Received: 30 March 2018 | Revised: 9 July 2018 | Accepted: 10 July 2018

DOI: 10.1111/jsr.12748

REGULAR RESEARCH PAPER

The impact of negative mood state on sleep‐related

attentional bias in insomnia

Senning Zheng1,2 | Juan Feng2 | Rongmao Lin2 | Youwei Yan2 | Renchuan Zhang3 |
Huiyuan Huang1 | Junjing Wang4 | Ruiwang Huang1

1
School of Psychology, Institute of Brain
Research and Rehabilitation (IBRR), Center Abstract
for the Study of Applied Psychology and Sleep‐related attentional bias is thought to play a role in the maintenance of insom-
MRI Center, Key Laboratory of Mental
Health and Cognitive Science of nia. However, this concept has been questioned by several studies that did not
Guangdong Province, South China Normal show the presence of sleep‐related attentional bias in clinical insomnia or poor
University, Guangzhou, China
2 sleepers. Our goal in the present study was to test whether the mood state of indi-
School of Psychology, Fujian Normal
University, Fuzhou, China viduals with insomnia affects the presence of sleep‐related attentional bias. To this
3
Fuzhou Neuro‐Psychiatric Hospital, end, 31 individuals with insomnia and 34 good sleepers were randomly assigned to
Fuzhou, China
4
a negative mood‐inducing condition or a control condition. They then completed a
Department of Applied Psychology,
Guangdong University of Foreign Studies, visual probe task with three types of pictorial stimuli (general threat, sleep‐related
Guangzhou, China
negative pictures and sleep‐related positive pictures). Vigilance, maintenance and
Correspondence the overall bias indexes were calculated based on the reaction time. We found indi-
Rongmao Lin, School of Psychology, Fujian
viduals with insomnia only showed a greater overall bias compared with good sleep-
Normal University, Fujian 350117, China.
Email: lrm990527@hotmail.com ers following a negative mood induction, regardless of the pictures presented. In
and
addition, we found that a negative mood state was significantly correlated with the
Ruiwang Huang, School of Psychology,
South China Normal University, Guangzhou overall attentional bias in good sleepers but not in individuals with insomnia. These
510631, China.
findings suggest that sleep‐related attentional bias in insomnia can be modulated by
Email: ruiwang.huang@gmail.com
mood state. This effect may reflect the dysregulation of top‐down attentional con-
Funding information
trol in individuals with insomnia.
The Humanities and Social Science
Foundation of Ministry of education, China,
Grant/Award Number: 13YJA190016; KEYWORDS
Ministry of Education of the People's autobiographical memory, insomnia, negative mood state, sleep‐related attentional bias
Republic of China

1 | INTRODUCTION
paradigm (Woods, Marchetti, Biello, & Espie, 2009; for review, see
Sleep‐related attentional bias (SAB), which refers to the tendency to Harris et al., 2015).
selectively attend to sleep‐related cues, is emphasized in several However, several studies have reported contradictory evidence
models as a core feature in the maintenance of insomnia (Espie, about the presence of SAB. Lundh, Froding, Gyllenhammar, Broman,
Broomfield, Macmahon, Macphee, & Taylor, 2006; Harvey, 2002). and Hetta (1997) used the EST to measure cognitive bias in patients
This concept has been supported by studies using the emotional with persistent insomnia, yet found an equivalent Stroop interfer-
Stroop task (EST) (Spiegelhalder, Espie, & Riemann, 2009), the flicker ence for sleep words between an insomnia group and a control
paradigm (Jones, Macphee, Broomfield, Jones, & Espie, 2005; March- group. In a study conducted by Spiegelhalder, Espie, Nissen, and Rie-
etti, Biello, Broomfield, Macmahon, & Espie, 2006), the visual probe mann (2008), a significant group difference with respect to SAB as
task (VPT) (Akram et al., 2018; Jansson‐Fröjmark, Bermås, & Kjellén, measured by the EST was found when comparing an insomnia group
2012; MacMahon, Broomfield, & Espie, 2006) and the Posner with an expert group (but not with a healthy control group).

J Sleep Res. 2018;e12748. wileyonlinelibrary.com/journal/jsr © 2018 European Sleep Research Society | 1 of 10

https://doi.org/10.1111/jsr.12748
2 of 10 | ZHENG ET AL.

