Chronotype and time of day effects on verbal and facial emotional Stroop task performance in adolescents

ABSTRACT

This study investigated chronotype and time of day effects on lexical and facial emotion processing tasks and explored relationships with sleep quality and mental health outcomes. Participants were 351 Taiwanese adolescents (204 males) aged 13 to 16 years of age who completed both a Facial-Emotional Stroop and a Lexical-Emotional Stroop task at 08:00–10:00 am or at 14:00–16:00 pm. Chronotype was measured using the Morningness-Eveningness Questionnaire (MEQ) and grouped into Morning (n = 46), Intermediate (n = 248) or Evening types (n = 57). Participants completed validated measures of sleep quality (PSQI) and mental health difficulties (DASS-21). The study observed independent effects of time of day and chronotype on the Facial-Emotional Stroop task. Independent of chronotype group, delayed reaction times to anger stimuli were observed in adolescents tested in the morning. Younger adolescents with an Evening chronotype showed delayed responses to anger faces independent of the time of testing. Facial-Emotional Stroop task performance correlated with reported sleep quality and mental health only in Evening chronotypes, with attenuated responses to anger versus neutral stimuli associated with poorer sleep quality and worse mental health outcomes. An exploratory simple mediation analysis indicated that the relationship between attenuated responses to anger and greater mental health difficulties was fully mediated by poorer sleep quality. This study provides evidence of diurnal and chronotype-related variation in facial threat-related processing in typical adolescent development. It also indicates how social emotional processing is vulnerable to sleep disruptions and is linked to the greater mental health problems observed in adolescents with an eveningness profile.

KEYWORDS:

• Adolescents
• chronotype
• morningness-eveningness
• emotion processing
• mental health
• sleep

Introduction

Adolescence is a period associated with significant changes in sleep patterns due to biological processes related to pubertal development. The circadian rhythm undergoes change during the adolescent period whereby there is a tendency for older adolescents to have later bedtimes and timing of sleep onset compared to that of younger adolescents (Carskadon et al. 1998). This behavioural pattern is attributed to both biological changes in the circadian system known as ‘phase delay’ toward an eveningness chronotype, and environmental and societal pressures that include early school start times, an increase in academic workload and extracurricular activities (Crowley et al. 2007). There is no change in the amount of sleep need, and therefore the amount of sleep adolescents get is significantly reduced compared to younger children (Carskadon et al. 1980), and as Huang et al. (2010) reported, insufficient sleep is one of the most commonly reported problems amongst Taiwanese adolescents. A delayed sleep onset has been linked to increased reported daytime sleepiness and greater mood problems in Taiwanese adolescents (Gau and Soong 1995), that is consistent with observations internationally (Short et al. 2020). It is now well recognized that adolescence is a period associated with insufficient sleep, and that individuals with a later evening chronotype are the most affected in terms of a shorter duration and poorer quality of sleep (Vollmer et al. 2017).

Sleep factors and chronotype both influence the development of emotion processing in adolescence. Research has shown that sleep restriction and disruptions can have a negative impact on the development of abilities to process facial emotional expressions (Soffer-Dudek et al. 2011), lead to altered experience of affect, difficulties in emotion regulation, and mood problems (Baum et al. 2014; Reddy et al. 2017). Independent of sleep factors, adolescents with a greater evening chronotype are also observed to have altered affective processing (Dagys et al. 2012), that may partly explain the poorer mental health outcomes observed in adolescents with a later evening preference (Randler 2011).

