The effect of fragmented sleep on emotion regulation ability and usage

ABSTRACT

Sleep has a profound effect on our mood, but insight in the mechanisms underlying this association is still lacking. We tested whether emotion regulation is a mediator in the relationship between fragmented sleep and mood disturbance. The effect of fragmented sleep on the emotion regulation strategies, including cognitive reappraisal, distraction, acceptance and suppression ability, was assessed. We further tested whether the use of these strategies, as well as rumination and self-criticism, mediated the association between fragmented sleep and negative and positive affect. Participants (N = 69) wore an actiwatch and filled in a sleep diary for 12 consecutive nights. They had one control night and one sleep fragmentation night. Emotion regulation ability was assessed with an experimental task. Usage of emotion regulation strategies and negative and positive affect were assessed four times during the day with a survey after the control and sleep fragmentation night. Cognitive reappraisal, distraction, acceptance and suppression ability did not differ between the sleep fragmentation and control condition. However, participants reported higher usage of rumination and distraction after the sleep fragmentation night and rumination significantly mediated the negative association between fragmented sleep and negative affect.

KEYWORDS:

• Sleep fragmentation
• emotion regulation
• negative affect
• positive affect

Sleep is vital for all aspects of human health and plays a crucial role in the homeostasis of positive and negative affect (Fairholme & Manber, 2015). Poor sleep quality, such as having non-restorative sleep due to fragmented sleep (i.e. brief awakenings) during the night, is associated with a reduction in positive affect (i.e. positive emotions such as happiness and joy) and an increase in negative affect (i.e. negative emotions such as sadness and anxiety) during the following day (Fairholme & Manber, 2015; Palmer & Alfano, 2017). However, why sleep quality has such a strong impact on our affect remains an open question. A growing literature suggests that the reduced ability to regulate one’s emotions might be an important mechanism in the relationship between poor sleep quality and negative and positive affect (Mauss et al., 2013; Muzur et al., 2002; Palmer & Alfano, 2017).

Emotion regulation refers to the thoughts we have and actions we perform to influence what emotions we experience, how intensely we experience them, and how we express them (Gross & John, 2003). Commonly used emotion regulation strategies are cognitive reappraisal (i.e. trying to see a situation in a more positive light), acceptance (i.e. being aware of and accepting emotions without feeling the need to change them), and distraction (i.e. shifting attention to neutral or positive stimuli; Gross & John, 2003). These strategies are often labelled as adaptive and are associated with increases in positive affect and decreases in negative affect (Gross & John, 2003). Other commonly used emotion regulation strategies are suppression (i.e. hiding in your expression what you are actually feeling), rumination (i.e. thinking persistently and negatively about a situation), and self-criticism (i.e. judging or evaluating oneself negatively; Clancy et al., 2020; Gross & John, 2003). These strategies are often labelled as maladaptive and are associated with decreases in positive affect, increases in negative affect, and are a risk factor for many mood and anxiety disorders (Clancy et al., 2020; Gross & John, 2003).

Several neurobiological findings suggest a close relationship between sleep quality and emotion regulation (Muzur et al., 2002; Yoo et al., 2007). Brain regions involved in emotion regulation are the Prefrontal Cortex (PFC), that plays a key role in top-down cognitive control (i.e. the ability to modify thoughts and behaviour in accordance with internal goals and intentions; Miller & Cohen, 2001), as well as the amygdala, a brain region important for emotion generation (Muzur et al., 2002). However, when sleep-deprived, there is a reduced functional connectivity between the PFC and the amygdala (Yoo et al., 2007), resulting in a reduced emotion-regulatory control of the PFC. These findings are in line with other neuroimaging studies that demonstrate that poor sleep impairs executive functions of the PFC (i.e. a key component of cognitive control), such as inhibitory and attentional control, which has direct consequences for emotion regulation ability (i.e. the effectiveness of the emotion regulation strategy to modify and improve mood: for a review see Gruber & Cassoff, 2014). For instance, impaired cognitive control following poor sleep could reduce the ability to cognitively reappraise, as during cognitive reappraisal one has to be able to shift attention to new information, hold the information in mind, and assume a new perspective (Muzur et al., 2002; Yoo et al., 2007).

