Content uploaded by Elaina Bolinger
Author content
All content in this area was uploaded by Elaina Bolinger on Nov 22, 2018
Content may be subject to copyright.
Contents lists available at ScienceDirect
Neuropsychologia
journal homepage: www.elsevier.com/locate/neuropsychologia
Sleep divergently affects cognitive and automatic emotional response in
children
Elaina Bolinger
a
, Jan Born
a,b,⁎
, Katharina Zinke
a,⁎⁎
a
Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, 72076, Germany
b
Centre for Integrative Neuroscience, University of Tübingen, 72076, Germany
ARTICLE INFO
Keywords:
Development
Children
Sleep
Emotion
EEG
Memory
ABSTRACT
Sleep enhances memory for emotional experiences, but its influence on the emotional response associated with
memories is elusive. Here, we compared the influence of nocturnal sleep on memory for negative and neutral
pictures and the associated emotional response in 8–11-year-old children, i.e., an age group with heightened
levels of emotional memory-related sleep features. During all sessions, emotional responses as measured by
subjective ratings, the late positive potential of the EEG (LPP) and heart rate deceleration (HRD) were recorded.
Sleep enhanced picture memory. Compared to dynamics across wakefulness, sleep decreased the emotional
response in ratings and the LPP, while increasing the emotional response in HRD. We conclude that sleep
consolidates immediate emotional meaning by enhancing more automatic emotional responses while con-
currently promoting top-down control of emotional responses, perhaps through strengthening respective neo-
cortical representations.
1. Introduction
Whereas research has revealed a benefit of sleep on emotional
memory (e.g. Hu et al., 2006;Nishida et al., 2009;Wagner et al., 2001),
sleep's influence on the emotional reactivity associated with a memory
has produced complex and seemingly disparate results. Sleep has been
shown to increase (Wagner et al., 2002), decrease (Pace-Schott et al.,
2011; van der Helm et al., 2011), and preserve (Baran et al., 2012;
Pace-Schott et al., 2011; Werner et al., 2015) the emotional tone as-
sociated with memories. Interestingly, a central role of rapid eye
movement (REM) sleep has consistently emerged, leading to two con-
flicting theories on its role in emotional processing. The Sleep to Forget,
Sleep to Remember (SFSR) theory proposes that REM sleep specifically
decreases emotional tone by providing an opportunity to process
emotional memories within a neurochemical environment that supports
memory consolidation in an emotionally neutral setting (Walker,
2009). The SFSR theory has received some support, with work sug-
gesting REM physiology is linked to decreases in emotional tone (van
der Helm et al., 2011). The second theory proposes that REM sleep
preserves emotional tone through consolidation of emotional salience
(Baran et al., 2012; Pace-Schott et al., 2011; Werner et al., 2015). In this
framework, REM sleep is assumed to strengthen the link between a
stimulus and its associated immediate emotional meaning (i.e., aversive
or safe), thereby leading to more robust emotional responses to the
stimulus. Emotional salience consolidation (ESC) might be associated
with the enhanced consolidation of emotional memory seen across
sleep.
Emotional scenarios elicit both physiological and behavioral re-
sponses. A prominent physiological response is the tendency of heart
rate to decrease during confrontation with an emotionally relevant
stimulus (Bradley et al., 2001). Encountering emotional stimuli also
elicits an enhanced central nervous response, which can be measured as
an increase in event-related response amplitude in the EEG (i.e. the late
positive potential at Pz, see Cuthbert et al., 2000). While heart rate
deceleration is more automatic in its features, in that it is preattentively
triggered and directly controlled by subcortical structures like the
amygdala and brainstem (Öhman et al., 2000), the late positive po-
tential (LPP) as well as the subjective emotional response (as assessed
by ratings) are modulated by an interplay between the amygdala and
prefrontal cortex (PFC, Hajcak et al., 2010;Ochsner et al., 2012). Im-
portantly, top-down cognitive processes readily influence both sub-
jective ratings and the LPP (Dennis and Hajcak, 2009; Hajcak et al.,
2010).
Sleep's influence on emotional tone might depend on the nature of
the emotional response considered, and more specifically, on the degree
to which it is influenced by cognitive processes. The level to which the
https://doi.org/10.1016/j.neuropsychologia.2018.05.015
Received 8 November 2017; Received in revised form 9 May 2018; Accepted 17 May 2018
⁎
Correspondence to: Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, Otfried-Müller-Str. 25, 72076 Tübingen, Germany.
⁎⁎
Correspondence to: Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, Silcherstr. 5, 72076 Tübingen, Germany.
E-mail addresses: jan.born@uni-tuebingen.de (J. Born), katharina.zinke@uni-tuebingen.de (K. Zinke).
Neuropsychologia 117 (2018) 84–91
Available online 19 May 2018
0028-3932/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
T
HRD differentially responds to emotional compared to neutral stimuli is
preserved over a nap (Pace-Schott et al., 2011) and HRD responses
decrease after a night of sleep in an emotionally-independent manner
(Cunningham et al., 2014). Thus, more automatically triggered emo-
tional responses might be preserved by sleep. However, more complex
emotional response measures, which recruit cognitive processes to a
greater extent, might become more accessible to top-down control and
thereby decrease over sleep. While some evidence from emotional
ratings supports this notion (van der Helm et al., 2011), other studies
examining sleep on subjective ratings measures have been contra-
dictory (Baran et al., 2012; van der Helm et al., 2011; Wagner et al.,
2001). The LPP, as another measure integrating emotional top-down
control, has thus far only been investigated in one sleep study where it
was related to the emotional enhancement in memory after REM-rich
sleep (Groch et al., 2013).
Children are an excellent model of enhanced emotional processing
during sleep as they exhibit heightened levels of a number of sleep
features which have been associated with emotional memory proces-
sing in adults (i.e. REM & NonREM theta and delta power: Kurth et al.,
2010;Nishida et al., 2009, and NonREM spindle density: Clawson et al.,
2016;Kaestner et al., 2013) and likewise a greater benefit of sleep on
emotional memory consolidation than adults (Prehn-Kristensen et al.,
2013). What is more, they tend to have more emotional sleep experi-
ences as evidenced by increased nightmare and night terror frequency
(Simonds and Parraga, 1982). Though emotion robustly influences
subjective ratings, LPP, HRD, and memory in children (Leventon et al.,
2014), the influence of sleep on the emotional tone associated with
memories has not yet been reported in this age group.