Similarly, another study using the EST showed that poor sleepers 2013), three types of pictures (general threat, sleep‐related negative
had a longer response time to sleep‐related words than non‐specific and sleep‐related positive) were presented during the VPT.
threat words, but no significant group difference in reaction time The aim of the current study was to test whether individuals
was found (Barclay & Ellis, 2013). The absence of SAB in the above with insomnia, compared to good sleepers, displayed preferential
studies may not simply be attributable to the insensitivity of the attention toward sleep‐related pictures following a negative mood
EST, because several researchers also failed to replicate the finding induction. Based on the evidence that the emotional state can mod-
of SAB using the VPT (Clarke et al., 2016; Lancee et al., 2017; ulate attentional processing (Sagliano et al., 2018), we hypothesized
Spiegelhalder et al., 2010). Among the eye‐tracking studies, whereas that, relative to good sleepers, individuals with insomnia would show
some demonstrated that individuals with insomnia were more likely sleep‐related attentional bias in a negative mood condition but not
to watch sleep‐related cues (Akram, Ellis, Myachykov, & Barclay, in a control condition. In addition, because patients with insomnia
2017; Beattie, Bindemann, Kyle, & Biello, 2017), another showed a were reported to show greater amygdala reactivity to sleep‐related
pattern of general performance deficits in the insomnia populations negative pictures (but not to general negative pictures) than good
(Woods, Scheepers, Ross, Espie, & Biello, 2013). Furthermore, a sleepers (Baglioni et al., 2014), we also hypothesized that any bias
recent functional magnetic resonance imaging (fMRI) study ques- identified in the insomnia group would be specifically toward sleep‐
tioned the concept of SAB, because the study did not reveal any related negative pictures (and possibly toward sleep‐related positive
threat‐ or craving‐related brain reactivity to sleep‐related words in pictures, but not toward general threatening pictures).
patients with insomnia versus good sleepers (Spiegelhalder et al.,
2016). Together, those findings indicate the need to further investi-
2 | METHOD
gate SAB, especially for potential moderating effects.
Mood state is a relatively transient emotional experience that
2.1 | Participants
influences many domains of cognitive processes, such as attention
(Sagliano et al., 2018). The associative network theory (Bower, 1981) The participants were recruited from the undergraduate population at
suggests that emotional state can enhance the salience of mood‐ Fujian Normal University. They completed a screening questionnaire
congruent information, which in turn biases subsequent attention of the Pittsburgh Sleep Quality Index (PSQI) and a clinical interview
allocation. Therefore, as sleep‐related cues are emotionally salient (conducted by RZ, an experienced psychiatrist) to determine the eligi-
for individuals with insomnia (Baglioni et al., 2010, 2014), it is likely bility and group allocation. Individuals with insomnia had to score >7
that SAB will be affected by one's mood state. In particular, SAB on the PSQI and meet the DSM‐5 criteria for insomnia disorder. Their
may be robust when individuals with insomnia are in a negative difficulties in sleep initiation, sleep maintenance or early morning
mood, considering that the cognitive model of insomnia suggests awakening needed to be present for more than 3 nights per week, for
that emotional distress is a precipitating factor for SAB (Harvey, at least 1 month, and cause significant daytime impairment. Good
2002). To date, many studies have taken the symptoms of depres- sleepers were required to score <5 on the PSQI, and to report no cur-
sion/anxiety as control factors when accounting for SAB (e.g. Akram rent or historical sleep problems. According to the interview, we
et al., 2018). However, insomnia could persist even after the remis- excluded the individuals who reported symptoms of a sleep disorder
sion of mood disorders (Ohayon & Roth, 2003), and several studies other than insomnia, existence of medical/mental disorders that might
found that SAB is not significantly correlated with depression/anxiety contribute to insomnia, substance abuse, or vision problems. Addition-
(Akram et al., 2018; Jansson‐Fröjmark et al., 2012). This evidence ally, the data of individuals with an error rate >10% in the VPT were
seem to imply that the association between SAB and emotion is removed from the analyses.
state dependent rather than trait dependent. Therefore, there is a Figure 1 shows the recruitment details for the present study.
need to investigate the impact of mood state on SAB using a direct The final sample consisted of 31 individuals with insomnia (mean
experimental manipulation. Such a study is very important in probing age = 19.55 years, SD = 0.89 years; 23 female/8 male) and 34 age‐
the condition when individuals with insomnia display a dysfunctional and gender‐matched good sleepers (mean age = 19.50 years, SD =
attentional pattern and could provide useful guidelines for a treat- 0.86 years; 24 female/10 male). Written informed consent was pro-
ment strategy for insomnia. vided by all the participants, and the research protocol was approved
To that end, a widely used and validated procedure, autobio- by the Fujian Normal University Ethics Committee.
graphical recall, was employed for mood induction (Jallais & Gilet,
2010). In general, this technique requires subjects to recall and write
2.2 | Manipulation
down an emotional event that they have experienced in the past. In
the present study, individuals with insomnia and good sleepers were Fourteen individuals with insomnia and 19 good sleepers were
recruited and randomly assigned to an insomnia‐related recall condi- assigned to an insomnia‐related autobiographical recall (IAR) condi-
tion (negative mood condition) or a reading recall condition (control tion, and 17 individuals with insomnia and 15 good sleepers were
condition). Then they completed a VPT to measure the magnitude of assigned to a reading‐related autobiographical recall (RAR) condition.
SAB. To ensure that SAB was driven by a sleep property (rather than The main reasons for choosing IAR for the negative mood induc-
by a negative emotional valence) of the stimuli (Barclay & Ellis, tion were as follows. First, the experience of insomnia is a major
ZHENG ET AL. | 3 of 10

F I G U R E 1 Pipeline for subject recruitment and manipulation assignment. IAR, insomnia‐related autobiographical recall; RAR, reading‐related
autobiographical recall

