Chronotype and time-of-day effects on spatial working memory in preschool children

Abstract

Study Objectives:

Spatial working memory (SWM) capacity subserves complex cognitive functions, yet it is unclear whether individual diurnal preferences and time-of-day influence SWM in preschool children. The main and interaction effects of chronotype and time-of-day on SWM and SWM differences in preschoolers with different chronotypes within each time-of-day group will be examined.

Methods:

We studied a subset of typically developing 4.5-year-olds taking part in a birth cohort study (n = 359). The Children’s Chronotype Questionnaire categorized children into morning-, intermediate-, and evening-types. Using a computerized neuropsychological test (Cambridge Neuropsychological Test Automated Battery), SWM was determined from the total number of between-search errors (ie, between search-total errors) and Strategy scores. Higher between search-total errors or lower Strategy scores indicated worse SWM. Time-of-day was categorized into late morning (10:00 am to 11:59 am), afternoon (12:00 pm to 3:59 pm), and late afternoon (4:00 pm to 6:30 pm). In a subsample (n = 199), caregiver-reported chronotype was validated using actigraphy-measured sleep midpoint.

Results:

After controlling for ethnicity, no significant main and interaction effects of chronotype and time-of-day on between search-total errors and Strategy scores were seen (all P > .05). However, evening-types outperformed morning-types (ie, lower mean between search-total errors) in the late afternoon (P = .013) but not in the late morning and afternoon (all P > .05). Actigraphy data in the subsample confirmed that evening-types had later sleep midpoints during weekdays and weekends (P < .001).

Conclusions:

Since evening-type preschoolers had better SWM in the late afternoon compared to morning-type preschoolers, this gives insights into optimal learning opportunities in early childhood education.

Citation:

Abdul Jafar NK, Tham EKH, Eng DZH, et al. Chronotype and time-of-day effects on spatial working memory in preschool children. J Clin Sleep Med. 2023;19(10):1717–1726.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Spatial working memory (SWM) capacity increases throughout childhood and subserves complex cognitive functions, yet it is unclear whether individual diurnal preferences and time-of-day influence SWM performance in preschool children. The current study examined the main and interaction effects of chronotype and time-of-day on SWM in preschoolers and determined whether SWM differed in preschoolers with different chronotypes within each time-of-day group.

Study Impact: This study indicates that late afternoon time of the day may differentiate SWM performance between evening-type compared to morning-type preschoolers. Future research in this area will offer further insights into understanding optimal learning opportunities in early childhood education.

INTRODUCTION

Spatial working memory (SWM) involves the ability to temporarily store and manipulate visuospatial (nonverbal) information that subserves complex cognitive functions.¹ SWM plays an important role in children’s neurodevelopment and in school achievement.^2,3 A major indication of cognitive development in children involves the increase in working memory capacity throughout childhood.^4–7 Working memory capacity is essential in allowing children to actively combine concepts and ideas and connecting novel information with long-term memory information.⁸ Thus, this facilitates more knowledge accumulation as children develop.⁸ Therefore, it is important to explore factors that potentially influence working memory in young children.

There are several factors that may impact performance on measures of cognitive ability in childhood, including individual characteristics (eg, sex, ethnicity, birth order, gestational age),^9–12 intelligence,¹³ maternal education,¹⁴ and sleep.^15–17 In addition to these factors, SWM may also be influenced by the function of chronotype and time-of-day.^18–21 Chronotype refers to individual diurnal preferences on a continuum from early to late sleep and wake times, which can influence the preferred time-of-day for optimal functioning.²¹ Morning-types tend to wake up early and are more active in the morning, while evening-types prefer waking up late and are more active in the late afternoon and evening.²² Intermediate-types have fewer extreme diurnal preferences.²² During development, young children tend to have morning peaks in arousal, while adolescents shift to eveningness, which continues through young adulthood, shifting back to morningness later in life.^22,23 Despite younger children’s (kindergarten and preschool) predisposition to morningness, individual differences in chronotype can still occur with children exhibiting evening or intermediate chronotypes.^23–25