With the reduced top-down cognitive control from the PFC following poor sleep it might also be more difficult to effectively use distraction or acceptance to regulate emotions (Cote et al., 2015). For instance, when using distraction, one has to be able to shift attention away from internal or external negative stimuli to neutral or positive stimuli (Derryberry & Reed, 2002). Evidence suggests that impaired sleep quality can result in an inability to shift attention away from negative stimuli (Cote et al., 2015). Attentional control also plays a role in acceptance (Cote et al., 2015). When using acceptance, people have to voluntarily focus attention on their current emotional experience and disengage attention from automatic thoughts (Cote et al., 2015).

Reduced sleep quality might also be associated with maladaptive emotion regulation strategies, such as rumination and self-criticism. Studies show a positive association between poor sleep quality and rumination (for a review see Clancy et al., 2020) and short sleep and increased self-criticism (Vincent et al., 2009). Disengaging attention from negative thoughts, faults, and mistakes is suggested to be more difficult after a night of poor sleep quality, as disrupted sleep impairs attentional control, possibly resulting in increases in rumination and self-criticism (Koster et al., 2011; Yoo et al., 2007).

Results about the association between sleep quality and suppression are still inconclusive. Impaired sleep quality is associated with more difficulty to inhibit responses to negative events (Dahl & Lewin, 2002). There is however also evidence against this inhibition hypothesis as the study of Minkel et al. (2011) demonstrated that participants were less emotionally expressive towards both positive and negative stimuli and that inhibition of expression was not impaired after sleep deprivation.

So far, most studies looking at the association between sleep quality and emotion regulation have been cross-sectional. Only few studies have looked at causal effects by manipulating sleep and assessing the effect on emotion regulation strategies (Reddy et al., 2017; Shermohammed et al., 2020; Zhang et al., 2019). Results of these studies were mixed. Zhang et al. (2019) found that sleep deprivation impaired the regulatory effect of both cognitive reappraisal and distraction as measured with an EEG recording, but they did not find an effect of sleep deprivation on suppression ability. However, two other studies could not detect an effect of a night of reduced sleep quantity on the ability to use cognitive reappraisal (Reddy et al., 2017; Shermohammed et al., 2020).

Besides having yielded mixed results, previous research on sleep and emotion regulation is characterised by a number of methodological limitations. Firstly, studies assessing the effect of sleep on emotion regulation have employed a sleep deprivation protocol and therefore reduced sleep quantity, i.e. the number of hours asleep (Reddy et al., 2017; Shermohammed et al., 2020; Zhang et al., 2019). Sleep deprivation might not be an ecologically valid indicator of disrupted sleep (Vandekerckhove & Cluydts, 2010). Disrupted sleep, also as part of several sleep disorders, is often characterised by sleep fragmentation, i.e. short arousals that interrupt sleep (Vandekerckhove & Cluydts, 2010). Sleep fragmentation reduces sleep quality, since the brief arousals disrupt the normal sleep cycle, while often leaving sleep quantity intact (Vandekerckhove & Cluydts, 2010). There is evidence that sleep quality might better predict adaptive emotion regulation than sleep quantity. For instance, Mauss et al. (2013) demonstrated that poor sleep quality, but not reduced sleep quantity, was associated with reduced ability to use cognitive reappraisal, although this study did not manipulate sleep.

Secondly, next to the call for studies to manipulate sleep quality, Palmer and Alfano (2017) stressed the importance of differentiating between the effect of sleep quality on emotion regulation ability and the frequency of deployment of emotion regulation strategies. Sleep quality might impair how well emotion regulation strategies can be applied, as well as how often they are actually used during the day. So far, longitudinal studies demonstrate that maladaptive emotion regulation (especially rumination) can be a mechanism in the association between sleep disturbance and depression (Kirschbaum-Lesch et al., 2021; Niu & Snyder, 2023; Palmer et al., 2018). However, it is still an open question on what time scale these mediating effects occur. So far, no study has tested it on a short time scale (i.e. the day level), namely whether usage of emotion regulation strategies during the day can explain the association between poor sleep quality and positive and negative affect the next day.

Next to the main focus on regulation ability, the accent of studies assessing the causal effect of sleep on emotion regulation has been on cognitive reappraisal (Mauss et al., 2013; Reddy et al., 2017; Shermohammed et al., 2020). Only one study assessed the impact on other emotion regulation strategies (i.e. distraction and suppression; Zhang et al., 2019). Focus has to be extended to other emotion regulation strategies, as emotion regulation is an umbrella term for many different strategies and sleep might impact each strategy in slightly different ways.