source of emotional distress for people with insomnia. This experi- point Likert scale (0–4). The Chinese version of the STAI shows
ence can cause a complex of negative emotional feelings, including acceptable psychometric properties (Shek, 1988).
worry, anxiety, sadness and even the fear of death (Rezaie, Khazaie, Depression was assessed using the Beck Depression Inventory
& Yazdani, 2016; Yung, Chung, Ho, Yeung, & Ng, 2016). Thus, induc- (BDI) (Beck, Steer, & Carbin, 1988), which consists of 21 items (rang-
ing such a specific mood may be a preliminary step in understanding ing from 0 to 3). Higher total scores (0–63) indicate more severe
how individuals with insomnia naturally respond to the common depressive symptoms. Its Chinese version has been shown to be
emotional distress related to sleep disturbance. Second, other nega- valid and reliable (Shek, 1990).
tive emotional events (e.g. an experience of fear) seem to be salient
enough to elicit attentional bias in healthy participants (Sagliano
2.3.2 | Mood state measure
et al., 2018). Indeed, the findings of Richardson, Gradisar, and Pul-
ford (2015) indicated that good sleepers could develop temporary The immediate mood state following manipulation was assessed by a
SAB after watching a traumatic film. Hence, we avoided using salient visual analogue scale (VAS) (Kjeldsen, Klausen, & Rosenberg, 2015).
emotional events to minimize confounding effects (i.e. both individu- This simple tool requires participants to read a sentence and then to
als with insomnia and good sleepers develop SAB). rate their subjective feeling by marking a corresponding location on
a 100‐mm horizontal line. The left endpoint of the line (0 mm) indi-
cates “not at all”, and its right endpoint (100 mm) indicates “very
2.3 | Measures
much”. In this study, six dimensions were considered for the follow-
ing reasons. The negativity (“I feel distress”) and the positivity (“I feel
2.3.1 | Baseline measures
pleasant”) dimensions were included to measure emotional experi-
The PSQI, a psychometrically sound instrument for assessing sleep ence. Additionally, because an insomnia‐related autobiographical
quality (Buysse, Reynolds, Monk, Berman, & Kupfer, 1989), contains recall might induce simulated experiences of insomnia, we also
19 self‐rated items grouped into seven components: subjective sleep assessed fatigue (“I feel fatigue”), anxiety (“I feel anxious”) and cog-
quality, sleep latency, sleep duration, sleep disturbances, habitual nitive arousal (“I feel my mind is running”). Finally, the vividness
sleep efficiency, use of sleeping medication and daytime dysfunction. dimension (“I recall the memory vividly”) was accessed because neg-
The score for each component ranges from 0 to 3 and can be ative sleep memories may be more accessible to individuals with
summed to calculate a global score (ranging from 0 to 21). Higher insomnia. The main reasons for choosing VAS were to minimize the
global scores indicate greater sleep disturbance. In the Chinese ver- priming effect of a standard measure (Ellis, Gardani, & Hogh, 2010)
sion (Zheng, Man, Wang, & Jun, 2016), a global score above 7 indi- and to reduce the amount of time consumed.
cates poor sleep quality.
Anxiety was assessed using the State‐Trait Anxiety Inventory
2.4 | Materials
(STAI) (Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983), a well‐
constructed instrument consisting of a state subscale (STAI‐S) and a Four categories of stimuli set were used in the VPT. Each stimuli set
trait subscale (STAI‐T). Each subscale includes 20 items rated on a 5‐ contains 24 picture pairs. In each pair, a target picture, specifically a
4 of 10 | ZHENG ET AL.

FIGURE 2 Schematic representation of the visual probe task (VPT)

general threat picture (T), a sleep‐related negative picture (SN) or a or 14:00–19:00 hours). When the participants arrived, a written
sleep‐related positive picture (SP), was matched with a control pic- informed consent form was provided. Then a short interview based
ture (C). For the control pairs, two control pictures were matched on DSM‐5 was conducted to ensure the presence of insomnia disor-
with each other. Similar to a previous study (Jansson‐Fröjmark et al., der. After the interview, the participants completed the STAI and
2012), all the stimuli were selected from the Internet and were eval- BDI measures.
uated by an independent group. The procedure for building the pic- All the participants were then assigned randomly to two condi-
tures is provided in the supplementary materials (Figure S1). tions. In the IAR condition, the participants were asked to recall a
bad night's sleep, which should be their own experience, excluding
cases that were caused by over‐excitation/excitement driven by posi-
2.5 | Visual probe task
tive emotion. In the RAR condition, the participants were instructed
The VPT was programmed using E‐Prime 2.0 (Psychology Software to retrieve a memory of a typical reading experience. The partici-
Tools, Pittsburgh, PA, USA) and was presented on a 14‐inch white pants were encouraged to recall the memory as vividly as possible
background screen. The viewing distance was about 50 cm. and to write down the details but were discouraged from inventing
Figure 2 shows the experimental paradigm for the VPT. Each trial a story. Each participant was given 7 min to perform the recall and
began with a 500‐ms fixation cross, followed by 500 ms of picture pairs. was instructed to rate their subsequent feelings using the VAS.
When the pictures disappeared, a probe of two horizontal dots (..) or Finally, the VPT began. When the participants fully understood
two vertical dots (:) appeared in either the left or right position, lasting the instructions, they had 12 practice trials without feedback, fol-
2500 ms or until a response key was pressed. The participants were lowed by 96 experimental trials. We collected the key response and
instructed to indicate the spatial arrangement of the probe as quickly as the reaction time (RT) for each trial. To minimize expectation effects,
possible, using the “F” key for horizontal dots and the “J” key for verti- an experimenter who was blind to the group allocation administrated
cal dots. The trial interval was 1000 ms. If a target picture (T/SN/SP) the manipulation and the task sessions.
was followed by a probe with the same spatial position, it was deemed At the end of the experiment, all the participants were debriefed
to be a consistent trial. Otherwise, it was deemed to be an inconsistent and received a gift to thank them for their participation. The individ-
trial. A trial with two control pictures was defined as a neutral trial. uals with insomnia were also given advice for improving their sleep.
To control the order effects, we randomly arranged the presenta-
tion of the picture pairs, the location (left/right) of the target picture
2.7 | Analyses
and the probe and the spatial arrangement (horizontal/vertical) of the
probe in each trial. We conducted χ2 tests and two‐factor analyses of variance (ANO-
VAs) to detect any between‐group differences or within‐group dif-
ferences in the demographic and baseline measurements. A
2.6 | Procedure
multivariate ANOVA was used to check the effect of the manipula-
The experiment was conducted in a quiet room at the School of Psy- tion on each dimension of the VAS (vividness, positivity, negativity,
chology, Fujian Normal University, during the daytime (08:30–12:00 anxiety, fatigue and cognitive arousal).
ZHENG ET AL. | 5 of 10

In line with a previous study (Spiegelhalder et al., 2010), errors T A B L E 1 The demographical information and baseline measures
and outliers (mean RTs that exceeded the individual means by three (means ± SD)
standard deviations) in the VPT were removed from the analyses. Insomnia Good sleepers
Then, the three bias indexes (Jansson‐Fröjmark et al., 2012) were
RAR
calculated based on the picture‐probe combinations using the follow- IAR (n = 14) RAR (n = 17) IAR (n = 19) (n = 15)
ing equations: Sex (M/F) 4/10 4/13 4/15 6/9