Previous studies in young adults that investigated chronotype and time-of-day effects on cognitive ability are inconsistent.^18,19,26,27 A meta-analysis by Preckel et al (2011) demonstrated that eveningness is positively correlated with cognitive ability.²⁶ For example, individuals who were more extreme evening-types tend to perform well on measures of processing speed, memory, and cognitive ability, even when those cognitive tasks were executed early in the morning.¹⁹ In contrast, other studies suggest that relations between chronotype and cognitive ability are variegated and contingent upon the type of cognitive domain.^18,27 For instance, a study reported that individual with eveningness tendencies performed significantly better on the spatial subtest of the Multidimensional Aptitude Battery IQ compared to any other subtest.¹⁸ In preschool children, Zarch et al (2018) showed that preschoolers generally performed the best for working memory in early morning (8:00 am) and the worst in the afternoon (1:00 pm), but performance significantly improved toward late afternoon (> 3:00 pm), compared to early afternoon.²⁸ Furthermore, morning-type preschoolers outperformed intermediate- and evening-type preschoolers in the early morning and vice versa in the late afternoon.²⁸ Taken together, these studies posit that, even in preschool children, differences in cognitive performance can be chronotype- and time-dependent.

Although most studies that examined the interaction between chronotype and time-of-day (ie, synchrony effect) showed better cognitive performance at optimal times of the day (ie, morning-types perform better in the morning while evening-types perform better in the evening) according to chronotype,^29–32 little is known about how chronotype and time-of-day contribute to SWM performance in preschool children. Therefore, in the present study, we aimed to assess (1) whether typically developing Singaporean preschoolers who are evening-types exhibit better SWM performance than those who are morning- or intermediate-types, (2) whether SWM performance overall and respective to each chronotype is better in the late afternoon (4:00 pm to 6:30 pm) than late morning (10:00 am to 11:59 am), or afternoon (12:00 pm to 3:59 pm), and (3) if there is a synchrony effect between chronotype and time-of-day in SWM performance. We hypothesized that evening-type preschoolers have better SWM performance compared to their morning- and intermediate-type counterparts, especially in the late afternoon time-of-day.

METHODS

Study design and sample

The current study was performed as part of the Growing Up in Singapore Toward healthy Outcomes (GUSTO) birth cohort study. Pregnant women ages ≥ 18 years (n = 1450) were recruited during the first trimester of pregnancy from two major public hospitals in Singapore: KK Women’s and Children’s Hospital and the National University Hospital. All births occurred between November 30, 2009, and May 1, 2011. GUSTO study participants and research procedures have been described further in detail in a previous publication.³³ In the current study, we included 1181 singleton infants and excluded infants whose mothers had twin pregnancies (n = 10). Nonparticipation at this stage (n = 259) was due to loss to follow-up. Out of the 1181 infants, only 487 mother–offspring pairs that participated in the cohort study attended the center visit at 4.5 years old. In the final analysis, n = 359 had completed data for both the Children’s Chronotype Questionnaire (CCTQ) and SWM task and met the following inclusion criteria: having a last recorded Apgar score ≥ 9 (either at 5 or 10 min), birth weight ≥ 2500 g to ≤ 4000 g, gestational age ≥ 37 weeks (Figure 1).

Figure 1. Study flow chart.

Flow chart describing the data available for CCTQ and SWM outcomes (ie, between search-total errors and Strategy scores). Thick solid arrows indicate the distribution of participants at each stage of the study and dotted arrows represent participants that were excluded from this study. CCTQ = Children’s Chronotype Questionnaire, SWM = spatial working memory.

Written, informed consent was obtained from the legal guardian of each child. The GUSTO study was approved by the National Health Care Group Domain Specific Review Board and the SingHealth Centralized Institutional Review Board.

Measures

CCTQ

The CCTQ was completed by caregivers in English, Chinese, Malay, or Tamil according to their dominant language during the center visit when children were 4.5 years of age (± 2 months). The CCTQ was forward and backward translated to Chinese, Malay, and Tamil by accredited professional translation services specializing in East Asian and local languages. These forward and backward translations were independently checked by at least two native Chinese, Malay, and Tamil English-speaking persons. Following consensus discussions with all translators, single Chinese, Malay, and Tamil versions were obtained. Internal consistency of the English version as well as the combined English and language translations for CCTQ were assessed by Cronbach’s α coefficients, which showed similar results for both the English as well as the combined translations.²⁴ The CCTQ includes a morningness/eveningness scale score that has been validated for use in 4- to 11-year-old children across various populations, including Asian populations in which scores ≤ 23 are classified as morning-types, scores of 24 to 32 as intermediate-types, and scores ≥ 33 as evening-types.^34–36

