Research reportSleep unbinds memories from their emotional context
Introduction
After a learning episode, still fragile memory traces are gradually converted with time into more stable representations in long-term memory (McGaugh, 1966; Müller and Pilzecker, 1900). Growing evidence suggests that memory consolidation, or the process of trace stabilization and strengthening, preferentially occurs during sleep (Gais and Born, 2004; Maquet, 2001; Stickgold, 2005). Besides merely stabilizing memories however, sleep also contributes in protecting them against retroactive interference (Ellenbogen et al., 2009, 2006). In a classical AB–AC interference paradigm, Ellenbogen et al. (2006) asked participants to memorize a list of word pairs (AB), after which they spent a period of time either sleeping (at night) or awake (during daytime). At the second session and before retrieval of list AB, half of the participants memorized novel word pairs (AC), in which the first word of the pair was the same as in list AB, hence creating interference, whereas the other half directly proceeded to cued recall of list AB. In the non-interference condition, results showed a trend for a better recall of list AB after sleep than wakefulness. In the interference condition however, the beneficial effect of sleep was highly significant with a much better recall of list AB just after learning list AC, suggesting that memories consolidated during sleep are more resistant to retroactive interference. This effect was confirmed in a subsequent study in which participants were tested first in non-interference then in interference conditions within the same retrieval session (Ellenbogen et al., 2009). In the above studies, the material used to elicit interference was intrinsic to the learned material (word, engram), hence creating lexical interference. Still, it is not known whether extrinsic, contextual information would also induce interference effects. It is also not known whether sleep would prevent such interference to occur, especially in the domain of emotional interference, although there is a known effect of sleep on the consolidation of emotional memories.
It is now well established by various neuroimaging, neuropsychological and neural stimulation studies that memory is influenced by emotional arousal both at the memory encoding and memory consolidation stages (for review see Hamann, 2001), in such a way that emotional experiences are better remembered than neutral ones (e.g., Cahill, 2000; Dolcos et al., 2005; McGaugh, 2004; Phelps, 2004; Walker and Tarte, 1963). Most interestingly, it has been shown that a retention interval longer than 24 h leads to greater emotional arousal enhancement effects on declarative memory (Kleinsmith and Kaplan, 1963; LaBar and Phelps, 1998; Sharot and Yonelinas, 2007), suggesting a possible implication of sleep and/or time elapsed awake. Moreover, Hu et al. (2006) compared recall performance for negative and neutral pictures after a full night of sleep versus a same interval spent awake across the day. They found an emotional memory benefit after post-training sleep, validating the hypothesis of a sleep-dependent process in emotional memory consolidation. Furthermore, to avoid circadian confounds, Atienza and Cantero (2008) compared memory performance for emotional and neutral pictures after 1 week in participants deprived or not of sleep on the first post-learning night. Although, as expected, memory for both picture types was impaired in the sleep deprivation condition, performance was surprisingly better for emotional pictures suggesting that emotional stimuli are less sensitive to sleep deprivation; alternatively that the six recovery nights preferentially strengthened emotional altered memories. However, Wagner et al. (2001) demonstrated in a selective sleep deprivation paradigm that the second half of a night of sleep, richer in rapid eye movement (REM) sleep, is actually responsible for the beneficial effect of sleep on emotional texts as compared to neutral ones. In this study, participants studied brief texts having an emotional or a neutral content, and were then allowed to sleep during an interval of 3 h either covering the first part of the night of sleep (early sleep), richer in slow-wave sleep (SWS), or the second half (late sleep) dominated by REM sleep, or were kept awake during the same periods. Results disclosed a selective enhancement of memory for emotional texts following late, REM-rich sleep only (Wagner et al., 2001), persisting up to 4 years later (Wagner et al., 2006). REM sleep involvement was also confirmed in a nap paradigm (Nishida et al., 2009), in which enhancement in emotional memory after a 90-min nap was correlated with the amount of REM sleep. Finally, Payne et al. (2008) investigated whether sleep enhances memory for the emotional experience as a whole, or acts separately on different emotional memory constituents by enhancing only the affective component. Participants were presented pictures of neutral or negative objects on a neutral background and then had a recognition test on object and background separately after 12 h spent awake during daytime or 12 nighttime hours including sleep. Results showed that negative objects were consolidated at the expense of their background in the sleep-delayed condition, whereas both components were altered at similar rates in the wake-delayed condition, suggesting that sleep preferentially enhances the emotional components of memories.
All of these studies have investigated how sleep impacts on the consolidation of an emotional item or an emotional object in a neutral background. However, it remains unclear what would happen if the object is emotionally neutral but the background arouses emotions, or if the item to-be-learned is neutral but associated with a non-neutral mood at encoding. Kenealy (1997) showed that participants who learn and recall material in the same mood have better recall performance than participants who learn and recall in different moods, suggesting that a change of mood from learning to recall interferes with recall capacities (Bartlett and Santrock, 1979; Kenealy, 1997).