Vigilance ¼ MeanRTðneutral trialÞ  MeanRTðconsistent trialÞ; Age 19.79 ± 0.80 19.35 ± 0.93 19.58 ± 0.84 19.40 ± 0.91
PSQI 10.21 ± 2.22 9.71 ± 1.57 3.42 ± 0.84 2.93 ± 0.96
Maintenance ¼ MeanRTðinconsistent trialÞ  MeanRTðneutral trialÞ; STAI‐S 43.50 ± 9.68 40.06 ± 11.44 31.84 ± 9.36 26.87 ± 5.57
STAI‐T 51.29 ± 6.90 46.24 ± 9.64 35.47 ± 8.44 33.60 ± 8.23
Overall bias ¼ MeanRTðinconsistent trialÞ  MeanRTðconsistent trialÞ: BDI 12.93 ± 6.90 14.00 ± 7.64 5.58 ± 5.10 5.53 ± 4.00

Note. IAR, insomnia‐related autobiographical recall; RAR, reading‐related

A higher vigilance value indicates that the individuals were more
autobiographical recall; PSQI, Pittsburgh Sleep Quality Index; STAI‐S,
vigilant about holding their attention on certain stimuli. A higher state subscale of State‐Trait Anxiety Inventory; STAI‐T, trait subscale of
maintenance value denotes that individuals had difficulty in disen- STAI; BDI, Beck Depression Inventory.
gaging from certain stimuli. A higher overall bias indicates a height-
ened vigilance or difficulty in shifting attention from certain threats
or both.
Three‐factor mixed‐measures ANOVAs were used to determine T A B L E 2 The effect of mood induction (visual analogue scale)
(means ± SD)
the impact of mood manipulation on attentional bias in individuals
with insomnia versus good sleepers. Group (insomnia versus good Insomnia Good sleepers
sleepers) and manipulation (IAR versus RAR) were selected as IAR RAR IAR RAR
between‐subject variables, picture type (T versus SN versus SP) as a (n = 14) (n = 17) (n = 19) (n = 15)
within‐subject variable, and bias indexes (vigilance, maintenance or Vividness 6.51 ± 1.64 6.24 ± 2.55 6.02 ± 2.21 6.75 ± 2.39
overall bias) as dependent variables. We also computed correlations Positivity 2.11 ± 2.28 5.73 ± 2.89 1.78 ± 2.50 6.37 ± 2.66
between the baseline measurements and each bias index. If any of Negativity 5.20 ± 2.75 2.25 ± 2.68 2.33 ± 2.96 0.88 ± 2.29
the correlations were significant, those baseline measures were Anxiety 2.52 ± 1.99 2.31 ± 2.69 1.44 ± 1.53 2.91 ± 2.34
taken as covariates in an ANCOVA. Fatigue 5.71 ± 1.85 4.37 ± 2.26 3.45 ± 2.87 2.68 ± 2.31
Further, the Pearson correlations between each bias index and
Cognitive 5.29 ± 2.09 4.14 ± 3.57 2.92 ± 2.40 3.07 ± 3.43
the dimension on the VAS were calculated separately for the insom- arousal
nia and the good sleeper groups. This was to check whether the
Note. IAR, insomnia‐related autobiographical recall; RAR, reading‐related
alteration in the attentional bias could be attributed to the emotional autobiographical recall.
state or to another state. Significance thresholds were set at
recall (Table 2). We also detected several main effects of group,
p < 0.05 (two‐tailed) for all analyses.
indicating that individuals with insomnia generally reported stron-
ger negative feelings (F1,61 = 9.89, p < 0.01), tiredness
3 | RESULTS (F1,61 = 10.93, p < 0.01) and mind running (F1,61 = 5.47, p < 0.05)
than good sleepers. No other main effect or interaction was
3.1 | Participants detected.

Table 1 lists the demographic data and the baseline measurements

for the participants. For sex distribution and age, no significant group 3.3 | Attentional bias
or condition difference was found. For all the baseline measure-
ments, the main effects of group were significant (Ps < 0.001). That Table 3 lists the basic task performance and bias indexes. As for task

is, individuals with insomnia subjectively rated themselves as worse performance, no significant effect was found for either the mean RT

in sleep quality, more anxious in state and in trait and more or error rate (two‐factor ANOVA).

depressed.

3.3.1 | Vigilance index

3.2 | Manipulation check
The mixed‐measures ANOVA revealed no significant main effect or
We found significant main effects of manipulation for both positiv- interaction with respect to the vigilance index. The score for the
ity and negativity (F1,61 = 39.86, p < 0.001; F1,61 = 10.65, p < 0.01, STAI‐S was then added to ANCOVAs as a covariate because it was
respectively), showing that insomnia‐related recall stimulated a less significantly correlated with vigilance toward general threats
positive and a more negative experience than did reading‐related (r = 0.254, p = 0.041). However, all the results remained the same.
6 of 10 | ZHENG ET AL.

T A B L E 3 The scores for task performance and three types of attentional bias in the visual probe task (means ± SD)
Insomnia Good sleepers

IAR (n = 14) RAR (n = 17) IAR (n = 19) RAR (n = 15)

Task performance
RT (ms) 619.49 ± 101.08 625.28 ± 79.40 615.30 ± 87.77 577.37 ± 56.84
ER (%) 4.21 ± 2.83 3.71 ± 2.76 2.84 ± 2.06 2.93 ± 2.63
Vigilance
V‐T 6.49 ± 37.33 −3.24 ± 52.51 11.76 ± 36.46 −12.07 ± 31.28
V‐SN 16.47 ± 40.65 12.79 ± 68.35 0.58 ± 46.82 21.30 ± 35.33
V‐SP 15.40 ± 40.49 13.32 ± 39.82 4.51 ± 54.20 6.95 ± 38.19
Maintenance
M‐T 7.49 ± 62.83 −15.83 ± 57.46 −19.36 ± 45.47 6.25 ± 44.30
M‐SN −0.63 ± 44.35 −8.97 ± 65.98 −14.26 ± 39.96 −0.94 ± 39.80
M‐SP 9.79 ± 36.09 −7.22 ± 55.01 −13.36 ± 33.90 1.23 ± 23.63
Overall bias
OB‐T 13.98 ± 70.96 −19.08 ± 55.83 −7.60 ± 57.64 −5.82 ± 38.48
OB‐SN 15.84 ± 69.05 3.82 ± 46.67 −13.68 ± 40.22 20.37 ± 48.91
OB‐SP 25.18 ± 30.25 6.10 ± 43.96 −8.85 ± 71.92 8.18 ± 38.44