SWM task

During the center visit, the child also participated in a computerized SWM task as part of a battery of cognitive tests that was administered by a trained research coordinator. The SWM task is a self-ordered searching task taken from the Cambridge Neuropsychological Test Automated Battery that enabled participants to maintain and update spatial information in working memory.³⁷ Participants were required to search through several boxes for hidden tokens. Only one token is hidden at one time as participants progressed through each trial. The token would never be hidden in the same box, and, thus, individuals were told not to return to the same boxes where a token had been previously found. Trials increased in difficulty according to the number of boxes (4, 6, and 8 boxes). Practice trials (3 boxes) were performed preceding test trials (4, 6, and 8 boxes) administration. Performance on the task was evaluated by quantifying (1) between-search (BS) errors in which the participant returned to search a box where a token had already been found and (2) strategy score that estimates the extent in which the individual follows a predetermined sequence by beginning with a specific box and then, once a token has been found, to return to that box to start a new searching sequence. BS-total errors were calculated by summing the total number of between-search errors from each task level of 4, 6, and 8 boxes. Low scores on BS-total errors indicate a small number of errors and hence better SWM performance. A low strategy score represents a more consistent use of the effective search strategy.

The SWM task took place on both weekdays and weekends (51.5% vs 48.5% out of n = 359). Task timings (ie, start and end times) to complete the SWM task were also recorded. Data from the SWM task was stratified into three time-of-day periods according to their task timings: late morning (10:00 am to 11:59 am), afternoon (12:00 pm to 3:59 pm), and late afternoon (4:00 pm to 6:30 pm).

Actigraphy

Sleep-wake behavior was observed in a subsample of 4.5-year-old children (n = 199) for 4 days, including at least 1 day of the weekend, using actigraphy (Actiwatch 2; Philips Respironics, Murrysville, PA). Data were obtained in 1-minute intervals and scored with Actiware software version 6.0.2 (Philips Respironics) using the automatic sensitivity threshold for sleep-wake staging. Chronotype is independent of sleep duration,³⁸ and the measurement of sleep midpoint was used as a single phase-marker to characterize individuals. Sleep midpoint was calculated from actigraphy data as the midpoint between sleep onset and sleep offset.^39,40 Further details about actigraphy data collection and coding procedures have been described in a previous publication.²⁴

Covariates and other data

During the antenatal period, socioeconomic and demographic data such as birth order, ethnicity, and maternal education were obtained from questionnaires administered to mothers by researchers. Birth outcomes such as birth weight, sex of child, Apgar scores, and gestational age were recorded by midwives at delivery. Night-sleep and day-sleep durations were derived from the Children’s Sleep Habits Questionnaire reported by caregivers during the center visit.⁴¹ Day- and night-sleep durations were calculated from the Children’s Sleep Habits Questionnaire from the questions, “Write in child’s bedtime/usual naptime” and “Write in the time of day child usually wakes in the morning/after nap” covering the past week.

Data were checked for extreme outliers and missing data. Of the 427 sample of preschool children who met the inclusion criteria, 3 were identified as having extreme outliers from boxplot distributions of BS-total errors and were removed from the analyses; n = 4 had missing data for maternal education, n = 13 for night-sleep duration (from Children’s Sleep Habits Questionnaire), and n = 29 for day-sleep duration (from Children’s Sleep Habits Questionnaire). Hence, the final analytical sample included 359 mother-offspring pairs (Figure 1).