In this perspective, taking into account the proposal that sleep protects declarative memory (e.g., word pairs) against interference (Ellenbogen et al., 2009, 2006), we wondered whether this protective effect embraces the emotional component of novel memories, in particular when the emotional component is extrinsic to the learned material such as a different mood at the encoding and the recall sessions. To test this hypothesis, we designed a within-subjects study in which participants learned neutral word pairs after a mood induction procedure (MIP) (Mayer et al., 1995), then slept or were kept awake on the first post-training night, and were finally tested after two recovery nights either in the same or in a different mood than at the learning phase. To account for this effect known as mood-state dependent retrieval, Bower (1981) proposed the Network Theory of Affect. According to this theory, a network of nodes representing interconnected semantic concepts would constitute memory. Emotions would be units or nodes also integrated in this semantic network, and linked with propositions or assertions describing events during which the specific emotion was aroused. Hence when an emotional node is activated, emotion-related nodes in the semantic network would be activated also promoting their accessibility in mind. Consequently, when a similar emotional context takes place during the recall session, the same emotional nodes than at the encoding session are reactivated, eventually activating stimuli presented during the learning phase. In this case, the emotional context cues the learned material, hence a facilitation effect. On the other hand, when the emotional context at recall is different from the one experienced at the learning session, this facilitation effect cannot take place. Furthermore, a different emotional context may activate emotion-related concepts that will compete with the learned stimuli, hence creating interference.
Section snippets
Participants
Twenty-six healthy participants gave written informed consent to participate in this study, which was approved by the local ethics committee. Due to irregular sleep pattern or floor effects in the memory task (recall score at delayed phase <10%), four participants were excluded from the statistical analyses. The 22 remaining participants (12 males, mean age ± SD 21.86 ± 2.38 years) met the following criteria: native French speakers, non-smokers, no history of neurological disorders or sleep
Sleep and vigilance parameters
A repeated measures ANOVA on mean reported sleep duration during the experiment with factors Sleep (RS vs SD) and Session (Learning vs Recall) did not reveal any main Sleep (F1,21 = 3.17, p = .09) or Session (F1,21 = 4.04, p = .06) nor interaction (F1,21 = .01, p = .9) effects, indicating similar mean sleep duration before learning and recall sessions in each condition. During the first night after learning in the sleep condition, mean sleep duration was 8.03 ± 1.41 h.
Repeated measures ANOVAs
Discussion
In the present study, we investigated how post-training sleep modulates the effect of contextual emotional interference on the delayed recall of verbal memories. First, our results replicate the finding that between-sessions congruency in the emotional context of neutral words enhances memory recall (Kenealy, 1997). Indeed, we found a group effect with higher recall when participants learned and recalled word pairs after the same MIP than when the MIP was different, suggesting a mood-state
Conclusion
In summary, we show here that sleep reduces the facilitating effect of emotional congruency between learning and retrieval contextual background, suggesting that post-training sleep releases recently learned memories from their emotional context. Thanks to the sleep-dependent decoupling of memory from its affective tag and the removing of the latter, the emotional reactivity associated with these items is reduced, consequently making memories less sensitive to emotional cueing.
Acknowledgment
The authors thank Isaline Caudron for help in data acquisition, and two anonymous Reviewers for their insightful and constructive comments on a prior version of this manuscript. GD, MG and RS are Research Fellows at the Belgian Fonds National de la Recherche Scientifique (FNRS). This study has been conducted with support of FNRSCC grant 1.5.184.10.F and FRSM grant 3.4.594.08.F.
References (56)
- et al.
Psychometric properties of Beck Depression Inventory: Twenty-five years of evaluation
Clinical Psychology
(1988) - et al.
The Pittsburgh Sleep Quality Index: A new instrument for psychiatric practice and research
Psychiatry Research
(1989) Neurobiological mechanisms of emotionally influenced, long-term memory
- et al.
Interfering with theories of sleep and memory: Sleep, declarative memory, and associative interference
Current Biology
(2006) Cognitive and neural mechanisms of emotional memory
Trends in Cognitive Sciences
(2001)Human emotion and memory: Interactions of the amygdala and hippocampal complex
Current Opinion in Neurobiology
(2004)- et al.
Brief sleep after learning keeps emotional memories alive for years
Biological Psychiatry
(2006) - et al.
Memory storage as a function of arousal and time with homogeneous and heterogeneous lists
Journal of Verbal Learning and Verbal Behavior
(1963) - et al.
Subjective and objective sleepiness in the active individual
International Journal of Neuroscience
(1990) - et al.
Modulatory effects of emotion and sleep on recollection and familiarity
Journal of Sleep Research
(2008)