Note. RT, reaction time; ER, error rate; V, vigilance index; M, maintenance index; OB, overall bias index; T, general threat; SN, sleep‐related negative;
SP, sleep‐related positive; IAR, insomnia‐related autobiographical recall; RAR, reading‐related autobiographical recall.

overall bias toward sleep‐related negative stimuli, those measures

3.3.2 | Maintenance index
were added as covariates in ANCOVAs. However, the ANCOVAs did
The largest effect showed by the mixed‐measures ANOVA was the not show any distinct results that differed from the primary
group × manipulation interaction (F1,61 = 3.66, p = 0.060, η2p = analyses.
0.057). Simple effect analyses showed a tendency (F1,31 = 3.95,
p = 0.056, η2p = 0.113) for individuals with insomnia to display stron-
3.4 | Correlational analyses
ger maintenance than good sleepers in the IAR condition, regardless
of stimuli type. The simple effect analyses also revealed that, among Among the individuals with insomnia, surprisingly, we did not find
the good sleepers, those who were in the IAR condition were more any significant correlation between emotional state (positivity or
likely to disengage from the target stimuli than those who were in negativity) and each bias index. However, the negativity in the good
the RAR condition, regardless of picture type (F1,32 = 4.01, sleepers was significantly negatively correlated with vigilance toward
p = 0.054, η2p = 0.111). Because no baseline measure was signifi- sleep‐related positive stimuli, overall bias toward sleep‐related posi-
cantly correlated with any of the maintenance indexes, we did not tive stimuli, mean vigilance and mean overall bias (rs <−0.350, Ps <
conduct an ANCOVA. 0.05; see Matrix B in Figure S2).
Another distinct correlational pattern was found for the individu-
als with insomnia versus the good sleepers (Matrix A in Figure S2).
3.3.3 | Overall bias index
Specifically, the cognitive arousal level in the individuals with insom-
The mixed‐measures ANOVA for overall bias only revealed a signifi- nia was positively associated with vigilance toward general threats,
cant group × manipulation interaction (F1,61 = 6.26, p = 0.015, sleep‐related negative stimuli and mean vigilance (rs > 0.309, Ps <
η2p = 0.093; see Figure 3). Simple effect analyses showed patterns 0.05). This correlational pattern was not found for the good sleep-
that were similar to those of the maintenance index; that is, individ- ers.
uals with insomnia displayed a stronger attentional bias than good
sleepers under the IAR condition (F1,31 = 5.85, p = 0.022,
4 | DISCUSSION
η2p = 0.159). Moreover, relative to the good sleepers in the RAR con-
dition, the good sleepers in the IAR condition displayed a tendency The current study examined whether individuals with insomnia, com-
toward attentional disengagement (F1,32 = 3.59, p = 0.067, pared with good sleepers, would display sleep‐related attentional
η2p = 0.101). Given that the score of the STAI‐S was significantly cor- bias following a negative mood manipulation. We showed that, only
related with the overall bias toward general threats (r = 0.282, under the IAR condition, the individuals with insomnia displayed a
p = 0.023), and that the scores of the STAI‐T (r = 0.244, p = 0.050) stronger overall bias relative to good sleepers. However, because no
and BDI (r = 0.303, p = 0.014) were positively associated with the significant three‐way interaction was found in all the primary
ZHENG ET AL. | 7 of 10

F I G U R E 3 Interactions of group × manipulation on three bias indexes (means ± SE). As no significant three‐way interaction was found,
each bias index presented was the mean value of the corresponding index of three stimuli types. IAR, insomnia‐related autobiographical recall;
RAR, reading‐related autobiographical recall; †p < 0.10; *p < 0.05

analyses, the hypothesis that attentional bias in insomnia is specifi- to those who experienced the control manipulation. Correlational
cally toward sleep‐related stimuli was not supported. analyses also suggested that good sleepers were more likely to dis-
engage from sleep‐related and negative information under emotional
distress. Thus, it seems that good sleepers apply a distraction strat-
4.1 | Attentional bias, emotion and insomnia
egy to cope with sudden stressful events. Appropriate attentional
The cognitive model of insomnia (Harvey, 2002) provided a system- distraction is deemed to be an effective regulator of emotion that
atic perspective for understanding the processes contributing to the can help individuals reduce emotional distress and other stress
maintenance of the disorder. This model suggested that excessive responses following a threatening exposure (Wolgast & Lundh,
negative cognitive activity leads to the arousal and distress, a “sleep‐ 2017). Unfortunately, in this study, such an adaptive strategy was
specific anxiety state”, which further precipitates the preferential seemingly lacking in the individuals with insomnia. As suggested by
attentional allocation to sleep‐related cues. In the present study, sig- Hur et al. (2015), this is probably because the presence of negative
nificant group differences with respect to attentional bias only emotional arousal suppresses their top‐down regulation. As a conse-
appeared following a specific negative mood manipulation (Figure 3). quence, individuals with insomnia find it difficult to inhibit sleep‐re-
Thus, this study provides the first empirical support for the sequence lated information under emotional distress, which manifests as a
between emotional distress and sleep‐related attentional bias pro- form of sleep‐related attentional bias relative to good sleepers. This
posed by the model. Furthermore, these findings also suggest that explanation was further supported by the evidence that individuals
the attentional bias in insomnia is an unstable cognitive feature that with insomnia are characterized by poor attentional control (Mitchell,
can be modulated by the mood state, which could partially explain Mogg, & Bradley, 2012) and emotional dysregulation (Palagini, Mor-
the mixed evidence on the presence of sleep‐related attentional bias. etto, Dell'osso, & Carney, 2017). If so, it would benefit psychological
There are two possible explanations for the effect of mood state. treatments of insomnia (e.g. cognitive behavioural therapy) to include
First, according to Bower (1981), it is because a negative mood state strategies targeting regulation of emotions. The effectiveness of such
enhances the salience of sleep‐related stimuli for individuals with a treatment in improving attention regulation has been demonstrated
insomnia. However, this bottom‐up sensory amplification explanation for generalized anxiety disorder (Renna et al., 2017).
is inconsistent with some of our findings. Specifically, we did not
find significant condition differences on the bias indexes among the
4.2 | Specific bias or general cognitive impairment
individuals with insomnia, and their bias indexes were not signifi-
cantly correlated with subjective emotional experience. In short, In contrast with a previous study (Baglioni et al., 2014), the hypothe-
mood state seems to be irrelevant to the magnitude of attentional sis that attentional bias in individuals with insomnia would specifi-
bias in individuals with insomnia. cally be toward sleep‐related cues was not supported in this study.
A second alternative explanation is related to the potential for One might doubt that the attentional bias that the insomnia group
top‐down dysregulation in people with insomnia. Our findings displayed was due to a general cognitive impairment caused by a
showed that, among the good sleepers, those who experienced the negative mood. However, we did not find significant group differ-
negative mood induction exhibited an overall avoidance bias relative ences in performance in the insomnia‐related recall condition. Based
8 of 10 | ZHENG ET AL.