Statistical analysis

Previous studies have associated child’s sex, maternal education, ethnicity, birth order, and sleep duration with cognitive development in young children.^{9–11,14–17} We therefore determine potential covariates using one-way analysis of variance (ANOVA) with Bonferroni-corrected post hoc pairwise comparisons (threshold P value < .017), Pearson correlations, independent sample t tests, and chi-square tests to compare participants’ characteristics (child’s sex, ethnicity, maternal education, birth order, night- and day-sleep durations) with chronotype (ie, CCTQ) and SWM outcomes (ie, BS-total errors and Strategy scores) (Table 1). To determine whether participants’ characteristics and study variables were influenced by test administration, chi-square tests and independent sample t tests were used to assess differences in participant characteristics, chronotype, and SWM outcomes on time-of-day (ie, late morning, afternoon, and late afternoon periods) and/or weekdays vs weekends. Since only ethnicity differed with respect to BS-total errors, in the main analysis, two-way ANOVA models, unadjusted and adjusted for ethnicity, were performed to test the main effects of chronotype and time-of-day as well as their interaction effects on SWM outcomes (Figure 2). Regardless of whether the interaction effects were significant, we still pursued our a priori hypothesis to analyze the differences in SWM performance in children with different chronotypes within each time-of-day group. Bonferroni-corrected post hoc pairwise comparisons (threshold P value < .017) were conducted when the main effect of chronotype or time-of-day showed significance. Out of those preschoolers who attended the center visit and met the inclusion criteria (n = 427), chi-square tests, two-sided Fisher’s exact test, and independent sample t tests were conducted to compare participants’ characteristics between those preschoolers who had completed data for both CCTQ and SWM (n = 359) and were included in the final analysis relative to those who had no data for either or both CCTQ and SWM (n = 68) and were not included in the final analysis (see Table S1 ^{(239.4KB, pdf)} in the supplemental material). Using a subsample of n = 199, chronotype was validated using Pearson correlations to determine the associations between chronotype (from CCTQ-morningness/eveningness scores) and sleep midpoint (from actigraphy) for all days (ie, weekdays and weekends), weekdays, and weekends, respectively. Pearson correlations were also used to investigate the relation between actigraphy sleep midpoint and SWM outcomes. Partial-eta-squared (η²_p) was used as a measure of effect size. Analyses were done with SPSS software version 26.0 (IBM, SPSS, Armonk, NY). Unless otherwise specified, all P values were two-sided, and the level of significance was considered at P < .05.

Table 1.

Comparison of participant characteristics across CCTQ, BS-total errors, and Strategy scores (n = 359).^†

Characteristics	Mean (SD)/n (%)††	CCTQ				BS-total Errors		Strategy Scores
		M-type (n = 37)	I-type (n = 215)	E-type (n = 107)	P *	Mean (SD)	P *	Mean (SD)	P *
Sex	359				.61		.43		.12
Male	174 (48.5)	19 (51.4)	105 (48.8)	50 (46.7)		72.87 (14.10)		37.37 (3.22)
Female	185 (51.5)	18 (48.6)	110 (51.2)	57 (53.3)		71.78 (12.04)		37.91 (3.25)
Maternal education	355				.005		.55		.88
<Postsecondary	111 (31.3)	16 (45.7)	74 (34.7)	21 (19.6)		73.38 (12.41)		37.66 (3.42)
Postsecondary	121 (34.1)	11 (31.4)	74 (34.7)	36 (33.6)		72.19 (13.97)		37.53 (3.30)
≥University	123 (34.6)	8 (22.9)	65 (30.6)	50 (46.8)		71.53 (12.70)		37.74 (2.95)
Ethnicity	359				.002		.001		.71
Chinese	203 (56.5)	15 (40.5)	111 (51.6)	77 (72.0)		70.09 (12.09)		37.74 (3.33)
Malay	103 (28.7)	15 (40.5)	67 (31.2)	21 (19.6)		75.85 (14.87)		37.63 (3.26)
Indian	53 (14.8)	7 (19.0)	37 (17.2)	9 (8.4)		73.91 (11.34)		37.32 (2.91)
Birth order	359				.36		.89		.33
First born	164 (45.7)	15 (40.5)	93 (43.3)	56 (52.3)		72.41 (13.34)		37.39 (3.27)
Second born	101 (28.1)	13 (35.1)	59 (27.4)	29 (27.1)		72.65 (11.77)		37.99 (3.21)
≥Third born	94 (26.2)	9 (24.4)	63 (29.3)	22 (20.6)		71.77 (14.03)		37.72 (3.23)
Night sleep duration (h)	9.67 (0.92)	9.70 (0.85)	9.75 (0.97)	9.52 (0.82)	.11		.84		.46
Day sleep duration (h)	1.36 (0.93)	0.93 (0.93)	1.43 (0.96)	1.36 (0.84)	.022		.59		.59

^†Values are presented as frequencies, percentages, means, and standard deviations (SD).

^††Missing data for n = 4 for maternal education, n = 13 for night sleep duration (from CSHQ), and n = 29 for day sleep duration (from CSHQ).