We investigated how sleep influences emotional reactivity in 8–11
year olds. Children encoded negative and neutral pictures and then
either stayed awake or slept in the laboratory during a 10-h retention
period. They then saw the original along with new pictures and were
asked to report if they recognized each picture. During encoding and
recognition testing, subjective (ratings of valence and arousal), central
nervous (LPP of the brain evoked response) and autonomic nervous
(HRD) system measures of the emotional response were assessed,
whereby the emotional response was defined as the response difference
between negative and neutral pictures. This metric provides a measure
of emotional reaction strength that is corrected for possible differences
in baseline reactivity (e.g. Coan and Allen, 2007;Phelps et al., 2001).
We expected that, compared to wake, measures of emotion that are
subject to greater cognitive control would decrease after sleep, while
measures of emotion that are less influenced by cognitive control would
be preserved. In addition, we hypothesized that consolidative influ-
ences of sleep (as measured by emotional response) would be specifi-
cally linked to REM sleep.
8:00h
8:00h 20:00h
Old or New?
Encoding Recognition
Old or New?
Recognition
Wake Condition
Encoding
Day
Awake
Sleep Condition
Encoding
Recognition
Night
Asleep
C) Experimental Design
A) Change in Emotional Response B) Recognition Accuracy
D) Emotional Picture Task
*
*
**
*
Fig. 1. A) Sleep and wake intervals differentially affected the change in emotional response from encoding to recognition. Whereas sleep decreased emotional
response in valence ratings, the ratings did not change across the wake interval. (Note that the y-axis has been inverted to facilitate interpretation of the valence
response: a positive change is a decrease in emotional response, n= 14). The emotional response of the late positive potential (LPP) of the event-related EEG at Pz
increased over a day of wake but was preserved over a night of sleep. Therefore, relative to wake, sleep decreased the emotional response in LPP (n= 12). The
emotional response in heart rate deceleration (HRD) decreased over the wake interval but was stable after a night of sleep. Thus, relative to wake, sleep increased the
emotional response in HRD (n= 15). B) Sleep enhanced recognition accuracy in general, while wake led to better memory for negative compared to neutral pictures
(n= 16). C) A recognition task was used in a within-subjects crossover design to examine the influence of post-encoding sleep vs wakefulness on memory and
associated emotional responses in 8–11 year old children. During Encoding, children viewed emotional pictures in either the evening (Sleep condition) or the
morning (Wake condition). After a 10-h retention interval, children performed a memory task during the Recognition session. During both sessions, subjective
ratings, the LPP of the EEG, and HRD were measured. D) During Encoding, children saw negative and neutral images that they rated with regard to subjective valence
and arousal using the SAM. During Recognition, children saw the original images as well as negative and neutral distractors. Trials were identical to the Encoding
session, with the exception of the additional question “Is this picture old or new?”which was asked after subjective ratings were performed. * represents p< .05.
E. Bolinger et al. Neuropsychologia 117 (2018) 84–91
85
2. Methods
2.1. Participants
Sixteen healthy children without sleep disturbances between the
ages of 8 and 11 years (mean ± SD: 9.25 ± 1.06 years, 8 females)
participated in the experiment. Sample size was based on a previous
within-subjects study addressing the influence of sleep on emotional
response in adults (Groch et al., 2013). All participants were physically
and mentally healthy according to parental report and the Child Be-
havior Checklist (Achenbach, 1991; Arbeitsgruppe Deutsche Child
Behavior Checklist, 1998), and had normal or corrected-to-normal vi-
sion. Participants had an average digit span forward of 5.57 ± 1.26
and backward of 3.94 ± 1.18 according to the digit span subscale of
the Wechsler Intelligence Scale (Petermann and Petermann, 2011).
Participants did not exhibit sleep disturbances according to parent and
child reports. The study was approved by the University of Tübingen
medical faculty ethics committee and both parents and children gave
informed verbal and written consent.
2.2. Experimental design and procedures
Recognition memory and emotional responses to pictures were
compared between a Sleep and a Wake condition, following a within
[HYPHEN]subjects crossover design where each child was tested on
both conditions with an interval of at least 12 days between each
condition. The order of conditions was randomly assigned to the chil-
dren, with 9 children of the final sample of 16 children starting with
Sleep condition. Each condition started with an Encoding session,
where the task pictures were presented the first time, followed by a 10-
h retention period in which the child either slept or was awake, and
then a Recognition session in which the original pictures were pre-
sented along with distractor pictures which had not been presented in
the Encoding session (Fig. 1C and D). In both sessions, valence and
arousal ratings (collected with the self-assessment manikin, SAM;
Bradley and Lang, 1994), the LPP, and HRD responses to the pictures
were measured.
Each child participated in an adaptation session before the experi-
ment proper. For this session, children and parents arrived at the lab at
18:00 h. Children filled out a questionnaire about their sleeping habits
and completed the digit span task. They also practiced the control tasks,
which were measured before each experimental condition, including a
5-min version of the Psychomotor Vigilance Test (PVT, Roach et al.,
2006) to assess vigilance, the Stanford Sleepiness Scale (SSS, Hoddes
et al., 1973) to assess sleepiness, the Positive and Negative Affect
Schedule (PANAS; Watson et al., 1988) to assess mood, and a five-point
Likert-scale assessing motivation (1 = absolutely not motivated, 5 =
extremely motivated). During this time, parents were asked to fill out
the questionnaires about the child's health and sleep behavior. Children
were then prepared for bed and electroencephalographic (EEG), elec-
trocardiographic (ECG), electromyographic (EMG), and electrooculo-
graphic (EOG) recordings. They then slept for ~9 h to ensure accli-
mation to the conditions of the sleep laboratory.