There are fewer studies that have investigated an interaction between chronotype and time of day, termed ‘synchrony effects,’ on emotional states and affective processing in adolescent samples. Randler and Weber (2015) found an overall increase in reported positive affect from morning to afternoon, lower positive affect in the evening-orientated compared to morning-orientated students, but no synchrony effects. Results also remained unchanged after controlling for sleep length the previous night. Díaz-Morales et al. (2015) reported the same pattern; mood (sense of pleasantness) improved from morning and afternoon in all chronotypes, and evening types remain lower than morning types throughout the day. Bettencourt et al. (2020) similarly reported the increase from morning to afternoon in positive emotional states (reports of subjective mood and wellbeing) and positive affect, again also finding evening types were lower than morning types on both measures, with no interaction observed. The chronotype group differences in positive affect also remained after covarying out sleep quality and sleep duration. The studies therefore consistently show an improvement in reported emotional experience (positive affect, wellbeing, mood) from morning to afternoon for all chronotypes, with evening types reporting a poorer emotional experience across the day.

Adult research has established synchrony effects on different aspects of cognition including attention, memory and executive functions (Schmidt et al. 2007). Prior studies with adolescents that have assessed cognitive processes also report synchrony effects, including on tests of fluid intelligence (Goldstein et al. 2007) and executive function (Hahn et al. 2012). It remains unclear if the lack of observed synchrony effects on emotional content are due to the use of self-report measures, or social cognitive and affective processes are consistent with the changes in reported mood and affect outlined above. Chronotype and time of day effects are potentially more likely to be observable in emotion processing tasks that tap attention and higher order cognitive skills. There is extensive past research addressing the development of emotional processing in adolescents, the influence of sleep factors and the implications for mental health. This study will therefore investigate if there is diurnal and chronotype-related variation in emotion processing in adolescents. It will further explore if performance is affected by sleep quality and relationships with reported mental health problems in Taiwanese adolescents.

Method

Participants

Participants were 352 Taiwanese adolescent students attending a Taipei urban high school. The sample was comprised of 148 (42%) girls and 204 boys (58%). The age range was 13 to 16 years old (M = 14.4, SD = 0.88). Ten classes were randomly selected from three grades. Two classes from Grade 7 (n = 93), two classes from Grade 8 (n = 156) and two classes from Grade 9 (n = 103). All participants were typically developing adolescents, and none reported having a developmental disorder or colour blindness.

Ethics

This study was approved University Ethics Committee. Parents provided consent and students provided assent prior to participation in the study.

Materials

Emotion processing tasks

Participants completed a Facial and Verbal version of the Emotional Stroop Task to assess cognitive control in response to emotional stimuli in the perceptual and verbal domains. The tasks have been previously described by Isaac et al. (2012) and were translated into Chinese.

Facial emotional stroop task

The Facial Emotional task stimuli consisted of a total of 160 emotional faces which comprised 10 different identities (5 males and 5 females) x 4 emotions (happy, neutral, anger, sad) x 4 colours (blue, green, red and yellow). Emotional faces were selected from Taiwan Corpora of Chinese emotions (Chen et al. 2013) in order to reduce cultural differences for Taiwanese participants. All facial stimuli were cropped, free from hair or other external accessories that could prevent any distractions during the task. Before presenting the stimulus, a fixation point was presented for 200 msecs followed by the stimulus, which was presented for 2000 ms to ensure participants had sufficient time to react. After participants press the key, feedback showed whether the response was correct, which lasted 500 msecs. An example experimental trial is displayed in Figure 1.

Figure 1. The diagram of Facial-Emotional Stroop Task. In this example, the stimulus is a male’s face of Blue Happy.

Lexical emotional Stroop task

The experiment stimuli consisted of emotion words categorized as positive, negative and neutral valence, each valence category contained five words and each word was printed in four colours; blue, green, red and yellow. The words used by Isaac et al. (2012) were translated into Chinese. Participants were asked to classify the colour by pressing a different button as fast as they could. For example, when participants saw a blue or green word they had to press “Q,” whereas when they saw a red or yellow word they had to press “P.” The trial sequence of events was identical to the Facial Emotional Task displayed in Figure 2. Task instructions were presented on screen and participants completed six practice trials before proceeding to 30 experimental trials. All stimuli were translated into Chinese and appeared in font DFKai_SB and in font size 96. The tasks were presented on a computer and colour words appeared against a black background. The presentation of the emotional-colour words was randomised.