^*P-values derived from independent sample t-tests and chi-square tests for categorical variables, and one-way analysis if variance with Bonferroni-corrected post hoc pairwise comparisons (threshold P value < .017) and Pearson correlation tests (pairwise) for continuous variables. CCTQ = Children’s Chronotype Questionnaire; CSHQ = Children’s Sleep Habits Questionnaire; BS-total errors = total number of between search errors; M-type = morning-type; I-type = intermediate-type; E-type = evening-type.

Figure 2. Bar graphs (unadjusted) of means and 95% confidence intervals of spatial working memory performance across chronotype by time-of-day.

(A) Two-way ANOVA test revealed overall significant effect of chronotype (F(2, 350) = 3.09, P = .047) but no significant effect of time-of-day (P = .88) on BS-total errors. (B) Two-way ANOVA test was not significant for chronotype (P = .15) and time-of-day (P = .74) on Strategy scores. Post hoc Bonferroni-adjustment results for multiple tests are indicated. ANOVA, analysis of variance, CI = confidence interval, BS-total errors = total number of between search errors.

RESULTS

Participant characteristics

Based on the chronotype categories defined using the CCTQ- morningness/eveningness scores,³⁴ our sample included morning- (n = 37, 10.3%), intermediate- (n = 215, 59.9%) and evening-type preschoolers (n = 107, 29.8%). Sample distribution across time-of-day comprised of late morning (n = 84, 23.4%), afternoon (n = 155, 43.2%), and late afternoon (n = 120, 33.4%). Chi-square tests and one-way ANOVA tests revealed no significant differences for all participants’ characteristics across time-of-day (all P > .05), except for day sleep duration (P = .009) in which preschoolers who attended the late afternoon session demonstrated significantly shorter day-sleep duration (mean = 1.17 h, standard deviation (SD) = 0.94) compared to those preschoolers who attended the afternoon session (mean = 1.52 h, SD = 0.91, P = .008). There were no significant differences in day-sleep duration between late morning (mean = 1.31 h, SD = 0.92) and afternoon (P = .30) as well as late afternoon (P = .94) sessions, respectively. Chi-square tests and independent sample t tests revealed no significant differences for all participants’ characteristics on weekdays vs weekends (all P > .05).

Table 1 compares CCTQ categories, BS-total errors, and Strategy scores across different participant characteristics. There were no statistically significant differences in CCTQ categories across child’s sex, birth order, and night sleep duration (all P > .05); however, we observed a greater proportion of evening-type compared to morning- and intermediate-type preschoolers belonging to the offspring of mothers with ≥ University education (46.8% vs 22.9%; 46.8% vs 30.6%, respectively), and of Chinese ethnicity (72.0% vs 40.5%; 72.0% vs 51.6%, respectively). Moreover, morning-type compared to intermediate-type preschoolers exhibited significantly shorter day-sleep duration (mean = 0.93, SD = 0.93 vs mean = 1.43, SD = 0.96, P = .017) but not significant when compared to evening-types (mean = 1.36, SD = 0.84, P = .078). There were no statistically significant differences in BS-total errors across child’s sex, maternal education, birth order, or night- and day-sleep durations (all P > .05), except for ethnicity (P = .001). Significantly lower mean BS-total errors were seen in Chinese compared to Malay preschoolers (mean = 70.09, SD= 12.09 vs mean = 75.85, SD = 14.87, P = .001) but were not significant when compared to Indian preschoolers (mean = 73.91, SD = 11.33, P = .17). There were no significant differences in Strategy scores for all participant characteristics (all P > .05).

There were no statistically significant differences in maternal education, ethnicity, birth order, or night- and day-sleep durations between those preschoolers who were included (n = 359; 84.1% of 427 participants) and those preschoolers who were not included in our final analysis (n = 68; 15.9% of 427 participants) (all P > .05). However, those preschoolers who were included in the final analysis more likely to be females than males (P = .034) (see Table S1 ^{(239.4KB, pdf)} ).