For the Sleep condition, children and parents arrived ~3 h before
the child's habitual bedtime. Children prepared for bed and recording
electrodes were applied (EEG, ECG, EMG and EOG). Children per-
formed the control and main tasks of the Encoding session and then
went to sleep for ~9 h. Upon waking, the children ate a light breakfast
while electrode impedances were checked. It was ensured that the
children were awake at least 30 min before beginning the Recognition
session, in order to avoid any influences of sleep inertia on task per-
formance.
For the Wake condition, children and parents arrived at the lab at
~7:30 h. The procedure was then identical to the Sleep condition with
the exception of a day of wake during the retention period. Children
wore an activity-tracking device (Actiwatch 2, Philips Respironics) to
ensure that they did not sleep during the wake period and families were
asked to abstain from strenuous activities over the course of the day.
Participants ate a light meal before starting the Recognition session.
2.3. Materials and task
A set of 444 pictures (taken from the International Affective Picture
System, IAPS; Lang et al., 2008, and other studies; Jackson et al., 2005;
Prehn-Kristensen et al., 2009) was used in this study to create two
unique versions of the task (List A and B, order balanced) for use in each
condition. Each task version consisted of 72 negative targets and 72
neutral targets, in addition to 36 negative foils and 36 neutral foils
which were used as distractor pictures during recognition (Fig. 1D). An
additional six positive pictures were included in each list with the sole
purpose of maintaining motivation and were not included in the ana-
lyses. A pilot study, wherein a larger cohort of children (n= 41) used
the Self-Assessment Manikin (SAM, Bradley and Lang, 1994) to rate the
pictures, confirmed that children generally rated the negative pictures
as more negative and more arousing (mean ± sem valence ratings:
negative = −1.68 ± 0.75, neutral = 0.78 ± 0.82, t(40) = −16.48,
p< .001; arousal ratings: negative = 4.58 ± 1.47, neutral
= 1.77 ± 1.18, t(40) = 12.32, p< .001).
During the Encoding session, target pictures were presented in a
pseudo-random order wherein consecutive iterations of picture cate-
gory (i.e. negative and neutral) were limited to three. A single trial
consisted of the following procession: a fixation circle (displayed for a
randomized period between 1 and 2 s), the picture (displayed for 1.5s),
a blank screen (displayed for 6 s), valence rating (i.e. valence SAM with
the question: “How unpleasant or pleasant did you feel while looking at
the picture?”, coded from −5 to 5), arousal rating (i.e. arousal SAM
with the question: “How calm or nervous did you feel while looking at
the picture?”, coded from 1 to 9), and finally an inter-trial interval of 2 s
(Fig. 1D). At the Recognition session, target and distractor pictures
were intermixed. The trial procession was identical except for the ad-
ditional question (following the ratings): “Is this picture old or new?”to
which children responded with either “Old”or “New.”Participants
were instructed to rate all pictures based on their momentary feelings
using a computer mouse and to not try to remember how they had rated
a picture if they had seen it before. The task was presented using E-
Prime®3.0 (Psychology Software Tools, Inc., Sharpsburg, Pennsylvania,
USA).
2.4. Behavioral analysis
Recognition memory scores were calculated as recognition accu-
racy, i.e. hit rate minus false alarm rate, for negative and neutral pic-
tures, separately. Average ratings for valence and arousal were calcu-
lated according to trial type, i.e. negative target at encoding, neutral
target at encoding, successfully remembered negative target at re-
cognition (negative hits), and successfully remembered neutral target at
recognition (neutral hits). Emotional response for all measures was
calculated as the difference between the response to negative images
and the response to neutral images at each session. This measure re-
flects the strength of an emotional response while correcting for pos-
sible differences in baseline reactivity. The analysis was limited to trials
where the picture was correctly remembered (hits) in recognition ses-
sions.
2.5. Electrophysiological recordings and analysis
Electrophysiological data was collected using Brain Vision hardware
and software (Brain Products GmBH, Gilching, Germany). During all
sessions, EEG was recorded at F3, Fz, F4, C3, Cz, C4, P3, Pz, and P4
(referenced to linked mastoids with FC1 and FC2 as grounds) according
to the 10–20 system using Ag/Cl electrodes. Two diagonally placed
EOG electrodes were used to measure eye movements. ECG bipolar
E. Bolinger et al. Neuropsychologia 117 (2018) 84–91
86
electrodes were placed on the lower left and upper right chest. Signals
were sampled at 500 Hz, and EEG electrode impedance was kept below
5kΩ.
Event-related potential (ERP) processing was performed with the
Brain Vision Analyzer 2.0 Software (Brain Products GmBH, Gilching,
Germany). EEG and EOG were first filtered between 0.3 and 35 Hz
using a zero-phase Butterworth filter (24 dB/oct) with a 50 Hz notch.
Trials were segmented from −200 ms before picture onset to 1500 ms
after picture onset and then subjected to linear detrending according to
Hennighausen et al. (Hennighausen et al., 1993). Baseline correction of
the ERP segments was performed using the 200 ms before picture onset.
ERP segments were rejected as artifacted when 1) a gradient > 40 µV/
ms was detected, 2) a voltage difference > 150 µV occurred within the
1700-ms segment, 3) a period of low activity (< .5 µV) was detected for
a period > 100 ms, or 4) an ocular artifact was apparent by visual in-
spection. ERP analyses concentrated on the LPP whose amplitude was
taken as the peak voltage at Pz between 600 ms and 1000 ms after the
picture onset (using the average voltage in this latency bin yielded es-
sentially the same results and are not reported here).
To determine heart rate deceleration (HRD), R waves of the QRS
complexes were first marked using a slope-based detection algorithm in
the Brain Vision Analyzer 2.0 Software. Trials were extracted according
to type and then visually inspected to exclude artifacted data. The
maximum R-R interval in the 5-s interval following picture onset was
subtracted from the mean R-R interval during the 1.5-s baseline period
before picture onset. LPP and HRD responses were averaged according
to trial type (negative and neutral targets at encoding, negative and
neutral hits at recognition,see Fig. 2 for traces of the original LPP and
HRD signals).