Figure 2. Mean (95% CIs) reaction times to the emotion stimuli in the Facial Stroop task in the group testing in the morning and in the afternoon.

Chronotype

Participants completed the Morningness-Eveningness Questionnaire (MEQ) (Horne and Östberg 1976). The MEQ includes 19 items that probe sleep and activity preferences, measured on a 4–5 points Likert scale that sum to a range between 16 and 86. Three groups were categorized based a score 59 and above for the Morning type, between 42 and 58 for the Intermediate type and a score of 41 and below for the Evening type. Horne and Östberg (1976), in a student population (N = 48,18–32 years), reported a Cronbach’s alpha of .84, which is considered good internal consistency.

Sleep quality

Participants completed the Pittsburgh Sleep Quality Index (PSQI) (Buysse et al. 1989) as a measure of sleep quality. The PSQI includes 19 items that ask about usual sleep in the past month that form seven components and a total score that was selected for use in this study. A higher total score indicates poorer sleep quality. Buysse et al. (1989) reported a Cronbach’s alpha of 0.83, indicating good internal consistency.

Mental health

Participants completed the Depression Anxiety Stress Scale-21 (DASS-21) (Antony et al. 1998). Items are scored 0–3 for each of 21 statements which indicates how much the statement applies to the respondent in the prior week. The instrument has seven items for each of three subscales of depression, anxiety and stress and a total composite score. A higher score indicates greater negative affectivity. The total composite score was selected for use in the present study. Cronbach’s alpha for the total score were reported as .93 (95% CI: .93–.94) in a large UK sample (Henry and Crawford 2005).

At the time of data collection, the three questionnaires were not available in Chinese. The translated Chinese versions used in the present study were reviewed and validated by a Chinese-English bilingual experienced in research with psychometric assessments.

Procedure

Participants were tested in a school’s computer lab, with the researcher and teacher present. In order to balance the number of classes with the time of day, half of the classes per grade were tested in the morning (8 a.m. to 9 a.m. or 9 a.m. to 10 a.m.), and the others were tested in the afternoon (2 p.m. to 3 p.m. or 3 p.m. to 4 p.m.), with 49% of the total sample tested in the morning. All participants completed the tasks in the same order.

Statistical analyses

Descriptive statistics were calculated for the study measures and group comparisons performed by MEQ groups and Time of Day to check for systematic differences in gender, age, PSQI and DASS scores. Pre-processing of the experimental task data found there were 7.8% errors in the lexical-emotional task and 6.9% errors in the facial-emotional task and the trials were removed before analyses of the reaction times. The experimental tasks have a hierarchical structure with trials nested within participants. The data was analysed using a two-level random intercept multilevel generalised linear mixed-effects model (GLMM) with gamma distribution and natural log link as appropriate for reaction times (Anderson et al. 2010). The level 1 data was specified as repeated measures of trials nested within participants. An autoregressive AR (1) covariance structure was selected as is appropriate for time series data. A random intercept was specified for participant with a scaled identity random covariance type. Preliminary analyses checked for effects of structural aspects of the experimental tasks (word colour/face colour or gender) and retained in the models if significant. The statistical analyses of the lexical experimental task included a fixed effect of MEQ (Morning, Intermediate, Evening), a fixed effect of Time of Day (AM, PM) a fixed effect of Emotion Valence (positive, negative, neutral). A full factorial model was specified, and interaction terms were retained if significant. Gender and age were added to the model as covariates. The data from the facial task was analysed using the same modelling approach but with a fixed effect of Emotion with 4 levels (Anger, Happy, Neutral, Sad). The alpha level in all reported comparisons from all models was Bonferroni adjusted. Developmental effects were explored by recalculating the final model from the full sample on a Younger Group (13- and 14-year-olds) and an Older Group (15- and 16-year-olds). The relationships between emotion processing task performance with sleep quality and mental health were explored using Pearson’s Product Moment correlations and simple mediation analyses using Process for SPSS. The data that support the findings of this study are openly available in Harvard Dataverse at https://doi.org/10.7910/DVN/LYR76K. All statistical analyses were performed in SPSS version 21.0 (IBM Corp., Armonk, NY).

