Examining the Interface of Children’s Sleep,Executive Functioning,and Caregiving Relationships

Early Caregiving Relationships and Child Executive Development

It is increasingly believed that early relational experiences should have an impact on brain development. This impact would be attributable to the brain’s remarkable plasticity during the first years of life, characterized by overproduction of synapses followed by a period of gradual pruning (Huttenlocher, 2002). During this period, experience is considered to determine to a large degree which synaptic connections persist and are strengthened by frequent use, and which are progressively eliminated as a result of inactivity (Kolb et al., 2012). In this context, one aspect of child development that appears susceptible to strong environmental influences is executive functioning (EF), a set of higher-order cognitive processes that primarily serve the self-regulation of behavior and emotion. EF is largely subsumed by brain structures located in the prefrontal cortex, the brain area that shows the most protracted postnatal development. Many have argued that this largely postnatal development leaves a substantial window of opportunity for environmental input to impact the development of frontal brain systems and related executive functions (e.g., Kolb et al., 2012).

As one of the earliest, most intense, and most enduring experiences of childhood, the parent-child relationship is perhaps the most salient aspect of young children’s environment. We therefore set out to investigate whether normative variation in the quality of early caregiving relationships was related to child executive development. A first study (Bernier, Carlson, & Whipple, 2010) suggested that the quality of maternal behavior during mother-infant interaction, assessed when children were 12 and 15 months old, was associated with their performance at 18 months and 2 years on tasks mostly requiring working memory and set shifting, a dimension often called conflict EF (Carlson, White, & Davis-Unger, 2014). In contrast, no convincing links were found between maternal behavior and the second EF dimension considered, impulse control. When the same children reached 3 years of age, we sought to expand on these findings by adding assessments of father-child interactions and mother-child attachment security (Bernier, Carlson, Deschênes, & Matte-Gagné, 2012). Several links were found among maternal behavior, paternal behavior, attachment, and children’s performance on conflict-EF tasks. The most robust links pertained to attachment security, which was found to predict conflict EF at 3 years of age above all other factors considered.

A disappointing aspect of these papers was our difficulty explaining the impulse-control dimension of child EF, which showed very few convincing relations to caregiving measures—in contrast to other studies that did find links between similar constructs, such as effortful control, and different aspects of parenting (e.g., Taylor, Eisenberg, Spinrad, & Widaman, 2013). One possible explanation is that moderating effects could have been at play in our sample. In searching for theoretically sound moderators, we were inspired by previous research. Notably, it is often reported that parenting bears more importance for children living in disadvantaged environments (e.g., Schonberg & Shaw, 2007). One proposed explanation for this is that in relatively less advantaged families, there are often fewer alternative resources around the child besides the parents, which places more responsibility on parents’ shoulders, whereas children in families of higher socioeconomic status (SES) may benefit from a greater diversity of influences (Beyers, Bates, Pettit, & Dodge, 2003). This suggested that caregiving measures might be related to child impulse control to a greater degree among the relatively lower-SES families of our sample.

Addressing this question with respect to impulse control at 3 years of age, we found that, indeed, some of the nonsignificant links with maternal behavior assessed when children were 1 year old were attributable to interaction effects: The hypothesized links were significant only among children living in lower-SES homes. Specifically, the impulse-control performance of lower-SES children caught up with that of their more advantaged peers when they were exposed to high levels of maternal sensitivity, evidencing a protective effect of maternal sensitivity (Rochette & Bernier, in press). In a second report tackling this question, although this time inspired by differential susceptibility theory (Belsky et al., 2007), we investigated whether the mothering-EF links were more pronounced among temperamentally difficult children (Rochette & Bernier, 2014). In this case as well, interaction effects were found, indicating that maternal behavior did clearly relate to child impulse control, albeit only among children with more difficult temperaments.

In sum, these reports suggest that in this sample, (a) higher-quality parenting and more secure mother-child attachment relate to better child conflict EF, and (b) higher-quality parenting also relates to better child impulse control, but only among more vulnerable children, namely those living in lower-SES homes and those with more difficult temperaments.

Unexplored until a few years ago, the links between parenting and child EF are now increasingly being documented (Hughes, Roman, & Ensor, 2014). Explanatory mechanisms for these associations are also being identified—for instance, child verbal ability (Matte-Gagné & Bernier, 2011) and psychophysiological regulation (Blair et al., 2011). Nearly all current findings, however, pertain to mothers. Given evidence that father-child relationships are associated with many other aspects of child functioning, the investigation of their role in child executive development is an important next step. Another area that, like EF, is highly dependent on neural maturation but also thought to be affected by maternal and paternal factors is the sleep-wake cycle.