The current study aims to address the three gaps in the literature by (1) manipulating sleep quality (instead of quantity) by fragmentating sleep and assessing the effect on emotion regulation ability, (2) assessing whether emotion regulation frequency mediates the effect of fragmented sleep on positive and negative affect, and (3) assessing the effect of fragmented sleep on the ability to use various strategies: cognitive reappraisal, distraction, acceptance, and suppression and the frequency with which these four strategies, as well as rumination and self-criticism, are used. Participants were exposed to a control night (normal night of sleep) and a sleep fragmentation night (i.e. woken up every 80 min by an alarm; following the procedure of Rosseland et al., 2018) to manipulate sleep quality. We expected that after the sleep fragmentation night participants had a reduced ability to use cognitive reappraisal, distraction, and acceptance compared to the control night. Given the mixed results for suppression, the effect of fragmented sleep on suppression ability was examined in an explorative fashion. We also expected that participants would report lower frequency of use of cognitive reappraisal, distraction, acceptance and higher usage of suppression, rumination, self-criticism after the sleep fragmentation night compared to the night with normal sleep and that this would result in higher negative affect and lower positive affect.

Method

Participants

Power analysis is complicated for mixed models (Matuschek et al., 2017). There is not yet an accurate method to calculate the power for a multilevel mediation analysis. We therefore only did a power calculation for assessing the effect of the sleep manipulation on emotion regulation ability. A number of 62 participants was sufficient to have a power of .80 with a small to medium effect size (f = .15), 2 groups, and 4 measurements for a repeated measures ANOVA in Gpower (Faul et al., 2007). A mixed model has higher power when missing data is present and it gives the same results as a repeated measures ANOVA. We, therefore, chose to use a mixed model. To account for possible dropouts, we aimed for 70 participants. University students were recruited via the University website. The final sample size consisted of 72 participants that received in exchange €25 or course credit. Three participants were excluded from all analyses, due to dropping out of the study in the first week. The final sample, therefore, consisted of 69 participants, consisting of 63 females and 6 males. Participants had a mean age of 20.86 years (SD = 2.88, range = 18–29). This study was preregistered (see https://osf.io/zb4he/). The research has been independently reviewed by the Ethics Committee Social Sciences (ECSS) of the Radboud University, and there was no formal objection.

Procedure

The study covered a period of 12 consecutive nights (see Figure S1 in the Supplementary File for an overview of the study). The study always started on a Sunday. After participants filled in the informed consent, they put an actiwatch around their non-dominant wrist to assess sleep objectively. They wore the actiwatch for 12 consecutive nights, and only detached it during showering/bathing and exercising. Subsequently, participants filled in the general online survey in Qualtrics assessing demographics, trait negative affect, and general sleep quality.

For the next 12 mornings, from Monday to Friday, participants filled in the Consensus Sleep Diary (CSD; Carney et al., 2012). Participants also pushed the event marker on the actiwatch when they went to bed and got out in the morning. On Thursday night, participants either had the sleep fragmentation night or the control night. In the sleep fragmentation night, participants were instructed to go to bed and get out of bed at their normal bed- and rise time. Normal bed and rise times were calculated per participant by taking the average of reported bed and rise times from Sunday night to Wednesday night from the CSD of the first week. In the sleep fragmentation night, participants were woken up every 80 min from the start of their bed time by an alarm set on their mobile phone. Participants were notified beforehand about the set alarms. When participants woke up from the alarm, they filled in an online filler survey of around 5 min (i.e. 60-item PANAS-X and 2 items from the Pittsburgh Sleep Quality Index), to make sure they stayed awake for a moment. In the control night, participants also went to bed at their average bed time, but only one alarm was set, namely at their average rise time in the morning. They were instructed to get out of their bed once the alarm went off.