Results

Group comparisons

The final sample was comprised of 351 participants, with 186 boys (57.4%) and 138 girls (42.6%). The mean age of students was 14.4 years (SD = 0.87 years old). The MEQ score classifications resulted in the following groups: the Morning Group (n = 41), the Intermediate Group (n = 229) and the Evening Group (n = 54). There were 170 participants tested in the morning (AM Group) and 183 in the afternoon (PM Group). There were no significant differences in gender, age, sleep quality and DASS Total scores between the AM Group and the PM Group.

Descriptive statistics and results of the MEQ group comparisons on age, gender, PSQI and DASS total scores are displayed in Table 1. Comparisons between the three MEQ groups showed the Evening Group had a significantly older mean age than the Morning group (p = .004). The Evening Group had higher reported DASS total scores (p = .008) and poorer reported sleep quality compared to both the Morning Group (p < .001) and the Intermediate Group (p = .002), who in turn had lower reported sleep quality than the Morning Group (p = .001).

Table 1. Descriptive statistics and chronotype group comparisons on demographic, sleep and mental health measures

	MEQ Group
Total N = 351	Morning (n = 46)	Intermediate (n = 248)	Evening (n = 57)	Statistic
Gender Male (%)	20 (56.5%)	148 (58.9%)	32 (56.1%)	χ^2 (2, N = 351) = 0.19, p = .906
Mean (SD) Age	14.0 (.89)	14.5 (.88)	14.5 (.83)	F (2,348) = 6.70, p = .001, ηp^2 = .039
Mean (SD) PSQI Total Scores	3.4 (2.7)	5.0 (2.7)	6.4 (2.7)	F (2,348) = 15.95, p < .001, ηp^2 = .084
Mean (SD) DASS Total Scores	35.4 (46.1)	48.2 (43.6)	63.1 (54.0)	F (2,348) = 4.65, p = .01, ηp^2 = .026

A Multivariate GLM tested for MEQ*Time of Day interaction effects on the PSQI and DASS total scores. The interaction term was significant (p = .018). Bonferroni adjusted post hoc tests revealed Evening types assessed in the afternoon reported significantly higher PSQI (p = .011) and DASS total scores (p = .014) compared to Evening types assessed in the morning. The interaction effect was independent of significant effects of gender (p < .001) on PSQI and DASS scores (higher in females) and effect of age (p < .004) on DASS scores (increased with age). The full results of the multivariate tests are presented in Table S1. The means (SE) for the DASS Total scores and the PSQI are displayed in Figure S1a and Figure S1b.

Lexical emotional Stroop task

The results of the GLMM model on the RT of correct trials are reported in Table 2. There were no significant main effects or interactions. Age was found to predict decreased reaction times (b = −0.04, SE = .119, t = −3.342, p = .001). There was a significant main effect of word colour and Bonferroni adjusted post hoc comparisons revealed RT was significantly increased when words were printed in green compared to red, yellow and blue (all p’s < .001). The GLMM models on the Younger Group and the Older Group replicated the nonsignificant findings found in the model on the full sample.