Sleep EEG and EOG were filtered in the same manner as the ERP
data. EMG was filtered with the same parameters but with a higher
bandpass frequency range (5–100 Hz). Sleep recordings were scored
according to standard criteria (Rechtschaffen, 1968) to obtain the time
spent in different sleep stages (wake, stages 1, 2, slow wave sleep
[SWS], and REM sleep in minutes and percentage of total sleep time
[TST]). The Brain Vision Analyzer 2.0 Software was used to process and
calculate power in the delta and theta bands during NonREM (S2 and
SWS) and REM, respectively. Artifact-free sleep epochs were cut into 6-
second segments with 2 s of overlap. A fast fourier transform using a
symmetric Hanning window was used to calculate power with a 0.2 Hz
resolution. Relative EEG power (%) at Fz (as done in previous work:
Prehn-Kristensen et al., 2013) was then calculated for each subject by
dividing the power in each frequency bin by the total power in the
spectrum for a specific sleep stage and multiplying by 100. Because
frequency band ranges can vary widely from person to person (e.g.
Klimesch et al., 1998;Pfurtscheller and Lopes da Silva, 1999), we fine-
tuned the determination of frequency ranges for the bands of interest by
visually determining for each participant the peak frequency for the
delta and spindle bands for NonREM sleep as well as theta during REM
sleep, and centering the band limits around this peak (Supplementary
Fig. S2). This procedure resulted in the following average bands:
0.4–2 Hz for the NonREM sleep delta band, 10.5–13.2 Hz for the
NonREM sleep spindle band, and 4.6–6 Hz for the REM sleep theta
band.
To detect discrete spindles, EEG was bandpass filtered within a
subject-specific spindle frequency range and spindle events were de-
tected according to an established thresholding algorithm (see Gais
et al., 2002 for details). Subject-specific spindle bands were visually
identified as the peak within the 9–15 Hz range (see Kurdziel et al.,
2013). Spindle counts were averaged over C3, Cz, and C4, similar to
Kaestner et al. (2013). Spindle density was calculated as the average
number of spindles per 30-s epoch for S2 sleep and SWS.
2.6. Statistical analyses
Physiological emotional response data from one participant had to
be excluded due to a technical failure. Ratings from two participants
were excluded from analyses because of performance values ex-
ceeding ± 2 standard deviations from the sample mean, suggesting that
they may have responded randomly to the subjective rating portion of
the task. It was nevertheless confirmed that they adequately performed
the memory task (i.e. recognition accuracies > 65%). The sleep elec-
trophysiology from frontal electrodes in one participant had to be ex-
cluded due to technical problems with the recordings. Data of 3 parti-
cipants were excluded from LPP analyses due to values that
exceeded ± 2 standard deviations from the mean.
Following the experimental within-subject cross-over design of the
study, changes in memory and emotional response were analyzed using
repeated measures analysis of variance (ANOVA) with the within-sub-
jects factor Sleep (Wake vs. Sleep). The influence of sleep on
Fig. 2. The influence of sleep on physiological measures of emotion. A) and B) show grand average LPP traces at Pz across wake and sleep, respectively. C) and D)
show the change in R-R interval relative to the 1.5 s baseline period before image onset (post onset minus pre onset). Note that a positive change R-R interval
represents a deceleration in heart rate and that the y-axis has been inverted to facilitate interpretation of the heart response. Black vertical lines represent image
onset.
E. Bolinger et al. Neuropsychologia 117 (2018) 84–91
87
recognition accuracy was addressed using an additional Emotion cate-
gory (negative vs. neutral) within-subjects factor. For the analysis of
emotional response, a within-subjects Session factor (Encoding vs.
Recognition) was used in addition to the Sleep factor. Bonferroni-cor-
rected paired t-tests were used to follow up significant ANOVA effects
(new α-level for comparison: 0.05/2 tests = 0.025). To determine the
specific nature of the influence of sleep, change scores in response over
the retention interval (defined by the difference between the respective
response at recognition minus the response at encoding) were calcu-
lated separately according to picture class (negative and neutral)
whenever a significant interaction between Sleep and Session was
found (Bonferroni-corrected paired t-tests as described above).
Pearson's correlations were used to test for relationships between
sleep features that have been associated with emotional preservation
(REM% (Wagner et al., 2001), REM theta power (Prehn-Kristensen
et al., 2013), S2 and SWS spindle density (Kaestner et al., 2013)) and
measures which did not exhibit a change in emotional response over
sleep (LPP and HRD, calculated as emotional response at recognition
minus emotional response at encoding) as well as emotional differ-
entiation in memory (i.e. the difference between negative recognition
accuracy and neutral recognition accuracy). We also tested for whether
emotional response measures which were preserved over sleep corre-
lated with emotional differentiation following sleep or wake. Bonfer-
roni correction was applied such that the required α-level for sig-
nificance was .05/16 tests = 0.003.
Influences of sleep on control variables were addressed using a Sleep
x Session ANOVA. Because this study was focused on possible influ-
ences of sleep on emotion and memory, follow up correlations were
calculated between control variables exhibiting Sleep effects (main ef-
fects or interactions) and the independent variables of interest (emo-
tional memory and emotional response for each measure).
3. Results
3.1. Recognition memory
In general, participants correctly recognized more target pictures
when they had slept after the encoding session than when they had
stayed awake (Sleep main effect for recognition accuracy: F(1,14)
= 8.52 p= .01, ƞ
p2
= 0.38, Fig. 1B). Also, negative images were gen-
erally better recognized than neutral ones (Emotion main effect: F(1,14)
= 17.51 p= .001, ƞ
p2
= 0.56). However, whereas better recognition
accuracy for negative compared to neutral pictures was found after
staying awake (p= .001), this was not the case for the Sleep condition
(p= .14, Sleep x Emotion interaction: F(1,14) = 6.17, p= .026, ƞ
p2
= 0.31), possibly reflecting a ceiling effect of the generally enhanced
recognition performance in the Sleep condition.