on the positive relationship between cognitive arousal and vigilance CONFLICT OF INTEREST
toward negative information (Matrix A in Figure S2), a specific bias
No conflicts of interest declared.
in insomnia is possible. On the other hand, perhaps, the absence of
specific bias in the individuals with insomnia may be attributed to
the fact that insomnia is highly linked with other emotional disorders AUTHOR CONTRIBUTION
(Ohayon, Caulet, & Lemoine, 1998), which makes it possible for indi-
SZ, JF, RL and YY designed the study, SZ, JF and RZ performed the
viduals with insomnia to display biases toward both sleep‐related
experiments, SZ analysed the data and drafted the first version of
cues and negative emotional information.
the manuscript, and RL, HH, JW and RH edited the manuscript. All
authors discussed the results and the interpretations.
4.3 | Limitations
Several limitations of the present study need to be acknowledged. ORCID
First, the results may not be generalizable to other populations,
Senning Zheng http://orcid.org/0000-0003-4835-4417
especially clinical patients, because the recruitment was limited to a
non‐clinical sample of college students. Second, although the vari-
ous categories of pictures were significantly different from each
REFERENCES
other and were exhibited in well‐matched pairs, the unexpectedly
high positive valence of the control pictures may have confounded Akram, U., Beattie, L., Ypsilanti, A., Reidy, J., Robson, A., Chapman, A.
the results in this study. In addition, recent studies (Akram et al., J., & Barclay, N. L. (2018). Sleep‐related attentional bias for tired
faces in insomnia: Evidence from a dot‐probe paradigm. Behavior
2018; Akram, Ellis, Myachykov, & Barclay, 2016) suggested that
Research and Therapy, 103, 18–23. https://doi.org/10.1016/j.brat.
sleep‐specific faces may be more proximal stimuli than words or 2018.01.007
images in investigating cognitive bias in insomnia. Accordingly, Akram, U., Ellis, J. G., Myachykov, A., & Barclay, N. L. (2016). Mispercep-
future studies should replicate our findings by using stimuli with a tion of tiredness in young adults with insomnia. Journal of Sleep
Research, 25, 466–474. https://doi.org/10.1111/jsr.12395
less distal nature. Third, in light of the fact that the function of
Akram, U., Ellis, J. G., Myachykov, A., & Barclay, N. L. (2017). Preferential
mood repair of positive autobiographical memory is well established attention towards the eye‐region amongst individuals with insomnia.
(Kovacs et al., 2015), the moderately positive emotional experience Journal of Sleep Research, 26, 84–91. https://doi.org/10.1111/jsr.
induced by the control condition may have blunted the measure- 12456
Baglioni, C., Lombardo, C., Bux, E., Hansen, S., Salveta, C., Biello, S., …
ment of attentional bias in this study. Finally, our experiment was
Espie, C. A. (2010). Psychophysiological reactivity to sleep‐related
carried out during two periods of time, morning and afternoon. emotional stimuli in primary insomnia. Behavior Research and Therapy,
Although the test time variable did not affect any primary results, it 48, 467–475. https://doi.org/10.1016/j.brat.2010.01.008
unexpectedly had main effects on reaction time and overall bias Baglioni, C., Spiegelhalder, K., Regen, W., Feige, B., Nissen, C., Lombardo,
C., … Riemann, D. (2014). Insomnia disorder is associated with
(see supplementary materials for details). As such, this variable
increased amygdala reactivity to insomnia‐related stimuli. Sleep, 37,
needs to be controlled by using a stricter design or relevant mea- 1907–1917. https://doi.org/10.5665/sleep.4240
sures in future research. Barclay, N. L., & Ellis, J. G. (2013). Sleep‐related attentional bias in poor
versus good sleepers is independent of affective valence. Journal of
Sleep Research, 22, 414–421. https://doi.org/10.1111/jsr.12035
5 | CONCLUSION Beattie, L., Bindemann, M., Kyle, S. D., & Biello, S. M. (2017). Attention
to beds in natural scenes by observers with insomnia symptoms.
The present study showed that when compared with good sleepers, Behavior Research and Therapy, 92, 51–56. https://doi.org/10.1016/
individuals with insomnia displayed attentional bias toward both j.brat.2017.02.001
sleep‐related and negative stimuli following negative mood manipula- Beck, A. T., Steer, R. A., & Carbin, M. G. (1988). Psychometric properties
of the Beck Depression Inventory: Twenty‐five years of evaluation.
tion. This may reflect a top‐down dysregulation in individuals with
Clinical Psychology Review, 8, 77–100. https://doi.org/10.1016/0272-
insomnia, which causes them difficulty in adopting an adaptive 7358(88)90050-5
attentional strategy. Our findings suggest that negative emotion is Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129–
an important factor impacting cognitive function in individuals with 148. https://doi.org/10.1037/0003-066X.36.2.129
Buysse, D. J., Reynolds, C. F. I., Monk, T. H., Berman, S. R., & Kupfer, D.
insomnia and extend our knowledge about the cognitive model of
J. (1989). The Pittsburgh sleep quality index: A new instrument for
insomnia. psychiatric practice and research. Psychiatry Research, 28, 193–213.
https://doi.org/10.1016/0165-1781(89)90047-4
Clarke, P. J., Bedford, K., Notebaert, L., Bucks, R. S., Rudaizky, D., Milkins,
ACKNOWLEDGEMENTS B. C., & MacLeod, C. (2016). Assessing the therapeutic potential of
targeted attentional bias modification for insomnia using smartphone
This study was supported by a grant from the Humanities and Social delivery. Psychotherapy and Psychosomatics, 85, 187–189. https://doi.
Science Foundation of the Ministry of Education, China (No. org/10.1159/000442025
13YJA190016). The authors appreciate Rhoda E. and Edmund F. Ellis, J., Gardani, M., & Hogh, H. (2010). Priming affects poor sleepers but
not normal sleepers on an insomnia ambiguity task. Journal of Sleep
Perozzi for their editing assistance.
ZHENG ET AL. | 9 of 10