Early Caregiving Relationships and Child Sleep

A central developmental task of infancy and early childhood is the establishment of the sleep-wake cycle. A key aspect of this development is consolidation of sleep into the night period, such that infants increasingly get most of their sleep during the night. Furthermore, total sleep duration decreases sharply over the second year of life, mainly because of a decrease in daytime sleep. Finally, infants and toddlers gradually benefit from more efficient sleep, characterized by longer periods of uninterrupted sleep (see Iglowstein, Jenni, Molinari, & Largo, 2003). For simplicity, we use here the term sleep maturation to refer to these developments overall.

It has been proposed that while constitutional factors can in part explain individual differences in sleep maturation, family factors should play an increasingly central role as children develop (e.g., Ferber, 1996). Using our longitudinal study as a platform to investigate these links to family factors, we have found evidence for both maternal and paternal contributions. Hence, maternal and paternal psychological adjustment (Bernier, Bélanger, Bordeleau, & Carrier, 2013), as well as the quality of maternal and paternal behavior during parent-child interactions (Bordeleau, Bernier, & Carrier, 2012a), were found to predict more advanced sleep maturation among toddlers and preschoolers.

These links raise the possibility that secure parent-child attachment could favor child sleep. Although relations between child attachment and sleep have been observed in a few studies (see Simard, Bernier, Bélanger, & Carrier, 2013, for a review), one concern is that these findings, as well as our own findings just described, were based on maternal reports of child sleep, which are subject to reporting biases (Sadeh, 2011). Objective assessment of children’s sleep can be achieved with different instruments, notably actigraphy, a method that relies on motor activity to estimate sleep duration (e.g., total time asleep) and quality (e.g., frequency of night awakenings). Given meta-analytic findings suggesting that sleep duration and quality show distinct relations to developmental outcomes (Astill, Van der Heijden, Van IJzendoorn, & Van Someren, 2012), these two parameters are considered to be distinct, important aspects of children’s sleep. Using an actigraphy scoring algorithm that we validated with young children (Bélanger, Bernier, Paquet, Simard, & Carrier, 2013), we found that preschoolers’ sleep duration and sleep quality were unrelated to mother-child attachment security as assessed with the Strange Situation (Simard et al., 2013). However, children showing more resistant attachment behavior were perceived by their mothers as having more fragmented sleep at night. Of note in that study, though, was the uneven distribution of attachment patterns, with about 75% of dyads classified as having a secure relationship. Therefore, we re-investigated this question with another measure, the Attachment Q-Sort, which yields a continuous score of attachment security. In this case, we found the expected links: Children more securely attached to their mothers at 15 months showed sleep of higher quality and duration when they reached 2 years of age (Bélanger, Bernier, Simard, Bordeleau, & Carrier, in press).

Hence, our work joins that of other teams (see El-Sheikh, 2011) in suggesting that young children’s sleep, while undoubtedly dependent on neural maturation, may be impacted by different aspects of the family environment, such as parental psychological adjustment, parenting, and mother-child attachment security. Given that these family factors relate to child EF as well, the question naturally came up as to whether child sleep and EF might be interrelated.

Child Sleep and Executive Functioning

Links between child sleep and EF can be expected, owing not only to a putative common underlying influence of family factors but also to the direct influence of sleep on EF, as demonstrated by experimental research with both adults and school-aged children (see Astill et al., 2012, for a review of underlying neurobiological mechanisms). Until recently, however, it was still unknown whether the links between sleep and EF could be observed before school age. Addressing this gap, we first reported that more mature sleep at 1 year of age was related to better EF at 18 months and 2 years of age (Bernier, Carlson, Bordeleau, & Carrier, 2010). Following up on the same group of children when they reached 4 years of age, we found that infants showing more mature sleep at age 1 continued to show superior EF performance 3 years later (Bernier, Beauchamp, Bouvette-Turcot, Carlson, & Carrier, 2013). These findings were functionally and developmentally specific, in that the effect of sleep held above that of general cognition and prior EF.

Overall, we and others have been finding that both sleep and caregiving experiences relate to child executive performance, but also that sleep and caregiving are interrelated. Thus, although the associations discussed thus far are bivariate and unidirectional, the real processes linking these areas of functioning are almost certainly more complex. These processes could and probably do take many forms, but one form that we have begun to address is that of interactive effects, in which children’s sleep modulates their receptivity to caregiving influences.