3.2. Subjective emotional response
Children rated negative images as more negative and more arousing
at both the Sleep (valence ratings: negative = −1.88 ± 0.78, neutral
= 0.56 ± 0.55, arousal ratings: negative = 5.8 ± 1.36, neutral =
2.44 ± 1.31) and Wake (valence ratings: negative = −1.80 ± 0.79,
neutral = 0.47 ± 0.40, arousal ratings: negative = 5.43 ± 1.68,
neutral = 2.39 ± 1.02) encoding sessions (all ps < .001). Sleep de-
creased the emotional response (response to negative images minus
response to neutral images) as measured by valence ratings of the
successfully remembered pictures (p= .007). In contrast, valence rat-
ings remained unchanged across a retention interval containing wake
(p= .53, see Fig. 1A for visualized Sleep x Session interaction: F(1,13)
= 4.78, p= .048, ƞ
p2
= 0.27; Sleep main effect: ns; Session main effect:
F(1,13) = 5.32, p= .038, ƞ
p2
= 0.29; see Fig. S3 in the Supplementary
Material for emotional response at encoding and recognition, sepa-
rately). Calculating change scores (response at recognition minus re-
sponse at encoding) separately for neutral and negative images revealed
that this interaction was driven by the fact that sleep produced a less
negative rating of negative images (mean difference ± sem:
+0.19 ± 0.13, see Supplementary Fig. S4) and, simultaneously, a
more negative valence rating of neutral images (−0.18 ± 0.06, t
(13) = 3.17, p= .007) whereas no such changes were observed across
the wake retention interval (negative: −0.02 ± 0.10, neutral:
−0.10 ± 0.08, p= .53). Arousal ratings exhibited a similar pattern of
decreased emotional response across Sleep compared with the Wake
condition, albeit less robustly (Sleep x Session interaction: F(1,13)
= 4.30, p= .058, ƞ
p2
= 0.25, Sleep & Session main effects: ns). The
influence of sleep on rating behavior did not appear to be driven by
circadian effects, as the subjective emotional response did not sig-
nificantly differ between the wake encoding session, which took place
in the morning, and sleep encoding session, which took place in the
evening (valence: p= .16; arousal: p= .11).
3.3. LPP emotional response
The emotional response as measured by the LPP of the event related
potential at Pz remained stable across sleep (p= .59) but increased
over the course of the wake period (p= .008; Sleep x Session interac-
tion: F(1,11) = 5.34, p= .041, ƞ
p2
= 0.33; Sleep and Session main
effects: ns). Thus, compared with the dynamics across the wake period,
sleep led to a relative decrease in the LPP emotional response (see
Fig. 1A for visualized Sleep x Session interaction). Correspondingly,
separate change scores for negative and neutral pictures indicated that
the LPP slightly increased across sleep regardless of emotional picture
category (negative: 2.81 ± 1.61 μV, neutral: 3.78 ± 1.10 μV,
p= .60), whereas the wake period produced a particularly strong in-
crease in LPP to negative pictures (6.9 ± 0.85 μV) as compared to
neutral pictures (2.95 ± 1.00 μV, t(11) = 3.21, p= .008). The emo-
tional response tended to be smaller at encoding in the morning (Wake
condition) than at encoding in the evening session (Sleep condition;
p= .063) suggesting that the increase in emotional response in LPP
across sessions in the Wake condition may partially reflect a circadian
influence (see Fig. S3 for an overview of the emotional responses of the
LPP at encoding and recognition, as well as additional LPP analyses in
the Supplementary material). To explore this possibility, we removed
the three participants with the lowest emotional response at the wake
encoding session in order to match emotional response at encoding (t
(8) = −1.49, p= .175). This analysis revealed essentially the same
pattern with a statistical trend for the Sleep x Session interaction (F(1,8)
= 3.93, p= .083, ƞ
p2
= 0.33), which suggests that time of day alone
did not exclusively drive the resultant data patterns.
3.4. HRD emotional response
In contrast to subjective ratings and LPP, the HRD emotional re-
sponse decreased across wake (p= .003) but not across sleep (p= .67;
Sleep x Session interaction: F(1,14) = 5.20 p= .039, ƞ
p2
= 0.27, Sleep
and Session main effects: ns). Thus, compared with the dynamics across
the wake period, sleep led to relative increase in HRD emotional re-
sponse (see Fig. 1A for visualized Sleep x Session interaction). Analysis
of separate change scores for negative and neutral pictures revealed
that this interaction was driven by a decrease in HRD response to the
negative pictures (mean ± sem: −14.09 ± 6.81 ms, t(14) = −3.54,
p= .003, see Supplementary Fig. S4) but not neutral pictures
(6.24 ± 6.85 ms) across the wake retention interval, whereas across
sleep HRD decreased regardless of emotional picture class (negative:
−15.73 ± 10.72 ms, neutral: −18.97 ± 8.23 ms, p= .91). HRD re-
sponses were comparable for the Encoding sessions of the Sleep and
Wake conditions ruling out substantial circadian effects (p= .13; see
Fig. S3 in the Supplementary material).
E. Bolinger et al. Neuropsychologia 117 (2018) 84–91
88
3.5. Correlations between emotional response, sleep parameters and
memory
The overnight change in the emotional HRD response (emotional
response to hits at recognition minus emotional response at encoding)
positively correlated with the degree of emotional memory differ-
entiation after sleep (r= 0.75, p= .001): participants who showed a
relative increase in HRD emotional response also showed a bigger dif-
ference between negative and neutral recognition accuracy. No other
correlations survived multiple testing correction at α= 0.003.
3.6. Sleep parameters and control variables
Sleep scoring of the polysomnographic recordings from the experi-
mental night indicated that the children slept normally (Supplementary
Fig. S1). In total, they slept around 8.58 h. Participants spent approxi-
mately 19.9 min, 162.8 min, 225.1 min and 99.6 min in sleep stages 1,
2, SWS, and REM, respectively. They entered SWS after ~4.5 min and
REM after ~95 min. Participants were awake for approximately 6.2 min
after sleep onset.