The Friday morning after the control and sleep fragmentation night, participants performed an online emotion regulation task. They started the task one hour after getting out of bed. Participants were instructed to turn their phone off and sit in a quiet room where they could not be interrupted. The emotion regulation task took around 45 min (See Figure S2 in de Supplementary File for a visualisation of the task). Participants started with watching a neutral film clip to put them in a similar emotional state. After the neutral film clip, participants filled in the emotion survey in which they rated the extent to which they experienced eight different emotions (i.e. disgust, anxiety, happiness, sadness, amusement, fear, surprise, and anger on a scale from 1 (not at all) to 9 (very much)) during the film clip. After filling in the emotion survey, participants watched a film clip that triggered sadness. This sad film clip served as a baseline to determine the effectiveness of the emotion regulation strategy and to index an individual's capacity for negative mood induction. Participants were instructed to carefully watch the film clip. After the film clip, they filled in the emotion survey and 2 control questions (“have you seen this film clip before?” and “did you close your eyes or look away during the film clip?” yes/no). Then participants performed a distraction task (i.e. drawing geometrical figures on a piece of paper) to recover from the sadness inducing film clip. They were instructed that the quality of the drawing was not important and that this task only served as a break. Next, participants watched another sad film clip. This time they received one of the four emotion regulation strategies instructions, namely cognitive reappraisal, distraction, acceptance, or suppression (derived from Campbell-Sills et al., 2006; Sheppes & Meiran, 2007; Troy et al., 2018). The instructions can be found in the Supplementary Files and were presented on screen and via audio. We had a maximum of four emotion regulation strategies (similar to the study of Lohani & Isaacowitz, 2014) and chose strategies with validated instructions from previous studies (Campbell-Sills et al., 2006; Sheppes & Meiran, 2007; Troy et al., 2018). Participants were instructed to implement the particular emotion regulation strategy while watching the sad film clip. After this film clip, participants filled in the emotion survey, the control questions and an additional question (“How hard did you try to follow the instruction?” 1 = not at all 9 = very much), and finally did the distraction task. This procedure was repeated until they had performed all four different emotion regulation strategies. The order of the four emotion regulation instructions was random, as well as the order of the five sadness-inducing film clips.

After finishing the task, participants were asked to fill in four surveys (i.e. at 12, 3, 6, and 9 pm) assessing the degree to which they used the emotion regulation strategies cognitive reappraisal, distraction, acceptance, suppression, rumination, and self-criticism and their momentary positive and negative affect. In the 9 pm survey, alcohol and caffeine consumption and daily stressors were assessed. This procedure was repeated the subsequent week, except participants got the other condition (i.e. sleep fragmentation night or control night). All parts of this study were conducted remotely (i.e. not in the laboratory). The order of the conditions was randomised. Finally, participants were debriefed and received money or course credits.

Materials

Emotion regulation task

Film clips. The neutral film clip depicted a scene from a nature documentary and was pre-tested in previous research to induce low levels of negative emotions (Bartolini, 2011). All sad film clips were also pre-tested in previous studies and elicited similar levels of sadness (Bartolini, 2011; Schaefer et al., 2010; See Table S3 in the Supplementary File for means and standard deviations of negative and positive emotions during each film clip). Sadness scores were significantly higher during the baseline sad film clip compared to neutral film clip for both the control condition, t(68) = −12.58, p < .001, and the sleep fragmentation night, t(68) = −10.65, p < .001 indicating a successful induction of sadness via the film clips. Sad film clips depicted scenes from movies with topics such as social exclusion or grief.

Actigraphy

Sleep was measured objectively via actigraphy (Actiwatch 2, Philips Respironics, Murrysville, USA). The Actiwatch 2 measures movement with an accelerometer and has a photopic light sensor. The Respironics Actiware 5 software (Philips Respironics, Murrysville, USA) was used to analyse the actigraphy data. This software uses the acceleration count values as measured by the Actiwatch 2 to estimate for each epoch (i.e. 60 s) whether someone is awake or asleep. How the actigraphy data were analysed can be found in the Supplementary File.

Consensus sleep diary

Subjective sleep was measured with the Consensus Sleep Diary (CSD; Carney et al., 2012). The CSD measures time into bed, time of trying to fall asleep, time it took to fall asleep, number of awakenings, duration of each awakening, time of final awakening, time getting out of bed, and subjective sleep quality. All items except for subjective sleep quality were open questions. Subjective sleep quality was measured with the item “How would you rate the quality of your sleep?” and answer options ranged from 1 = very poor to 7 = very good. The CSD has been shown to be valid and reliable (Maich et al., 2018).

Emotion regulation survey

The frequency that participants engaged in six different emotion regulation strategies were assessed with one item each. Participants were asked how they handled their emotions since finishing the experimental task or the last survey. Cognitive reappraisal was assessed with the item “I controlled my emotions by changing the way I think about a situation I’m in”. Distraction with “I distracted myself”. Acceptance with “I accepted my emotions”. Suppression with “I controlled my emotions by not showing them”. Rumination with “I ruminated about my feelings without a result”. Finally, self-criticism with “I criticized myself for my feelings” (Gross & John, 2003; Nittel et al., 2018). Answer options ranged from 1 = not at all to 10 = very much.

