. The influence of participants slept at home in the presence of a lavender scent compared to a learning on sleep slow oscillations and associated spindles and control condition without any interventions. Results showed that partici- ripples in humans and rats. Eur. J. Neurosci. 29, 1071–1081 (2009). pants were satisfied with the device and found it easy to use. Moreover, 5. Staresina, B. P. et al. Hierarchical nesting of slow oscillations, spindles participants reported better sleep as well as better mood in the next morning. and ripples in the human hippocampus during sleep. Nat. Neurosci. Future studies will have to test the applicability of this device for TMR. 18, 1679–1686 (2015). 6. Lewis, P. A. & Bendor, D. How targeted memory reactivation Conclusions and limitations promotes the selective strengthening of memories in sleep. Curr. Biol. The last 3-4 years of TMR research have provided substantial advances in 29, R906–R912 (2019). understanding the effects, mechanisms and applicability of TMR as a tool to 7. Hu, X., Cheng, L. Y., Chiu, M. H. & Paller, K. A. Promoting memory influence memory consolidation during sleep. Memory strengthening consolidation during sleep: A meta-analysis of targeted memory effects continue to be found across multiple memory systems, such as reactivation. Psychol. Bull. 146, 218–244 (2020). npj Science of Learning | (2024)9:31 8 https://doi.org/10.1038/s41539-024-00244-8 Review article 8. Klinzing, J. G. & Diekelmann, S. Chapter 31 - Cued Memory 30. Schreiner, T. & Rasch, B. Boosting vocabulary learning by verbal Reactivation: A Tool to Manipulate Memory Consolidation During cueing during sleep. Cereb. Cortex 25, 4169–4179 (2015). Sleep. in Handbook of Sleep Research (ed. Dringenberg, H.) 30, 31. Diekelmann, S., Biggel, S., Rasch, B. & Born, J. Offline consolidation 471–488 (Elsevier, 2019). of memory varies with time in slow wave sleep and can be accelerated 9. Rasch, B., Buchel, C., Gais, S. & Born, J. Odor cues during slow-wave by cuing memory reactivations. Neurobiol. Learn. Mem. 98, sleep prompt declarative memory consolidation. Science (80-.). 315, 103–111 (2012). 1426–1429 (2007). 32. Siefert, E. M. et al. Memory reactivation during sleep does not act 10. Rudoy, J. D., Voss, J. L., Westerberg, C. E. & Paller, K. A. holistically on object memory. bioRxiv (2023). Preprint at https://doi. Strengthening individual memories by reactivating them during sleep. org/10.1101/2023.12.14.571683. Science 326, 1079 (2009). 33. Bar, E. et al. Local targeted memory reactivation in human sleep. Curr. 11. Schönauer, M., Geisler, T. & Gais, S. Strengthening procedural Biol. 30, 1435–1446.e5 (2020). memories by reactivation in sleep. J. Cogn. Neurosci. 26, 34. Gao, C., Fillmore, P. & Scullin, M. K. Classical music, educational 143–153 (2014). learning, and slow wave sleep: A targeted memory reactivation 12. Rakowska, M., Abdellahi, M. E. A., Bagrowska, P., Navarrete, M. & experiment. Neurobiol. Learn Mem. 171, 1–24 (2020). Lewis, P. A. Long term effects of cueing procedural memory 35. Vidal, V. et al. Odor cueing during sleep improves consolidation of a reactivation during NREM sleep. Neuroimage 244, 118573 (2021). history lesson in a school setting. Sci. Rep. 12, 1–7 (2022). 13. Rakowska, M. et al. Cueing motor memory reactivation during NREM 36. Wick, A. & Rasch, B. Targeted memory reactivation during slow-wave sleep engenders learning-related changes in precuneus and sleep vs. sleep stage N2: no significant differences in a vocabulary sensorimotor structures. bioRxiv (2022). Preprint at https://doi.org/10. task. Learn. Mem. 30, 192–200 (2023). 1101/2022.01.27.477838. 37. Picard-Deland, C., Aumont, T., Samson-Richer, A., Paquette, T. & 14. Schechtman, E. et al. Multiple memories can be simultaneously Nielsen, T. Whole-body procedural learning benefits from targeted reactivated during sleep as effectively as a single memory. Commun. memory reactivation in REM sleep and task-related dreaming. Biol. 4, 1–13 (2021). Neurobiol. Learn. Mem. 183, 107460 (2021). 15. Schechtman, E., Witkowski, S., Lampe, A., Wilson, B. J. & Paller, K. A. 38. Veldman, M. P. et al. Somatosensory targeted memory reactivation Targeted memory reactivation during sleep boosts intentional enhances motor performance via hippocampal-mediated plasticity. forgetting of spatial locations. Sci. Rep. 10, 1–9 (2020). Cereb. Cortex 33, 3734–3749 (2023). 16. Schechtman, E. et al. Sleep reactivation did not boost suppression- 39. Nicolas, J. et al. Targeted memory reactivation during post-learning induced forgetting. Sci. Rep. 11, 1–17 (2021). sleep does not enhance motor memory consolidation in older adults. 