Results from control tests are summarized in Supplementary
Table 1. Motivation, vigilance, negative mood, and sleepiness in the
Encoding and Recognition sessions were not differentially influenced by
the Sleep and Wake conditions (ps > .26, for the respective main and
interaction effects). Motivation was higher during the Encoding ses-
sions than the Recognition sessions (Session main effect: F(1,15)
= 4.77, p= .045). Positive mood (PANAS-Positive) exhibited a similar
trend (Session main effect: F(1,15) = 3.99, p= .064), and was higher
during the Sleep than Wake condition (Sleep main effect: F(1,15)
= 7.82, p= .014). There were no consistent correlations between
PANAS-Positive scores and any of the emotional response or memory
measures (ps≥.059), excluding any substantial contributions of this
factor to the effects of sleep on emotional responses or recognition
performance.
4. Discussion
Beyond confirming that sleep enhances memory for pictures (e.g.
Baran et al., 2012), our study in ~10 year old children revealed a
distinct pattern of sleep effects on the emotional response associated
with these memories. Compared with a period of daytime wakefulness,
nocturnal sleep decreased the emotional response as measured by be-
havioral ratings (valence) and the LPP (by preserving a response that
would otherwise increase over wake) while increasing the emotional
response as measured by HRD (by preserving a response that would
otherwise decrease over wake). It therefore appears that sleep decreases
responses that are more subject to cognitive control, while increasing
responses that are generated more automatically (i.e. less sensitive to
top-down influences).
This study provides support for the body of research showing that
sleep enhances memory (Diekelmann and Born, 2010), though it should
be considered that without an immediate recognition test after en-
coding it is impossible to tell whether encoding strength was influenced
by time of day. It should be noted, however, that many studies which
included immediate recognition sessions in order to control for circa-
dian influences show a similar improvement in memory performance
after sleep (e.g. Prehn-Kristensen et al., 2013). Unlike previous studies
(Hu et al., 2006; Nishida et al., 2009; Prehn-Kristensen et al., 2009)we
did not find a preferential enhancement of negative over neutral stimuli
by sleep. This discrepancy likely reflects a ceiling effect, as average hit
rates for negative pictures in the Sleep condition reached a maximum of
91%.
Our finding that sleep decreased emotional response in valence
ratings in children, together with previous findings in adults from van
der Helm et al. (2011), suggests that sleep leads to a decrease in the
explicit judgment of the aversiveness of a stimulus. Overall, however,
studies of the effects of sleep on valence ratings to negative stimuli
reveal a rather heterogeneous picture (Baran et al., 2012; Wagner et al.,
2002), likely reflecting the complex integration of physiological signals
and concurrent behavioral goals that are inherent to cognitive emo-
tional appraisals (Schachter and Singer, 1962). Such aspects of cogni-
tive control may be more sensitive to subtle changes in experimental
context or the composition of sleep.
The effects of sleep on valence ratings were paralleled by a relative
decrease in the emotional LPP response across sleep when compared to
wakefulness. Specifically, the emotional LPP response remained at a
rather constant level across sleep, while increasing across wake.
Though it should be emphasized that circadian factors may have par-
tially influenced this pattern, these results are in accordance with
functional magnetic resonance imaging (fMRI) work reporting a de-
crease in emotional reactivity as measured by amygdala activity across
sleep (van der Helm et al., 2011). Importantly, both the LPP and
amygdala activity have been linked to emotional reactivity and pre-
frontal cortex-driven emotion regulation in children and adults
(Babkirk et al., 2015; Dennis and Hajcak, 2009; Hajcak et al., 2010;
Ochsner et al., 2012). The relatively diminished LPP after sleep there-
fore likely reflects increased emotion regulation which itself might be a
consequence of a sleep-induced enhancement of the cortical re-
presentation of the stimulus (Gais et al., 2007; Liu et al., 2016;
Takashima et al., 2006).
Opposing the down-regulating effects of sleep on the emotional
response in subjective ratings and LPP, sleep produced a relative en-
hancement in the emotional heart rate deceleration (HRD) response
compared to dynamics across wake. Specifically, sleep was associated
with preservation of emotional HRD response, which otherwise de-
creased across wake. Our findings in children agree with previous
findings in an adult study which likewise showed that HRD in response
to emotional pictures decreases across wake but is preserved across a
nap (Pace-Schott et al., 2011). Change scores also confirmed that the
preserving effect of sleep on HRD accompanied a stimulus category-
independent decrease in HRD across sleep, as seen in adults
(Cunningham et al., 2014).
Interestingly, even though a ceiling effect was seen in our memory
measure, the degree of emotional memory enhancement and the degree
of emotional HRD response preservation across the retention period
were positively correlated in the Sleep but not Wake condition. This
suggests that a similar underlying mechanism is responsible for pre-
serving the emotional HRD response and emotional memory, and in
extension, that the immediate emotional meaning of an event is in-
herently intertwined in the memory representation. Previous work has
shown that the enhanced consolidation of the entirety of an emotional
memory is linked to REM sleep (Menz et al., 2016), however, in our
study respective correlations for REM% and REM sleep theta power
with emotional preservation did not reach significance after correction
for multiple tests.
Our study is unique in showing that rather than exerting a uniform
influence, a night of sleep differentially affects automatic and cognitive
aspects of the emotional response. In combination, the present findings
suggest a scenario in which sleep affects memory and the associated
emotional response via two different mechanisms: On the one hand,
sleep preserves the immediate emotional meaning of events, leading to
preservation of automatic emotional responses like HRD. On the other
hand, sleep generally acts to strengthen declarative components of a
memory in a way that may also make them more accessible to retrieval
and various other cognitive functions. This episodic memory con-
solidation process leads to the strengthening of the neocortical re-
presentation of the event and has been linked to NonREM sleep, as well
as the associated slow wave and spindle activity (Diekelmann and Born,
2010). Indeed, recent work has shown that overnight consolidation of
emotional memories is associated with preservation of autonomic
emotional reactivity and reorganization of memory-associated activity
from hippocampal to neocortical structures (Liu et al., 2016). The
E. Bolinger et al. Neuropsychologia 117 (2018) 84–91
89
strengthening of respective cortical representations might improve the
top-down control of emotional responses, seen here as diminished rat-
ings of negative valence and a relative decrease in the emotional LPP
response after sleep.