Table 2. Fixed and random effects on the reaction time data for the lexical task

	F value	p value
Fixed Effects (df)
Emotion (2)	1.70	.184
MEQ (2)	0.31	.969
Time of Test (1)	0.60	.414
Age (1)	12.72	<.001
Gender (1)	1.49	.222
Word Colour (3)	35.6	<.001
Random Effects	Estimate (SE)	Z	p value [95% CI]
Participant	0.033 (0.003)	11.89	<.001 [.028, .038]
Residual
AR1 Diagonal	0.070 (0.001)	66.41	<.001 [.068, .072]
AR1 Rho	0.125 (0.012)	10.47	<.001 [.101, .148]

Facial emotional Stroop task

The results of the GLMM model on the RT of correct trials are reported in Table 3. The Emotion main effect was significant and post hoc tests showed slower RT to Anger compared to Sad (p = .034) and Neutral (p = .015). This was qualified by an Emotion x Time of Day interaction. Figure 2 displays the means (SEs) reaction times for AM and PM Groups for each Emotion condition. In the AM Group the RT in Anger trials was significantly longer than Happy (p = .022), Sad (p = .003) and Neutral (p < .001). No further interactions were significant. An older age predicted faster reaction times (b = −0.04, SE = .119, t = −3.342, p = .001). There was a significant main effect of display face colour and Bonferroni adjusted post hoc comparisons revealed RT was significantly increased when faces were presented in green compared to red, yellow and blue (all p’s < .001).

Table 3. Fixed and random effects on the reaction time data for the facial task

	F value	p value
Fixed Effects (df)
Emotion (3)	3.86	.009
MEQ (2)	0.19	.825
Time of Test (1)	0.12	.728
Emotion x Time of Test (3)	4.14	.006
Emotion x MEQ (6)	1.88	.080
Age (1)	24.48	<.001
Gender (1)	2.23	.135
Face Colour (3)	31.90	<.001
Random Effects	Estimate (SE)	Z	p value [95% CI]
Participant	0.027 (0.002)	11.84	<.001 [.023, .031]
Residual
AR1 Diagonal	0.070 (0.001)	68.04	<.001 [.068, .072]
AR1 Rho	0.075 (0.011)	6.63	<.001 [.053, .097]

The GLMM in the Younger Group showed the Emotion x MEQ interaction was significant (p = .023). Figure 3 displays the means (SEs) reaction times for MEQ Groups for each Emotion condition. Post hoc tests showed the Evening Group had longer RTs on Anger trials compared to all other emotions (all p’s < .03). The Emotion x Time of Day interaction was no longer significant (p = .11). No other changes in significance to the GLMM parameter estimates reported in the main were found. The full results of the model are displayed in Table S2. The GLMM computed on the data of the Older Group showed that with the exception of the effect of Face Colour (p < .001) no other significant effects remained (see Table S3). The inclusion of PSQI and DASS Total scores as covariates did not predict RTs or modify the effects found.

Figure 3. Mean (95% CIs) reaction times to the emotion stimuli in the Facial Stroop task in Younger Morning, Intermediate and Evening types (collapsed across morning and afternoon testing).

Chronotype, emotion processing, sleep quality and mental health

To test for associations with the slowed RT response to Anger, each participant’s mean RT predicted values were derived from the GLMM and Anger RT was adjusted by subtracting mean Neutral RT to represent a difference score (A-N). Simple correlations showed the A-N scores negatively correlated with PSQI and DASS Total scores in the Evening Chronotype group (Table 4). A smaller difference between Anger RT and Neutral RT was associated with higher PSQI and DASS-21 total scores. The correlations between A-N and PSQI and DASS-21 scores were non-significant in the Morning and Intermediate groups. The relationship between PSQI and DASS-21 scores was significant in the Morning Chronotype (r = .70 (46) p < .001), Intermediate Chronotype (r = .53 (248) p < .001) and Evening Chronotype Group (r = .62 (57) p < .001).