17. Borghese, F. et al. Targeted memory reactivation during REM sleep in J. Sleep Res. 1–17 (2023). https://doi.org/10.1111/jsr.14027 patients with social anxiety disorder. Front. Psychiatry 13, 40. Cairney, S. A., Durrant, S. J., Hulleman, J. & Lewis, P. A. Targeted 1–16 (2022). memory reactivation during slow wave sleep facilitates emotional 18. Shimizu, R. E. et al. Closed-loop targeted memory reactivation during memory consolidation. Sleep 37, 701–707 (2014). sleep improves spatial navigation. Front. Hum. Neurosci. 12, 41. Xia, T. et al. Updating memories of unwanted emotions during human 1–14 (2018). sleep. Curr. Biol. 33, 309–320.e5 (2023). 19. Pereira, S. I. R. et al. Cueing emotional memories during slow wave 42. Hutchison, I. C., Pezzoli, S., Tsimpanouli, M. E., Abdellahi, M. E. A. & sleep modulates next-day activity in the orbitofrontal cortex and the Lewis, P. A. Targeted memory reactivation in REM but not SWS amygdala. Neuroimage 253, 119120 (2022). selectively reduces arousal responses. Commun. Biol 4, 1–6 (2021). 20. Legendre, G. et al. Reinstatement of emotional associations during 43. Goldstein, A. N. & Walker, M. P. The role of sleep in emotional brain human sleep: an intracranial EEG study. bioRxiv (2022). Preprint at function. Annu. Rev. Clin. Psychol. 10, 679–708 (2014). https://doi.org/10.1101/2022.06.24.497499. 44. Yuksel, C. et al. Emotional memories are enhanced when reactivated 21. Göldi, M. et al. Increased neuronal signatures of targeted memory in slow wave sleep, but impaired when reactivated in REM. bioRxiv reactivation during slow-wave up states. Sci. Rep. 9, 1–10 (2019). (2023). Preprint at https://doi.org/10.1101/2023.03.01.530661. 22. Ngo, H. V. & Staresina, B. P. Shaping overnight consolidation via slow- 45. Simon, K. C. N. S., Gómez, R. L. & Nadel, L. Losing memories during oscillation closed-loop targeted memory reactivation. Proc. Natl sleep after targeted memory reactivation. Neurobiol. Learn. Mem. Acad. Sci. 119, e2123428119 (2022). 151, 10–17 (2018). 23. Schwartz, S., Clerget, A. & Perogamvros, L. Enhancing imagery 46. Whitmore, N. W., Bassard, A. M. & Paller, K. A. Targeted memory rehearsal therapy for nightmares with targeted memory reactivation. reactivation of face-name learning depends on ample and Curr. Biol. 32, 4808–4816.e4 (2022). undisturbed slow-wave sleep. npj Sci. Learn 7, 1–6 (2022). 24. Talamini, L. M. & Juan, E. Sleep as a window to treat affective 47. Whitmore, N. W. & Paller, K. A. Sleep disruption by memory cues disorders. Curr. Opin. Behav. Sci. 33, 99–108 (2020). selectively weakens reactivated memories. Learn. Mem. 30, 25. van der Heijden, A. C., van den Heuvel, O. A., van der Werf, Y. D., 63–69 (2023). Talamini, L. M. & van Marle, H. J. F. Sleep as a window to target 48. Ngo, H. V. V., Martinetz, T., Born, J. & Mölle, M. Auditory closed-loop traumatic memories. Neurosci. Biobehav. Rev. 140, 104765 (2022). stimulation of the sleep slow oscillation enhances memory. Neuron 26. Neumann, F., Oberhauser, V. & Kornmeier, J. How odor cues help to 78, 545–553 (2013). optimize learning during sleep in a real life-setting. Sci. Rep. 10, 49. Wang, J. Y., Heck, K. L., Born, J., Ngo, H. V. V. & Diekelmann, S. No 1–8 (2020). difference between slow oscillation up- and down-state cueing for 27. Göldi, M. & Rasch, B. Effects of targeted memory reactivation during memory consolidation during sleep. J. Sleep. Res. 31, e13562 (2022). sleep at home depend on sleep disturbances and habituation. npj Sci. 50. Abdellahi, M. E. A., Koopman, A. C. M., Treder, M. S. & Lewis, P. A. Learn 4, 1–7 (2019). Targeting targeted memory reactivation: characteristics of cued 28. Whitmore, N. W., Harris, J. C., Kovach, T. & Paller, K. A. Improving reactivation in sleep. Neuroimage 266, 119820 (2023). memory via automated targeted memory reactivation during sleep. J. 51. Antony, J. W. et al. Sleep spindle refractoriness segregates periods of Sleep Res. 31, e13731 (2022). memory reactivation. Curr. Biol. 28, 1736–1743 (2018). 29. Squire, L. R. & Zola, S. M. Structure and function of declarative and 52. Cairney, S. A., Guttesen, A. áV., El Marj, N. & Staresina, B. P. Memory nondeclarative memory systems. Proc. Natl Acad. Sci. Usa. 93, consolidation is linked to spindle-mediated information processing 13515–13522 (1996). during sleep. Curr. Biol. 28, 948–954 (2018). npj Science of Learning | (2024)9:31 9 https://doi.org/10.1038/s41539-024-00244-8 Review article 53. Antony, J. W., Schönauer, M., Staresina, B. P. & Cairney, S. A. Sleep sleep on memory consolidation? Curr. Neurol. Neurosci. Rep. 14, spindles and memory reprocessing. Trends Neurosci. 