Affect

Both negative and positive affect was measured. Negative affect was measured with 5 items derived from the PANAS-X (i.e. sad, anxious, tense, angry, and blue; Goldstein et al., 2014). Positive affect was measured with 5 items derived from the study of Gill et al. (2017) that took items from the Profile of Mood State Short Form (POMS-SF) (i.e. energetic, lively, cheerful, relaxed, and happy). Responses ranged from 1 = not at all to 10 = extremely. The items of negative affect in this sample had adequate within-person reliability (α = .59) and high between-person reliability (α = .92). For positive affect, within-person reliability (α = .79) and between person reliability (α = .84) were high. Both the items measuring negative affect and positive affect have been proven reliable and valid (Gill et al., 2017; Goldstein et al., 2014).

Control variables and demographics

We assessed five confounding factors (i.e. gender, caffeine consumption, alcohol consumption, daily stressors, and trait negative affect). Details on these measures can be found in the pre-registration (see https://osf.io/zb4he/).

Data analysis

Experimental task

A detailed description of how the randomisation and manipulation check were performed can be found in the Supplementary File. In the experimental task, participants were excluded (n = 3) if they reported no sadness (i.e. a score of 1) during the baseline sad film clip. When participants had a score lower than three on the effort measure of a emotion regulation strategy (“how hard did you try to follow the instruction?”), that participants’ emotion regulation strategy was excluded in the analyses (cognitive reappraisal (n = 12), distraction (n = 6), acceptance (n = 14), and suppression (n = 13)). We only used sadness scores in the analysis. The other emotions served as filler items. Emotion regulation ability was calculated for each emotion regulation strategy separately (i.e. cognitive reappraisal, acceptance, distraction, and suppression) by subtracting the sadness score of the regulated film clip from the sadness score of the baseline sad film clip.

Subsequently, for the ability to deploy each emotion regulation strategy, we ran separate multilevel models using the lme4 package in R to test whether cognitive reappraisal ability, distraction ability, acceptance ability, and suppression ability differed between Condition (0 = control and 1 = sleep fragmentation). We used multilevel analyses, as the data has a hierarchical structure with days (Level 1) nested within persons (Level 2). The multilevel model employed a maximum likelihood estimation method. Condition was a fixed effect, as well as the control variables caffeine consumption (standardised), alcohol consumption (standardised), daily stressors (standardised), gender (0 = Female, 1 = Male), and trait negative effect (standardised). The intercept varied across participants. To correct for multiple testing, all p-values were adjusted using the p.adjust function with the Benjamini and Hochberg method.

Surveys

We used the R package “lavaan” to perform a multilevel mediation analysis with Condition as independent variable, the six emotion regulation strategies as mediators, and negative and positive affect as outcome variables. We used multilevel analysis, as time points were nested within days within subjects (i.e. 3 level model). The intercept and slope varied by participant and a maximum likelihood estimation method was used. All p-values were adjusted using the Benjamini and Hochberg method.

Results

Sleep manipulation

Overall, the randomisation check was successful and baseline sleep was similar prior to the control and sleep fragmentation night. The details of the results of the randomisation check and the manipulation check for baseline sleep can be found in the Supplementary File. A paired samples t-test showed that participants had significantly lower self-reported sleep quality, higher self-reported number of minutes awake, and higher self-reported frequency of awakenings, and lower actigraphy measured sleep efficiency, higher WASO and higher sleep fragmentation during the sleep fragmentation night compared to the control night (see Table S1 in the Supplementary File). Actigraphy measured sleep latency and number of minutes asleep did not differ between the control and sleep fragmentation night. Overall, these results showed that the manipulation was successful.

Main analyses

Multilevel models were performed to test whether cognitive reappraisal, acceptance, distraction, and suppression ability differed after the control and sleep fragmentation night. The ability to use each of the four emotion regulation strategies after the control night was not different from the ability after the sleep fragmentation night (See Table 1 for the results).

Table 1. Means, standard deviations, and the multilevel results of cognitive reappraisal, acceptance, distraction, and suppression ability (i.e. the sadness change score) for the control and sleep fragmentation (SF) condition.