Beyond Bivariate Links: Interactive and Reciprocal Relations

It is increasingly being documented that environmental influences on child functioning can be amplified or attenuated by certain biological factors (Belsky et al., 2007). El-Sheikh, Hinnant, Kelly, and Erath (2010) proposed that one such factor might be sleep. Indeed, the restorative properties of sleep may facilitate children’s capacity to take advantage of positive caregiving influences, whereas poor sleep may interfere with children’s receptivity to environmental input by making them fatigued or irritable. We first sought to examine this idea with respect to the well-documented association between maternal sensitivity and children’s socio-emotional adjustment. Results indicated that maternal sensitivity interacted with mother-reported infant sleep duration, both assessed when children were 1 year old, such that the protective function of sensitivity against subsequent internalizing and externalizing symptoms, assessed at 4 years of age, was found only for children who slept more at night (Bordeleau, Bernier, & Carrier, 2012b). We then used actigraphy to examine the moderating role of child sleep in the connections between maternal sensitivity and three other of its well-documented outcomes: child attachment security, theory of mind, and EF. In line with the previous findings, this study indicated that maternal sensitivity related in expected ways to these outcomes only among children showing more mature sleep (Bernier, Bélanger, Tarabulsy, Simard, & Carrier, 2014). Taken together, these two studies suggest that adequate child sleep may facilitate the unfolding of the normal developmental processes linking parenting to child outcomes.

In addition to such interactive effects, transactional links involving reciprocal influences appear likely. Sophisticated longitudinal studies are beginning to emerge that provide unprecedentedly robust tests of developmental processes—for instance, by relying on several waves of data for each construct, controlling for autoregressive effects and within-time relations, or focusing on changes in predictors and outcomes (see Taylor et al., 2013, for examples pertaining to parenting and child effortful control and Kelly & El-Sheikh, 2014, for examples pertaining to sleep and socio-emotional adjustment). As more studies of that level of methodological and statistical refinement continue to appear, the field will make important strides toward estimating the real, population-level magnitude and direction of associations between key developmental constructs.

Looking Ahead: The Holy Grail of Causality

Nearly all studies that we have described above, whether ours or others’, have been based on nonexperimental designs. Consequently, the relations observed may not be indicative of causal processes. For instance, the links that we have reported between child sleep and EF, which we interpreted as implying an influence of sleep on EF, could just as well suggest that both are indices of neural maturation at different ages. Alternatively, these associations could be due to a common influence of the family environment. Furthermore, connections to parenting may not be causal in nature either, and could reflect shared genetic characteristics between parents and children. In fact, given experimental studies showing that child sleep (Mindell, Kuhn, Lewin, Meltzer, & Sadeh, 2006) and EF (Diamond & Lee, 2011) can be improved by behavioral intervention, and genetically informative designs suggesting that they are both partly heritable (Polderman et al., 2007; Touchette et al., 2013), gene-environment interactions are almost certainly at play.

In order to demonstrate causal relations between caregiving and child sleep or EF, and to isolate the putative role of brain development in these associations, the field will need to move toward experimental designs. Although ambitious, this goal appears realistic in light of recent studies showing that evidence-based intervention can be used to enhance the quality of parent-child interactions and documenting resulting improvements in child functioning (e.g., Kochanska, Kim, Boldt, & Nordling, 2013). In fact, a groundbreaking study showed that an intervention aimed at enhancing sensitivity among mothers of preterm babies improved infant cerebral white-matter development (Milgrom et al., 2010). A most promising avenue would be to use this type of evidence-based parenting intervention to experimentally enhance the quality of care children receive. One can then assess ensuing improvements in child outcomes, and estimate the (most likely partial) mediating role of changes in brain structure or function in these putative improvements. Given intriguing evidence suggesting that some aspects of children’s genetic makeup make them more (or less) likely to benefit from parenting interventions (Bakermans-Kranenburg, Van IJzendoorn, Mesman, Alink, & Juffer, 2008), the combination of genetically informative designs and intervention studies appears especially exciting.

In sum, joining forces with colleagues working in other disciplines appears crucial for pushing the developmental field forward. Combining the elaborate longitudinal designs and careful observational measures that are the hallmark of developmental psychology with genotyping and brain-imaging techniques, along with parenting interventions, will provide unprecedented answers to the exciting but complex questions pertaining to the interface of children’s relationships, biology, and cognition. In all likelihood, we have hardly scratched the surface.