Table 4. Results of correlations between Anger minus Neutral difference score (A-N) and Sleep Quality (PSQI) and Mental Health (DASS-21 Total)

		PSQI Total	DASS-21 Total
Chronotype Group		r (p value)	r (p value)
Morning (n = 46)	A – N	−.10 (.49)	.06 (.68)
Intermediate (n = 248)	A – N	.02 (.77)	−.05 (.48)
Evening (n = 57)	A – N	−.38 (.003)	−. 36 (.005)

A simple mediation analysis was performed to test if PSQI mediated the relationship between A-N and the DASS-21 total scores as the outcome. The analysis revealed that when controlling for sleep quality, the A-N scores were not a significant predictor of mental health outcomes, b = −0.46, t (54) = −1.31, p = .19. The standardized indirect effect was −.21. The significance of this effect was tested using bootstrapping. Unstandardized indirect effects were computed for each of 5000 bootstrapped samples, at the 95% confidence interval. The unstandardized indirect effect was −.66, and the 95% confidence interval ranged from −1.14, −.28. Thus, the indirect effect was statistically significant.

Discussion

The study found no evidence of synchrony effects between chronotype group and time of day in performance on the emotional processing tasks. The largest effect observed was the significantly delayed reactions to anger expressions in the facial Emotional Stroop task in the group tested in the morning. This attentional interference to anger displays was also evident in younger evening chronotypes independent of the time of day and was not related to sleep quality or mental health. In contrast, it was the relative absence of this threat-related attentional interference that characterised emotion processing in evening chronotypes with greater mental health difficulties. A mediation analysis found this relationship to be fully explained by the poorer sleep quality reported by this group.

The findings from the emotional Stroop tasks suggest there is an attentional sensitivity in the morning hours to facial signals of threat. The anger-related interference observed here could reflect the experience of negative affect reported by adults in the morning hours that dissipate through the day until evening time (Ayuso-Mateos et al. 2013). There is however a dearth of studies on diurnal variation in facial emotion processing. An imaging study conducted with a large sample of young adults observed increased amygdala reactivity to emotional faces that included threat stimuli (anger, fearful) when the scan was conducted in the morning hours (Baranger et al. 2017). This was attributed to diurnal variation in hypothalamic-pituitary-adrenal (HPA) axis and the related peak in cortisol after awakening that decreased throughout the day. In a study with adults, the administration of cortisol has been observed to increase, as opposed to delay, reaction times to all emotional stimuli in the emotional Stroop (Bertsch et al. 2011), and higher levels of basal cortisol are associated with avoidance away from angry faces (Van Honk et al. 1998). Attentional vigilance toward angry facial cues, in contrast, is associated with increased testosterone to cortisol ratios that are involved in social aggression and motivation for social dominance (Montoya et al. 2012). There is emergent theoretical and empirical evidence from animal models of an interplay between the circadian system and the expression of anger and aggression (Hood and Amir 2018). The dual-hormone hypothesis also predicts that testosterone increases status-seeking and social dominant behaviours when cortisol levels are lower. In early to mid-adolescence, cortisol levels decrease and the circadian rhythm flattens with reduced morning levels and a relative increase in the evening time that is observed with trait-like stability (Shirtcliff et al. 2012). Testosterone levels decline throughout the day, and there is a high level of shared within-person variability in the diurnal rhythms of testosterone and cortisol (Harden et al. 2016). Chronotype may moderate this hormone-behaviour relationship and could be evidenced by altered attentional responses to social emotional cues. Adolescents with greater eveningness are more likely to self-report an anger-oriented temperament, independent of sleep factors (Jankowski and Linke 2020), and there is also evidence of a vulnerability toward aggressive and antisocial behaviour in children and adolescents with an eveningness profile (Schlarb et al. 2014). This anger-related bias therefore would appear consistent with these other behavioural observations.

Prior developmental research on facial emotion processing has observed a marked increase in recognition of anger facial expressions in early to mid-adolescence (Thomas et al. 2007). Anger facial expressions signal social dominance and perceptual sensitivity to displays of anger are relevant for the negotiation of social hierarchies (Blair 2003). There is also a heightened importance of status amongst peers in the early and mid-adolescent period (Oberle et al. 2010). The persistent attentional sensitivity to anger displays in the younger evening chronotypes may be evidence of an adaptive influence of circadian delayed phase on social attention. One prior study reported intermediate and later phase preference predicted higher subjective social status amongst peers that is consistent with this hypothesis (Lunn et al. 2021). In the present study, younger evening chronotypes showed a similar response pattern to the overall sample. It is proposed here therefore that chronotype could partly explain some trait-level variation in social attention processes involved in typical adolescent social development.