Such a twofold influence of sleep - preserving automatic emotional
response and enhancing top-down control of emotional response -
might help to bridge the divergent predictions of the ESC and the SFSR
theories that have been proposed to explain effects of sleep on emo-
tional memory (Baran et al., 2012; Pace-Schott et al., 2011; Walker,
2010; Werner et al., 2015). In support of the ESC theory we found that
sleep increases (via preservation of a response that would otherwise
decrease over wake) emotional responses that are more automatic in
nature (HRD). In support of the SFSR theory, we found that sleep leads
to a relative decrease in more cognitive emotional responses, i.e. those
that may employ top-down regulation mechanisms like subjective rat-
ings and the LPP (via an absolute decrease in valence ratings and pre-
servation of the LPP response which otherwise increases over wake).
Thus, our study shows that rather than acting as a uniform entity,
emotional response emerges from the interaction between automatic
emotional reaction systems and cognitive processes, each of which is
divergently influenced by sleep.
Acknowledgements
This work was supported by the Deutsche Forschungsgemeinschaft
SFB 654 (Plasticity and Sleep). We would like to thank Alexander
Prehn-Kristensen and Christian Wiesner for preparing the task and
analyzing the results of a pilot study, Hong-Viet Ngo for providing the
spindle-detection algorithm code, and Cristin Clar and Lilliam Griselda
Hernandez-Reyes for their help with data collection.
Conflict of interest statement
The authors declare no competing financial interests.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the
online version at http://dx.doi.org/10.1016/j.neuropsychologia.2018.
05.015.
References
Achenbach, T.M., 1991. Child Behavior Checklist. University of Vermont, Burlington, VT.
Arbeitsgruppe Deutsche Child Behavior Checklist, 1998. Elternfragebogen über das
Verhalten von Kindern und Jugendlichen; deutsche Bearbeitung der Child Behavior
Checklist (CBCL/4-18). Einführung und Anleitung zur Handauswertung. Köln:
Arbeitsgruppe Kinder-, Jugend- und Familiendiagnostik.
Babkirk, S., Rios, V., Dennis, T.A., 2015. The late positive potential predicts emotion
regulation strategy use in school‐aged children concurrently and two years later. Dev.
Sci. 18, 832–841.
Baran, B., Pace-Schott, E.F., Ericson, C., Spencer, R.M., 2012. Processing of emotional
reactivity and emotional memory over sleep. J. Neurosci. 32, 1035–1042.
Bradley, M.M., Codispoti, M., Cuthbert, B.N., Lang, P.J., 2001. Emotion and motivation I:
defensive and appetitive reactions in picture processing. Emotion 1, 276.
Bradley, M.M., Lang, P.J., 1994. Measuring emotion: the self-assessment manikin and the
semantic differential. J. Behav. Ther. Exp. Psychiatry 25, 49–59.
Clawson, B.C., Durkin, J., Aton, S.J., 2016. Form and function of sleep spindles across the
lifespan. Neural Plast. 2016.
Coan, J.A., Allen, J.J., 2007. Handbook of Emotion Elicitation and Assessment. Oxford
university press.
Cunningham, T.J., Crowell, C.R., Alger, S.E., Kensinger, E.A., Villano, M.A., Mattingly,
S.M., Payne, J.D., 2014. Psychophysiological arousal at encoding leads to reduced
reactivity but enhanced emotional memory following sleep. Neurobiol. Learn. Mem.
114, 155–164.
Cuthbert, B.N., Schupp, H.T., Bradley, M.M., Birbaumer, N., Lang, P.J., 2000. Brain po-
tentials in affective picture processing: covariation with autonomic arousal and af-
fective report. Biol. Psychol. 52, 95–111.
Dennis, T.A., Hajcak, G., 2009. The late positive potential: a neurophysiological marker
for emotion regulation in children. J. Child Psychol. Psychiatry 50, 1373–1383.
Diekelmann, S., Born, J., 2010. The memory function of sleep. Nat. Rev. Neurosci. 11,
114–126.
Gais, S., Albouy, G., Boly, M., Dang-Vu, T.T., Darsaud, A., Desseilles, M., Rauchs, G.,
Schabus, M., Sterpenich, V., Vandewalle, G., 2007. Sleep transforms the cerebral
trace of declarative memories. Proc. Natl. Acad. Sci. 104, 18778–18783.
Gais, S., Mölle, M., Helms, K., Born, J., 2002. Learning-dependent increases in sleep
spindle density. J. Neurosci. 22, 6830–6834.
Groch, S., Wilhelm, I., Diekelmann, S., Born, J., 2013. The role of REM sleep in the
processing of emotional memories: evidence from behavior and event-related po-
tentials. Neurobiol. Learn. Mem. 99, 1–9.
Hajcak, G., MacNamara, A., Olvet, D.M., 2010. Event-related potentials, emotion, and
emotion regulation: an integrative review. Dev. Neuropsychol. 35, 129–155.
Hennighausen, E., Heil, M., Rösler, F., 1993. A correction method for DC drift artifacts.
Electroencephalogr. Clin. Neurophysiol. 86, 199–204.
Hoddes, E., Zarcone, V., Smythe, H., Phillips, R., Dement, W., 1973. Quantification of
sleepiness: a new approach. Psychophysiology 10, 431–436.
Hu, P., Stylos-Allan, M., Walker, M.P., 2006. Sleep facilitates consolidation of emotional
declarative memory. Psychol. Sci. 17, 891–898.
Jackson, P.L., Meltzoff, A.N., Decety, J., 2005. How do we perceive the pain of others? A
window into the neural processes involved in empathy. NeuroImage 24, 771–779.
Kaestner, E.J., Wixted, J.T., Mednick, S.C., 2013. Pharmacologically increasing sleep
spindles enhances recognition for negative and high-arousal memories. J. Cogn.
Neurosci. 25, 1597–1610.
Klimesch, W., Russegger, H., Doppelmayr, M., Pachinger, T., 1998. A method for the
calculation of induced band power: implications for the significance of brain oscil-
lations. Electroencephalogr. Clin. Neurophysiol. 108, 123–130.