	Control		SF
	M	SD	M	SD	B	SE	χ^2	p
Reappraisal	0.57	0.73	0.78	0.98	.36	0.16	5.28	.09
Distraction	0.76	1.09	0.53	0.95	-.14	0.20	0.49	.65
Acceptance	-0.12	0.74	-0.03	0.87	.16	0.18	0.89	.65
Suppression	0.65	1.11	0.59	1.11	.06	0.23	0.09	.76

Exploratory analyses

Exploratory analyses were performed with the outcome variables cognitive reappraisal, acceptance, distraction, and suppression ability. We added exploratory analyses to test the stability of the results. An exploratory analysis was run with sleep fragmentation as indicated by actigraphy as predictor instead of condition. Results remained insignificant, ps > .13. Exploratory analyses were also run (1) excluding film clips that people had seen before (i.e. 8% of the film clips) or looked away from (i.e. 10% of the film clips) and (2) assessing whether results were different for participants receiving a different order of the control and sleep fragmentation night. Results remained the same, ps > .38 (see Table S4–S6 in the Supplementary Files).

Surveys

Table 2 displays the output of the multilevel mediation analysis and Table S2 in the Supplementary File displays all means and standard deviations.

Table 2. Results of the multilevel mediation analysis testing emotion regulation frequency as a mediator for the effects of sleep fragmentation on positive and negative affect.

Total effect		B	S.E.	p-value
C → NA		.18	.11	.21
C → PA		-.65	.15	<.001
Indirect effect		B	S.E.	p-value
C → Reap → NA		-.003	.01	.65
C → Acc → NA		.02	.02	.39
C → Dis → NA		.03	.02	.17
C → Rum → NA		.09	.04	.04
C → Sup → NA		.01	.01	.43
C → Crit →NA		-.03	.02	.30
C → Reap → PA		-.05	.04	.30
C → Acc → PA		-.04	.04	.38
C → Dis → PA		-.06	.03	.11
C → Rum → PA		-.02	.02	.36
C → Sup → PA		-.03	.03	.37
C → Crit → PA		-.004	.01	.76
Predictor	Outcome	B	S.E.	p-value
C	Reap	-.28	.22	.30
C	Dis	.61	.23	.03
C	Acc	-.21	.19	.37
C	Crit	-.30	.18	.18
C	Rum	.47	.18	.03
C	Sup	.27	.21	.30
Predictor	Outcome	B	S.E.	p-value
Reap	NA	.01	.02	.53
Dis	NA	.05	.02	.05
Acc	NA	-.08	.02	.002
Crit	NA	.08	.03	.04
Rum	NA	.18	.03	<.001
Sup	NA	.05	.03	.18
Reap	PA	.17	.03	<.001
Dis	PA	-.09	.03	.01
Acc	PA	.21	.04	<.001
Crit	PA	.01	.04	.74
Rum	PA	-.05	.04	.30
Sup	PA	-.11	.04	.01

Note: Significant results (p < .05) are bold.

C: condition; NA: negative affect; PA: positive affect; reap: cognitive reappraisal; acc: acceptance; dis: distraction; rum: rumination; crit: self-criticism; sup: suppression.

First, the total effect was estimated for positive affect and negative affect. Participants experienced lower positive affect after the sleep fragmentation night as compared to the control condition. No significant difference was found for negative affect.

Results showed that participants reported higher rumination and distraction after the sleep fragmentation night compared to the control night. There were no differences in frequency of use of cognitive reappraisal, acceptance, self-criticism, and suppression after the sleep fragmentation night compared to the control night.

Higher frequency of self-criticism and rumination were associated with higher negative affect. Higher frequency of acceptance was associated with lower negative affect. Furthermore, higher use of cognitive reappraisal and acceptance were associated with higher positive affect, whereas higher use of distraction and suppression were associated with lower levels of positive affect.

The multilevel model showed that rumination negatively mediated the effect of condition on negative affect. That is, people reported more rumination after the sleep fragmentation night compared to the control night, which in turn predicted higher negative affect. The indirect effects of the other types of emotion regulation strategies (i.e. cognitive reappraisal, distraction, acceptance, suppression, and self-criticism) were not statistically significant for negative affect. The indirect effects of all emotion regulation strategies (i.e. cognitive reappraisal, acceptance, distraction, rumination, suppression, and self-criticism) for positive affect were not statistically significant. Thus, the relation between condition and positive affect was not mediated by the emotion regulation strategies.

Discussion

Over the last few decades, research has demonstrated that poor sleep quality has a profound effect on our mood and can causally contribute to the development of mood and anxiety disorders (Fairholme & Manber, 2015; Palmer & Alfano, 2017). These findings raised the question why sleep has such a profound impact on our mood. The present study tested whether the ability and use of emotion regulation strategies (i.e. cognitive reappraisal, distraction, acceptance, suppression, rumination, and self-criticism) might be a mechanism in the relationship between poor sleep quality and mood disturbance.