The absence of the threat-related interference in evening types with poorer sleep quality is a pattern consistent with neurophysiological studies on the effects of sleep deprivation on emotion processing in adults. An ERP study using emotional picture stimuli observed attenuated differences in the processing of neutral versus emotion pictures after sleep deprivation, attributed to reduced attentional discrimination between the neutral and emotional stimuli (Alfarra et al. 2015). Chayo et al. (2017) also observed similar effects in testing in the evening, whereby attentional processing of neutral stimuli was of similar magnitude to the emotional. The authors argued the increased homeostatic sleep pressure in the evening time likely resulted in the same interference in attention resource allocation as sleep deprivation.

The delayed phase preference in adolescence involves changes in both circadian timing and homeostatic sleep pressure (Borbély 1982; Hagenauer et al. 2009). A slower increase in homeostatic sleep pressure during wakefulness has been observed in young adult evening types (Taillard et al. 2003), and the rate of accumulation in homeostatic sleep pressure also slows through the adolescent period (Crowley et al. 2014). A slower rate of increase in homeostatic sleep pressure in evening chronotypes could be one potential explanation for the maintained difference between anger and neutral signals observed in the younger group. A testable hypothesis is chronotype-related differences in homeostatic sleep pressure contribute to diurnal variation in emotion processing in typical development, and evening types may also be more vulnerable to disruptions with poorer sleep quality.

The high co-occurrence of lower sleep quality and increased mental health difficulties is reported is typical adolescent populations worldwide (Gradisar et al. 2011), and note the reported problems in the present sample were in the mild to moderate range. The attenuated emotional processing in evening types is in contrast to the threat-related attentional biases more frequently observed in clinical populations with sleep dysfunction (Blake et al. 2018). It raises the possibility that reduced attentional discrimination between emotion signals is an acquired deficit from the accumulative sleep deprivation with increasing age seen in adolescence (Loessl et al. 2008).

Limitations

The use of a between-subjects design for the time-of-day testing factor precluded comparisons on emotion processing by the same individual in the morning and afternoon. The chronotype and time of day interaction on PSQI and DASS instruments suggests evening types may display diurnal changes in mood-congruent memory bias that requires further external validation. Prior research on threat-related attentional biases included fearful stimuli. The lack of a fear stimulus in the present study precluded comparisons between the two threat-related signals. The translated version of the Verbal Emotional Stroop Task was not validated and may explain the lack of effects found in this study. Time of the test was restricted to the daytime and not evening as the study was conducted in school setting during the school hours. The MEQ scores were transformed into a chronotype group at point of data entry and raw scores were not recorded digitally for use as a continuous variable.

Conclusion

The behavioural study has found evidence of diurnal variation in threat-related attentional bias in a typical population of adolescents. The attentional capture to anger displays in the morning hours could reflect heightened sensitivity to social emotional cues that communicate information about social aggression and dominance. This may be developmentally relevant given the importance of social hierarchies during this developmental period, but this hypothesis requires further comparisons with younger children and adults. Although there is evidence of lower mood and reported affect it the morning, there is limited information on patterns of anger expression or aggression across the day that requires further investigation. The relative absence of attentional responses to social emotional cues in those with greater eveningness observed here contrasts with threat-related attentional biases observed in clinical populations. It is plausible that attenuated social attention processes may distinguish evening type subgroups and reflect different mechanisms between sleep factors, emotion regulation and the increased mental health difficulties frequently observed in adolescents.

Acknowledgements

We would like to thank Ming-Der Senior High School for their participation in this research.

Disclosure statement

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

Additional information

Funding

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