Chronotype and SWM

Figure 2 depicts the distribution of chronotype by time-of-day on mean BS-total errors and Strategy scores, respectively. In the unadjusted model, two-way ANOVA results revealed overall significant main effect of chronotype on mean BS-total errors (F(2, 350) = 3.09, P = .047, η²_p = 0.017) with evening-type preschoolers (n = 107, mean = 70.37, SD = 12.45) committing fewer mean BS-total errors relative to morning-type preschoolers (n = 37, mean = 76.32, SD = 13.46, P = .016) but not relative to intermediate-type preschoolers (n = 215, mean = 72.58, SD = 13.19, P = .13) (Figure 2A). No significant mean differences in BS-total errors committed overall between intermediate-type preschoolers relative to morning-type preschoolers (P = .11) (Figure 2A). However, when adjusted for ethnicity, no significant main effect of chronotype on mean BS-total errors were observed (P = .15). Overall, no significant main effect of chronotype on mean Strategy scores was seen in either unadjusted (P = .72) or adjusted models (P = .77).

Time-of-day and SWM

Mean BS-total errors across time-of-day were late morning (mean = 72.10, SD = 12.93), afternoon (mean = 73.18, SD = 13.23), and late afternoon (mean = 71.33, SD = 13.0). Mean strategy scores across time-of-day were late morning (mean = 37.98, SD = 3.75), afternoon (mean = 37.42, SD = 3.13), and late afternoon (mean = 37.71, SD = 3.01). Chi-square tests and independent sample t tests revealed no significant differences for CCTQ categories across late morning (morning-type: n = 9, 24.3%; intermediate-type: n = 55, 25.6%; evening-type: n = 20, 18.7%), afternoon (morning-type: n = 15, 40.5%; intermediate-type: n = 89, 41.4%; evening-type: n = 51, 47.7%), and late afternoon (morning-type: n = 13, 35.2%; intermediate-type: n = 71, 33.0%; evening-type: n = 36, 33.6%) time-of-day (P = .69) as well as on weekdays (morning-type: n = 16, 43.2%; intermediate-type: n = 115, 53.5%; evening-type: n = 54, 50.5%) vs weekends (morning-type: n = 21, 56.8%; intermediate-type: n = 100, 46.5%; evening-type: n = 53, 49.5%, P = .50). Mean BS-total errors and Strategy scores also did not differ significantly between weekdays (mean = 72.96, SD = 12.90; mean = 37.52, SD = 3.22, respectively) and weekends (mean = 71.62, SD = 13.25; mean = 37.78, SD = 3.28, respectively) (all P > .05). In the unadjusted model, two-way ANOVA results revealed overall no significant main effect of time-of-day on mean BS-total errors (P = .88) or Strategy scores (P = .74) (Figure 2A and Figure 2B, respectively). Results remained not significant after adjusting for ethnicity (P = .95 and P = .71, respectively).

Synchrony effect (chronotype*time-of-day) and SWM

There were no significant interaction (synchrony) effects between chronotype and time-of-day on mean BS-total errors (P = .46) and Strategy scores (P = .31) (Figure 3). Results remained not significant after adjusting for ethnicity (P = .31 and P = .28, respectively).

Figure 3. Line graphs depicting no significant interaction effects between chronotype and time-of-day on spatial working memory performance.

(A) Mean BS-total errors (P = .46) and (B) Mean Strategy score (P = .31) from unadjusted models. BS-total errors = total number of between search errors.

Given that a previous study found time-of-day effects between different chronotypes,²⁸ we further examine the differences in SWM in children with different chronotypes within each time-of-day group. With respect to only late afternoon time-of-day, we found significant overall mean differences in BS-total errors between chronotype (F(2, 117) = 4.43, P = .014, η²_p = 0.017) (Figure 2A). Bonferroni-corrected post hoc pairwise comparisons showed that evening-types (n = 36, mean = 68.14, SD = 11.50) relative to morning-types (n = 13, mean = 80.31, SD = 10.36) but not with intermediate-types (n = 71, mean = 71.31, SD = 13.51, P = .67), committed significantly lesser mean BS-total errors in the late afternoon (P = .011) (Figure 2A). Significance of results remained even after adjusting for ethnicity (F(2, 116) = 4.32, P = .015, η²_p = 0.069), with evening-types (n = 36, adjusted mean = 68.59, SD = 2.14) committing lesser mean BS-total errors relative to morning-types (n = 13, adjusted mean = 80.53, SD = 3.50, P = .013). There were no mean differences in BS-total errors between chronotype in the late morning (P = .57) or afternoon (P = .83).