Research, 19, 27–30. https://doi.org/10.1111/j.1365-2869.2009. Psychiatry, 39, 185–197. https://doi.org/10.1016/S0010-440X(98)

00792.x 90059-1
Espie, C. A., Broomfield, N. M., Macmahon, K. M., Macphee, L. M., & Ohayon, M. M., & Roth, T. (2003). Place of chronic insomnia in the
Taylor, L. M. (2006). The attention‐intention‐effort pathway in the course of depressive and anxiety disorders. Journal of Psychiatric
development of psychophysiologic insomnia: A theoretical review. Research, 37, 9–15. https://doi.org/10.1016/S0022-3956(02)00052-
Sleep Medicine Reviews, 10, 215–245. https://doi.org/10.1016/j.smrv. 3
2006.03.002 Palagini, L., Moretto, U., Dell'osso, L., & Carney, C. (2017). Sleep‐
Harris, K., Spiegelhalder, K., Espie, C. A., Macmahon, K. M., Woods, H. C., related cognitive processes, arousal, and emotion dysregulation
& Kyle, S. D. (2015). Sleep‐related attentional bias in insomnia: A in insomnia disorder: The role of insomnia‐specific rumination.
state‐of‐the‐science review. Clinical Psychology Review, 42, 16–27. Sleep Medicine, 30, 97–104. https://doi.org/10.1016/j.sleep.2016.11.
https://doi.org/10.1016/j.cpr.2015.08.001 004
Harvey, A. G. (2002). A cognitive model of insomnia. Behavior Research Renna, M. E., Seeley, S. H., Heimberg, R. G., Etkin, A., Fresco, D. M., &
and Therapy, 40, 869–893. https://doi.org/10.1016/S0005-7967(01) Mennin, D. S. (2017). Increased attention regulation from emotion
00061-4 regulation therapy for generalized anxiety disorder. Cognitive Therapy
Hur, J., Miller, G. A., McDavitt, J. R., Spielberg, J. M., Crocker, L. D., and Research, 42, 121–134. https://doi.org/10.1007/s10608-017-
Infantolino, Z. P., … Heller, W. (2015). Interactive effects of trait and 9872-7
state affect on top‐down control of attention. Social Cognitive and Rezaie, L., Khazaie, H., & Yazdani, F. (2016). Exploration of the experi-
Affective Neuroscience, 10, 1128–1136. https://doi.org/10.1093/sca ence of living with chronic insomnia: A qualitative study. Sleep
n/nsu163 Science, 9, 179–185. https://doi.org/10.1016/j.slsci.2016.07.001
Jallais, C., & Gilet, A. L. (2010). Inducing changes in arousal and Richardson, C., Gradisar, M., & Pulford, A. (2015). The development of
valence: Comparison of two mood induction procedures. Behavior insomnia or the plasticity of good sleep? A preliminary study of acute
Research Methods, 42, 318–325. https://doi.org/10.3758/BRM.42.1. changes in sleep and insomnia resulting from an analogue trauma.
318 Behavioral Sleep Medicine, 13, 19–35. https://doi.org/10.1080/
Jansson‐Fröjmark, M., Bermås, M., & Kjellén, A. (2012). Attentional bias 15402002.2013.829065
in insomnia: The dot‐probe task with pictorial stimuli depicting day- Sagliano, L., Trojano, L., Di Mauro, V., Carnevale, P., Di Domenico, M.,
time fatigue/malaise. Cognitive Therapy and Research, 37, 534–546. Cozzolino, C., & D'Olimpio, F. (2018). Attentional biases for threat
https://doi.org/10.1007/s10608-012-9486-z after fear‐related autobiographical recall. Anxiety Stress and Coping,
Jones, B. T., Macphee, L. M., Broomfield, N. M., Jones, B. C., & Espie, C. 31, 69–78. https://doi.org/10.1080/10615806.2017.1362297
A. (2005). Sleep‐related attentional bias in good, moderate, and poor Shek, D. T. (1988). Reliability and factorial structure of the Chinese ver-
(primary insomnia) sleepers. Journal of Abnormal Psychology, 114, sion of the State‐Trait Anxiety Inventory. Journal of Psychopathology
249–258. https://doi.org/10.1037/0021-843X.114.2.249 and Behavioral Assessment, 10, 303–317. https://doi.org/10.1007/
Kjeldsen, H. B., Klausen, T. W., & Rosenberg, J. (2015). Preferred pre- BF00960624
sentation of the visual analog scale for measurement of postopera- Shek, D. T. (1990). Reliability and factorial structure of the Chinese ver-
tive pain. Pain Practice, 980–984. https://doi.org/10.1111/papr. sion of the Beck Depression Inventory. Journal of Clinical Psychology,
12344 46, 35–43. https://doi.org/10.1002/1097-4679(199001)46:1aabbb
Kovacs, M., Yaroslavsky, I., Rottenberg, J., George, C. J., Baji, I., Benák, I., 35:AID-JCLP2270460106aaabb3.0.CO;2-W
… Kapornai, K. (2015). Mood repair via attention refocusing or recall Spiegelhalder, K., Baglioni, C., Regen, W., Kyle, S. D., Nissen, C., Hennig,
of positive autobiographical memories by adolescents with pediatric‐ J., … Riemann, D. B. (2016). Reactivity and selective attention to