Results showed that fragmented sleep did not significantly affect the ability to use cognitive reappraisal, distraction, acceptance, and suppression. We also did not detect an effect of fragmented sleep on the frequency of use of cognitive reappraisal, acceptance, self-criticism, and suppression. Results showed that after the sleep fragmentation night participants reported lower positive affect, but not higher negative affect. This is consistent with findings from previous studies that show that poor sleep often shows a stronger effect on positive affect than negative affect (Finan et al., 2015; Talbot et al., 2010). For the relationship between emotion regulation and affect, we found that in line with our hypotheses, lower acceptance, higher self-criticism, and higher rumination were related to higher negative affect. Higher cognitive reappraisal, higher acceptance, and lower suppression were associated with higher positive affect. These results suggested that self-criticism and rumination can prolong and intensify negative emotional processing and without cognitive reappraisal or acceptance that help to see negative events and feelings in a more positive light, there is additionally a lack of positive affect (Boemo et al., 2022).

Contrary to what was hypothesised, we found that higher distraction was related to lower positive affect and that a fragmented night of sleep resulted in more usage of distraction. In line with our hypothesis, a fragmented night of sleep resulted in more usage of rumination. Higher rumination also mediated the association between fragmented sleep and higher negative affect. This mediating effect of rumination was not found for positive affect. We did not find support for a mediating role of cognitive reappraisal, distraction, acceptance, suppression, and self-criticism for both positive and negative affect.

Overall, the findings of this study provided limited support for the idea that disrupted sleep reduces the ability to use emotion regulation strategies. However, the finding that rumination mediated the negative association between fragmented sleep and negative affect does align with the literature that shows that disengaging attention from negative thoughts is suggested to be more difficult after a night of poor sleep, as disrupted sleep impairs attentional control (Koster et al., 2011). This finding is also in line with models that demonstrate that rumination and distraction might play an important role in the relationship between poor sleep and more severe mood disruption, such as depression (Gradisar et al., 2022; Kirschbaum-Lesch et al., 2021; Niu & Snyder, 2023; Nolen-Hoeksema, 1991; Palmer et al., 2018). Over a longer time period, it is shown that rumination following poor sleep can predict the onset of depressed mood (Gradisar et al., 2022; Kirschbaum-Lesch et al., 2021; Niu & Snyder, 2023; Palmer et al., 2018). The present study was the first study to show that on a very short time scale, namely with one night of sleep fragmentation, rumination increases and is associated with higher negative affect the next day. Sustained negative affect is a key component of depression (Erk et al., 2010). It might therefore be possible that the mood impairing effects of rumination following poor sleep over a longer time period could lead to the onset of depressive feelings. Next to rumination, distraction is also suggested to play a role within the link between disrupted sleep and depression (Nolen-Hoeksema, 1991). Distraction is suggested to reduce negative mood by taking an individual’s mind off negative feelings and thoughts (Nolen-Hoeksema, 1991). It can serve as a counterpart to rumination (Nolen-Hoeksema, 1991). We found that after a fragmented night of sleep participants reported higher distraction. It could be the case that people used distraction more often due to having more ruminative thoughts after sleep fragmentation. However, regarding the mood repairing effects of distraction we found mixed results. On the one hand, we did not find support for an effect of sleep fragmentation on distraction ability. However, there were signs that distraction might be more maladaptive than adaptive with the self-report data, as higher levels of reported distraction were associated with lower levels of positive affect. Future research could look more into the specific mechanism of distraction in the relationship between disrupted sleep and positive and negative affect.

The results suggest that participants selected different emotion regulation strategies, but that ability to implement emotion regulation strategies was not impaired after a night of fragmented sleep. Interestingly, research focusing on people with Major Depressive Disorder (MDD), a mood disorder that is often accompanied by sleeping problems (Mendlewicz, 2009), shows that this group only had impairments in the selection of emotion regulation strategies (e.g. more rumination), but that their ability to implement emotion regulation strategies remained intact (Liu & Thompson, 2017). These findings hint at the idea that emotion regulation selection might be an important factor linking sleep and mood disturbance.