Validation of chronotype and SWM with sleep midpoint

In our subsample (n = 199), the actigraphy-estimated average sleep midpoint (SD in hours) was 03:25 (0:50). Average sleep midpoints for morning- (n = 23, 11.6%), intermediate- (n = 113, 56.7%), and evening-type (n = 63, 31.7%) preschoolers during all days were, respectively, 02:48 (0:46), 03:25 (0:44), and 03:40 (0:56). Average sleep midpoints for morning-, intermediate-, and evening-type preschoolers during weekdays were, respectively, 02:41 (0:44), 03:17 (0:44), and 03:31 (0:55) and during weekends were, respectively, 03:07 (0:59), 03:44 (0:54), and 04:04 (01:09). Hence, preschoolers in our subsample exhibited later sleep midpoints during weekends compared to weekdays, regardless of their chronotype. Pearson correlation results revealed that later chronotype was associated with later sleep midpoints for all days (r = .36, P < .001), during weekdays (r = .34, P < .001) as well as weekends (r = .33, P < .001). No significant correlations were seen between actigraphy sleep midpoints with BS-total errors and Strategy scores for all days (r = –.028, P = .69; r = .051, P = .47), weekdays (r = –.021, P = .76; r = .051, P = .48) as well as weekends (r = –.045, P = .53; r = .06, P = .40), respectively.

DISCUSSION

We found no significant differences in SWM performance in typically developing preschoolers after adjusting for ethnicity. When examining SWM performance by chronotype and time-of-day, we found that only in the late afternoon, evening-types demonstrated significantly lesser mean BS-total errors than morning-types but not with intermediate-types. This implies that differences in SWM performance were mainly observed between extreme chronotypes (ie, morning- vs evening-types) and most evident in the late afternoon time of the day (4:00 pm to 6:30 pm). To the best of our knowledge, this is the first cohort study to examine, in typically developing multiethnic Asian (Singaporean) preschool children, the effects of chronotype and time-of-day on SWM.

Although most studies in young adult populations (mean ages 15–23 years) showed that eveningness were more likely to outperform morningness on working memory and cognitive ability,^19,26 our findings overall did not show significant differences in SWM performance in children with different chronotypes. Despite not reaching significance, our evening-type preschoolers still perform marginally better in terms of committing fewer mean BS-total errors compared to morning-types, even in the late morning. This finding implies that perhaps differences in SWM performance between chronotypes are less evident in younger children compared to young adults due to the marked variability in sleep timing during the early years of life.³⁶

The time-of-day along with the synchrony effect on cognitive performance has been well established in previous research.^18,42,43 Our study, however, did not find any significant interaction effects between chronotype and time-of-day, suggesting that these two variables are not adjusted to each other and morning-, intermediate-, and evening-type preschoolers did not show SWM performance differences during different times of the day. Notably, when we examine differences in SWM in children with different chronotypes within each time-of-day group, our evening-type preschoolers perform better than morning-types (ie, lower mean BS-total errors) in the late afternoon. In addition to our study, another study in preschool children also showed that late afternoon or morning (early or late) rather than afternoon may distinguish working memory performance between evening- and morning-types.²⁸ Thus, to be independent of chronotype effects, future assessments on working memory in preschool children may consider administrating the cognitive task in the afternoon (between 12 pm and < 4 pm). Our null interaction findings may result from generalized assumptions on the optimal functioning times based on chronotype, from the use of varying cognitive measures, or from interindividual differences in arousal levels.⁴⁴ Hence, future studies should explore factors that can influence individual differences in chronotype and time-of-day preferences across various cognitive domains.

Additionally, a strength of this study was that, in a subsample of participants, actigraphy was used to validate the chronotype (derived from CCTQ) and SWM outcomes of preschoolers. Our evening-types indeed had later sleep midpoints during weekdays and weekends compared to morning- and intermediate-types. We also found that regardless of their chronotype, our preschoolers generally had later sleep midpoints and much later during weekends than weekdays. This validates our finding that our sample comprised of predominantly evening-types (29.8%) than morning-types (10.3%), which is different from previous studies that found younger children to be predominantly morning-types.^34–36,45 Although we found no significant correlations between actigraphy sleep-midpoint and SWM outcomes, our findings showed a trend toward preschoolers with later sleep midpoints to attain better SWM performance (ie, lower BS-total errors and higher Strategy scores). More research is needed to validate objective sleep measures with SWM performance in young children. Later sleep midpoints may be attributed to parental involvement in children’s bedtimes in which bedtime discipline during childhood can modulate the extent of shift to eveningness, hence affecting their chronotype.^46,47 For example, parents may influence children’s bedtimes by keeping children awake at later times, especially during weekends when social/family-related time such as dining outside in the late evening are practiced culturally.⁴⁸ Additional studies are needed to understand the cultural or parental factors that potentially contribute to the higher-than-expected proportion of evening-types in our preschool sample.