onset major depression. Journal of Child Psychology and Psychiatry, 56, sleep‐related words in patients with chronic insomnia. Behavioral
1108–1117. https://doi.org/10.1111/jcpp.12376 Sleep Medicine, 2016, 1–15. https://doi.org/10.1080/15402002.2016.
Lancee, J., Yasiney, S. L., Brendel, R. S., Boffo, M., Clarke, P. J. F., & Sale- 1253014
mink, E. (2017). Attentional bias modification training for insomnia: A Spiegelhalder, K., Espie, C., Nissen, C., & Riemann, D. (2008). Sleep‐re-
double‐blind placebo controlled randomized trial. PLoS One, 12, lated attentional bias in patients with primary insomnia compared
e0174531. https://doi.org/10.1371/journal.pone.0174531 with sleep experts and healthy controls. Journal of Sleep Research, 17,
Lundh, L.‐G., Froding, A., Gyllenhammar, L., Broman, J.‐E., & Hetta, J. 191–196. https://doi.org/10.1111/j.1365-2869.2008.00641.x
(1997). Cognitive bias and memory performance in patients with per- Spiegelhalder, K., Espie, C., & Riemann, D. (2009). Is sleep-related atten-
sistent insomnia. Scandinavian Journal of Behaviour Therapy, 26, 27– tional bias due to sleepiness or sleeplessness? Cognition & Emotion,
35. https://doi.org/10.1080/16506079708412033 23, 541–550. https://doi.ory/10.1080/02699930802022129
Macmahon, K. M., Broomfield, N. M., & Espie, C. (2006). Attention bias Spiegelhalder, K., Kyle, S. D., Feige, B., Prem, M., Nissen, C., Espie, C. A.,
for sleep related stimuli in primary insomnia and delayed sleep phase & Riemann, D. (2010). The impact of sleep‐related attentional bias on
syndrome using the Dot‐Probe Task. Sleep, 29, 1420–1427. https://d polysomnographically measured sleep in primary insomnia. Sleep, 33,
oi.org/10.1093/sleep/29.11.1420 107–112. https://doi.org/10.1093/sleep/33.1.107
Marchetti, L. M., Biello, S. M., Broomfield, N. M., Macmahon, K. M., & Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G.
Espie, C. A. (2006). Who is pre‐occupied with sleep? A comparison of A.. (1983). Manual for the State‐Trait Anxiety Inventory (STAI). Palo
attention bias in people with psychophysiological insomnia, delayed Alto, CA: Consulting Psychologists Press.
sleep phase syndrome and good sleepers using the induced change Wolgast, M., & Lundh, L. G. (2017). Is distraction an adaptive or maladap-
blindness paradigm. Journal of Sleep Research, 15, 212–221. https://d tive strategy for emotion regulation? A person‐oriented approach.
oi.org/10.1111/j.1365-2869.2006.00510.x Journal of Psychopathological and Behavioral Assessment, 39, 117–127.
Mitchell, L., Mogg, K., & Bradley, B. P. (2012). Relationships between https://doi.org/10.1007/s10862-016-9570-x
insomnia, negative emotionality and attention control. Sleep and Bio- Woods, H., Marchetti, L. M., Biello, S. M., & Espie, C. A. (2009). The clock
logical Rhythms, 10, 237–243. https://doi.org/10.1111/j.1479-8425. as a focus of selective attention in those with primary insomnia: An
2012.00567.x experimental study using a modified Posner paradigm. Behavior
Ohayon, M. M., Caulet, M., & Lemoine, P. (1998). Comorbidity of mental Research and Therapy, 47, 231–236. https://doi.org/10.1016/j.brat.
and insomnia disorders in the general population. Comprehensive 2008.12.009
10 of 10 | ZHENG ET AL.

Woods, H. C., Scheepers, C., Ross, K. A., Espie, C. A., & Biello, S. M. SUPPORTING INFORMATION
(2013). What are you looking at? Moving toward an attentional time-
line in insomnia: A novel semantic eye tracking study. Sleep, 36, Additional supporting information may be found online in the
1491–1499. https://doi.org/10.5665/sleep.3042 Supporting Information section at the end of the article.
Yung, K. P., Chung, K. F., Ho, F. Y., Yeung, W. F., & Ng, T. H. (2016).
The experience of chronic insomnia in chinese adults: a study
using focus groups and insomnia experience diaries. Behavioral
Sleep Medicine, 14, 406–428. https://doi.org/10.1080/15402002. How to cite this article: Zheng S, Feng J, Lin R, et al. The
2015.1017097 impact of negative mood state on sleep‐related attentional
Zheng, B., Man, L. I., Wang, K. L., & Jun, L. V. (2016). Analysis of the reli-
bias in insomnia. J Sleep Res. 2018;e12748. https://doi.org/
ability and validity of the Chinese version of Pittsburgh sleep quality
index among medical college students. Journal of Peking University, 10.1111/jsr.12748
48, 424–428.