Despite the effect of fragmented sleep on frequency of distraction and rumination, we could not detect an effect on the ability and frequency of use of the other emotion regulation strategies. It might be the case that one night of sleep fragmentation does not substantially impact emotion regulation ability and frequency. So far, there are a couple of studies with a sleep deprivation manipulation of one night in a healthy sample finding limited support for the effect of sleep on emotion regulation ability (Reddy et al., 2017; Shermohammed et al., 2020; Zhang et al., 2019). Effects of poor sleep quality are cumulative (Van Dongen et al., 2003). So, several nights of poor sleep might show a stronger effect of emotion regulation ability and frequency than one poor night. For instance, the study of Mauss et al. (2013) demonstrated that sleep quality over the last week showed a stronger association with cognitive reappraisal ability than sleep in the past 24 h. Next to that, participants slept on average about 6.5 h in the control night. This is average for a college student population, but it is still not close to the recommended sleep duration time of 8 h (Van Dongen et al., 2003). Multiple weeks of sleeping around 6 h is associated with impaired cognitive functioning (Van Dongen et al., 2003). It might therefore be possible that there was a floor effect due to an already sleep deprived population, as the emotion regulation ability scores showed little mood repairing effect in both the control and sleep fragmentation condition. Even though the sleep manipulation was successful, the difference of sleep in the control and sleep manipulation condition could be tested in a target population with a higher average of hours of sleep and by deploying partial sleep deprivation or fragmentation procedures of multiple nights.

The absence of effects of fragmented sleep on emotion regulation ability and frequency might also be explained by the fact that the manipulation of sleep was not under precise control in this study. A limitation of this study is that it is unknown what part of the sleep cycle was interrupted by our sleep manipulation. Every night, an individual goes through four to six sleep cycles of light, deep, and REM sleep (Feinberg, 1974). REM and deep sleep seem to play an important role in the link between poor sleep and emotion dysregulation (Riemann et al., 2012; Roepke & Ancoli-Israel, 2010). During REM sleep, emotional networks (i.e. limbic and paralimbic areas) are activated that help us regulate emotional events (Riemann et al., 2012). Sufficient deep sleep is important for good functioning of the prefrontal cortex (PFC) (Roepke & Ancoli-Israel, 2010). As stated before, impaired higher order cognitive functions of the PFC might impair adaptive emotion regulation and increase maladaptive emotion regulation (Muzur et al., 2002; Yoo et al., 2007). People cyclically go through light, deep, and REM sleep every 90–120 min (Feinberg, 1974). In the present study, participants were woken up every 80 min. It is unknown which sleep stages were mainly disrupted. Emotion regulation ability and frequency might have been less disrupted if participants were mainly woken up during light sleep (Riemann et al., 2012). Future research could assess sleep via polysomnography to underpin the role of REM and deep sleep in adaptive and maladaptive emotion regulation.

Another limitation of this study is that the majority of participants was female. This limits the generalisability of the findings, as men and women differ in which specific emotion regulation strategies they mainly use (e.g. Nolen-Hoeksema & Aldao, 2011) and with their sleep, as women report more sleeping problems and often report stronger effects of sleep on their functioning (Krishnan & Collop, 2006). Furthermore, we used instructions derived from previous studies (i.e. Campbell-Sills et al., 2006; Sheppes & Meiran, 2007; Troy et al., 2018). However, in the current study, participants were performing the task at home. This uncontrolled, distraction-prone environment might have affected the results.

Another possible limitation of this study could be that we assessed the impact of sleep fragmentation on subjective ratings of emotion regulation in the experimental task. In the study of Zhang et al. (2019) participants reported adaptive emotion regulation, while objective, neurophysiological measures showed the opposite. It was suggested that there might be a dissociation between subjective ratings and objective, neurophysiological measures. This discrepancy between subjective and objective measures after sleep deprivation is also apparent in other studies showing that sleep deprived individuals often underestimate the impact of sleep deprivation on their cognitive or attentional abilities (Van Dongen et al., 2003). Future studies are needed to assess whether such discrepancy is apparent for emotion regulation and how this might impact emotional well-being.

To our knowledge, the present study is the first one that gains insights into the effects of fragmented sleep on both the ability and the use of emotion regulation. It highlights the role of specific emotion regulation strategies in the relation between fragmented sleep and mood and demonstrated that people ruminate and distract themselves more after a night with fragmented sleep. It also supports the notion that improving sleep quality might be an important target for treatment of mood disturbance by reducing the selection of maladaptive emotion regulation strategies. Future research could assess the effect of longer partial sleep deprivation manipulations on different measures of emotion regulation strategies by assessing both emotion regulation and sleep objectively. Empirical research on the effect of fragmented sleep on emotion regulation is still limited and more research is necessary to make solid conclusions about the effects of fragmented sleep on emotion regulation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study will be openly available in https://osf.io/zb4he/.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.