Prior studies of chronotype and time-of-day on cognitive performance have mainly investigated morning- and evening-types but not intermediate-types.²⁶ Moreover, time-of-day was not defined consistently across studies and were mostly based on school session times. Given that majority of younger children were of intermediate-types (59.9% in our study),²³ it is also important to examine chronotype and time-of-day effects in this group to gain a comprehensive insight into interindividual differences in cognitive performance. Further, performance should be measured specific to age and not categorized broadly as young children since there exists age-related progression in cognitive ability levels on frontal lobe tasks.⁴⁹ For example, it has been shown that 4-year-olds performed worse than 5- to 7-year-olds on working memory measures, while 8-year-olds outperformed younger children in their ability to solve complex problems.⁴⁹ Our findings also add to the literature by including late afternoon (4:00 pm to 6:30 pm) time-of-day. Unlike kindergarten or primary school children, preschoolers are yet to be exposed to the formal education system and fixed school start times. Thus, our study highlights the importance of considering chronotype and time-of-day in early childhood education.

This study also has its limitations. Although we have controlled for ethnicity in the main analyses, we acknowledge that there may be other residual confounding variables (eg, alertness level, performance IQ) that may also affect working memory.^13,43,50 Controlling for ethnicity is important given that our study sample comprises of multiethnic Asian (Singaporean) preschoolers. Given the influences of digital technology and parental involvement, sleep patterns in young children are often affected by their daily behaviors, such as screen time. Research has shown that greater daily screen time is linked to lower sleep duration in children.⁵¹ Hence, it may be difficult to affix a particular chronotype to a child without accounting for type and duration of daily activities that can influence sleep patterns throughout the development stages. Those children who attended the center visit were noted for difficulties in task understanding and execution, and such cases have been excluded prior to data analysis. Indeed, our sample was of typically developing children with no known diagnosis of learning difficulties. The comparatively small sample size of morning-types (n = 37) may not be able to detect differences in performance between time-of-day due to the lack of power. Although the mean effect size was significant when comparing differences in performance between morning- and evening-types in the late afternoon, the small sample size (n = 13) of morning-types indicated that this finding should be treated with caution. If, however, a larger and more comparable sample size is available for morning-types, we would still expect evening-types to perform better than morning-types in the late afternoon albeit with larger effect size. Perhaps this time, we may notice a synchrony effect between chronotype and time-of-day in SWM performance as shown by the graphical trend toward interaction in our results (Figure 3B). Despite the possibility of chance findings, our results in which evening-types outperformed morning-types in a working memory task are highly consistent with previous studies.^19,28

In conclusion, our findings suggest that late afternoon time of the day may differentiate SWM performance between evening-type compared to morning-type preschool children. Future research in this area will offer further insights into understanding optimal learning opportunities in early childhood education.

ABBREVIATIONS

BS

between search

CCTQ

Children’s Chronotype Questionnaire

SD

standard deviation

SWM

spatial working memory

DISCLOSURE STATEMENT

All authors have seen and approved the final manuscript. Work for this study was performed at the Singapore Institute for Clinical Sciences, Agency for Science, Technology and Research (A*STAR) and supported by the Singapore National Research Foundation under its Translational and Clinical Research (TCR) Flagship Programme and administered by the Singapore Ministry of Health’s National Medical Research Council (NMRC), Singapore-NMRC/TCR/004-NUS/2008; NMRC/TCR/012-NUHS/2014. Additional funding is provided by the Singapore Institute for Clinical Sciences, A*STAR, Singapore. Dr. Yap-Seng Chong received lecture fees from companies that sell nutritional products. They are part of an academic consortium that has received research funding from Abbott Nutrition, Nestec, and Danone. The other authors report no conflicts of interest.