Kurdziel, L., Duclos, K., Spencer, R.M., 2013. Sleep spindles in midday naps enhance
learning in preschool children. Proc. Natl. Acad. Sci. 110, 17267–17272.
Kurth, S., Jenni, O.G., Riedner, B.A., Tononi, G., Carskadon, M.A., Huber, R., 2010.
Characteristics of sleep slow waves in children and adolescents. Sleep 33, 475.
Lang, P., Bradley, M., Cuthbert, B., 2008. International Affective Picture System (IAPS):
Affective Ratings of Pictures and Instruction Manual (Technical Report A-8).
University of Florida, Gainesville, FL.
Leventon, J.S., Stevens, J.S., Bauer, P.J., 2014. Development in the neurophysiology of
emotion processing and memory in school-age children. Dev. Cogn. Neurosci. 10,
21–33.
Liu, Y., Lin, W., Liu, C., Luo, Y., Wu, J., Bayley, P.J., Qin, S., 2016. Memory consolidation
reconfigures neural pathways involved in the suppression of emotional memories.
Nat. Commun. 7, 13375.
Menz, M.M., Rihm, J.S., Büchel, C., 2016. REM sleep is causal to successful consolidation
of dangerous and safety stimuli and reduces return of fear after extinction. J.
Neurosci. 36, 2148–2160.
Nishida, M., Pearsall, J., Buckner, R.L., Walker, M.P., 2009. REM sleep, prefrontal theta,
and the consolidation of human emotional memory. Cereb. Cortex 19, 1158–1166.
Ochsner, K.N., Silvers, J.A., Buhle, J.T., 2012. Functional imaging studies of emotion
regulation: a synthetic review and evolving model of the cognitive control of emo-
tion. In: Ann. N. Y. Acad. Sci. pp. 1251.
Öhman, A., Hamm, A., Hugdahl, K., 2000. Cognition and the autonomic nervous system:
orienting, anticipation, and conditioning. In: Cacioppo, John T., Tassinary, Louis G.,
Berntson, Gary G. (Eds.), Handbook of Psychophysiology, 2nd ed. Cambridge
University Press, Cambridge, pp. 533–575.
Pace-Schott, E.F., Shepherd, E., Spencer, R.M., Marcello, M., Tucker, M., Propper, R.E.,
Stickgold, R., 2011. Napping promotes inter-session habituation to emotional stimuli.
Neurobiol. Learn. Mem. 95, 24–36.
Petermann, U., & Petermann, F. 2011. WISC-IV: Wechsler Intelligence Scale for Children-:
Deutschsprachige Adaptation nach D. Wechsler: Pearson Assessment and
Information.
Pfurtscheller, G., Lopes da Silva, F.H., 1999. Event-related EEG/MEG synchronization and
desynchronization: basic principles. Clin. Neurophysiol. 110, 1842–1857.
Phelps, E.A., O'Connor, K.J., Gatenby, J.C., Gore, J.C., Grillon, C., Davis, M., 2001.
Activation of the left amygdala to a cognitive representation of fear. Nat. Neurosci. 4,
437–441.
Prehn-Kristensen, A., Goder, R., Chirobeja, S., Bressmann, I., Ferstl, R., Baving, L., 2009.
Sleep in children enhances preferentially emotional declarative but not procedural
memories. J. Exp. Child Psychol. 104, 132–139.
Prehn-Kristensen, A., Munz, M., Molzow, I., Wilhelm, I., Wiesner, C.D., Baving, L., 2013.
Sleep promotes consolidation of emotional memory in healthy children but not in
children with attention-deficit hyperactivity disorder. PLoS One 8, e65098.
Rechtschaffen, A., 1968. A manual of standardized terminology, techniques and scoring
system for sleep stages of human subjects. Public Health Serv.
Roach, G.D., Dawson, D., Lamond, N., 2006. Can a shorter psychomotor vigilance task be
used as a reasonable substitute for the ten‐minute psychomotor vigilance task?
Chronobiol. Int. 23, 1379–1387.
Schachter, S., Singer, J., 1962. Cognitive, social, and physiological determinants of
emotional state. Psychol. Rev. 69, 379.
Simonds, J.F., Parraga, H., 1982. Prevalence of sleep disorders and sleep behaviors in
children and adolescents. J. Am. Acad. Child Psychiatry 21, 383–388.
Takashima, A., Petersson, K.M., Rutters, F., Tendolkar, I., Jensen, O., Zwarts, M.,
McNaughton, B., Fernandez, G., 2006. Declarative memory consolidation in humans:
a prospective functional magnetic resonance imaging study. Proc. Natl. Acad. Sci.
USA 103, 756–761.
van der Helm, E., Yao, J., Dutt, S., Rao, V., Saletin, J.M., Walker, M.P., 2011. REM sleep
depotentiates amygdala activity to previous emotional experiences. Curr. Biol. 21,
2029–2032.
Wagner, U., Fischer, S., Born, J., 2002. Changes in emotional responses to aversive pic-
tures across periods rich in slow-wave sleep versus rapid eye movement sleep.
Psychosom. Med. 64, 627–634.
Wagner, U., Gais, S., Born, J., 2001. Emotional memory formation is enhanced across
E. Bolinger et al. Neuropsychologia 117 (2018) 84–91
90
sleep intervals with high amounts of rapid eye movement sleep. Learn. Mem. 8,
112–119.
Walker, M.P., 2009. The role of sleep in cognition and emotion. Ann. N. Y. Acad. Sci.
1156, 168–197.
Walker, M.P., 2010. Sleep, memory and emotion. Prog. Brain Res. 185, 49–68.
Watson, D., Clark, L.A., Tellegen, A., 1988. Development and validation of brief measures
of positive and negative affect: the PANAS scales. J. Pers. Soc. Psychol. 54,
1063–1070.
Werner, G.G., Schabus, M., Blechert, J., Kolodyazhniy, V., Wilhelm, F.H., 2015. Pre‐to
postsleep change in psychophysiological reactivity to emotional films: late‐night REM
sleep is associated with attenuated emotional processing. Psychophysiology 52,
813–825.
E. Bolinger et al. Neuropsychologia 117 (2018) 84–91
91