42, 1–3 (2019). 430 (2014). 54. Schechtman, E., Heilberg, J. & Paller, K. A. Memory consolidation during sleep involves context reinstatement in humans. Cell Rep. 42, Acknowledgements 112331 (2023). This work was funded by a grant from the Deutsche 55. Forcato, C. et al. Reactivation during sleep with incomplete reminder Forschungsgemeinschaft (DFG) to SD (FOR 5434, DI 1866/6-1) as well as a cues rather than complete ones stabilizes long-term memory in doctoral research fellowship from the Deutscher Akademischer humans. Commun. Biol. 3, 1–13 (2020). Austauschdienst (DAAD) to JC. 56. Carbone, J. et al. The effect of zolpidem on targeted memory reactivation during sleep. Learn. Mem. 28, 307–318 (2021). Author contributions 57. Carbone, J., Bibián, C., Born, J., Forcato, C. & Diekelmann, S. J.C. and S.D. both conceived, wrote and revised the manuscript. Targeted memory reactivation is not more effective during slow-wave sleep than in sleep stage 2. BioRxiv (2023). Preprint at https://doi.org/ Funding 10.1101/2023.10.19.563121. Open Access funding enabled and organized by Projekt DEAL. 58. Pereira, S. I. R. et al. Rule abstraction is facilitated by auditory cuing in REM sleep. J. Neurosci. 43, 3838–3848 (2023). Competing interests 59. Abdellahi, M. E. A., Koopman, A. C. M., Treder, M. S. & Lewis, P. A. The authors declare no competing interests. Targeted memory reactivation in human REM sleep elicits recurrent, temporally compressed reactivation. Neuroimage 266, Additional information 119820 (2022). Supplementary information The online version contains 60. Picard-Deland, C. & Nielsen, T. Targeted memory reactivation has a supplementary material available at sleep stage-specific delayed effect on dream content. J. Sleep. Res. https://doi.org/10.1038/s41539-024-00244-8. 31, 1–9 (2022). 61. Abdellahi, M. E. A., Koopman, A. C. M., Treder, M. S. & Lewis, P. A. Correspondence and requests for materials should be addressed to Targeted memory reactivation in human REM sleep elicits detectable Susanne Diekelmann. reactivation. Elife 12, 1–14 (2023). 62. Salvesen, L., Capriglia, E., Dresler, M. & Bernardi, G. Influencing Reprints and permissions information is available at dreams through sensory stimulation: a systematic review. Sleep Med. http://www.nature.com/reprints Rev. 74, 101908 (2024). 63. Haar Horowitz, A., Cunningham, T. J., Maes, P. & Stickgold, R. Publisher’s note Springer Nature remains neutral with regard to Dormio: A targeted dream incubation device. Conscious. Cogn. 83, jurisdictional claims in published maps and institutional affiliations. 102938 (2020). 64. Weinhold, S. L. et al. Auditory stimulation in-phase with slow Open Access This article is licensed under a Creative Commons oscillations to enhance overnight memory consolidation in patients Attribution 4.0 International License, which permits use, sharing, with schizophrenia? J. Sleep. Res. 31, e13636 (2022). adaptation, distribution and reproduction in any medium or format, as long 65. Gvozdanovic, G., Schoch, S., Stämpfli, P., Seifritz, E. & Rasch, B. as you give appropriate credit to the original author(s) and the source, Neural correlates of sleep-induced benefits on traumatic memory provide a link to the Creative Commons licence, and indicate if changes processing. Hum. Brain Mapp. 44, 3506–3518 (2023). were made. The images or other third party material in this article are 66. Chen, L., Wang, Y., Zhang, Y. & Wang, Y. A new application of TMR: A included in the article’s Creative Commons licence, unless indicated study on implicit self-esteem. Curr. Psychol. 42, 5159–5168 (2023). otherwise in a credit line to the material. If material is not included in the 67. Ziqing, Y. et al. Reactivating positive personality traits during sleep article’s Creative Commons licence and your intended use is not permitted promotes positive self-referential processing. bioRxiv (2024). Preprint by statutory regulation or exceeds the permitted use, you will need to at https://doi.org/10.1101/2022.11.27.518064. obtain permission directly from the copyright holder. To view a copy of this 68. Amores, J., Dotan, M. & Maes, P. Development and study of ezzence: licence, visit http://creativecommons.org/licenses/by/4.0/. A modular scent wearable to improve wellbeing in home sleep environments. Front. Psychol. 13, 1–19 (2022). © The Author(s) 2024 69. Ackermann, S. & Rasch, B. Differential effects of non-REM and REM npj Science of Learning | (2024)9:31 10