Mother–infant interactions and regional brain volumes in infancy: an MRI study

Parent–infant interactions are critical for child development. For instance, sensitive and responsive early care is linked to optimal behavioural and cognitive outcomes (Cabrera et al. 2011; Lugo-Gil and Tamis-LeMonda 2008); in contrast, parental insensitivity increases the risk of children developing psychopathology in later life (Murray et al. 2010). Although the biological mechanisms mediating these associations are not entirely understood, it is generally agreed that the human brain is most vulnerable to environmental influences (De Bellis et al. 2001; Schore 2001)—including parent–infant interactions (Rifkin-Graboi, et al. 2015)—early in development.

For example, elevated levels of stress hormones stemming from early-life adversity are thought to lead to altered brain development through the accelerated loss of neurons, disrupted pruning, inhibition of neurogenesis (Teicher et al. 2006; Tupler and De Bellis 2006), and perhaps also altered anatomical ‘connectivity’ (Sarkar et al. 2014). Prior reports also suggest that early childhood maltreatment is associated with later fronto-limbic abnormalities (Belsky and de Haan 2011; Hart and Rubia 2012); smaller corpus callosum and total brain volumes, and increased ventricular volumes (De Bellis et al. 2002; Teicher et al. 2004). It has also been suggested that the male brain is particularly vulnerable to such insults (De Bellis and Keshavan 2003; Tupler and De Bellis 2006). For example, smaller cerebral volumes have been reported in older male children exposed to childhood maltreatment (Belsky and de Haan 2011).

However, existing studies in humans mainly document outcomes following extreme adversity in infancy (i.e., institutional rearing or severe maltreatment), and are retrospective in design. Also, the high prevalence of psychopathology (72%) in these retrospective analyses of older cohorts (De Bellis et al. 2001) makes it difficult to determine whether the structural brain differences observed explain the aetiology of psychopathology or are caused by it and/or its treatment [for example, medication exposure may confound interpretation (Tupler and De Bellis 2006)]. Thus, the results of these high-risk samples do not reveal how normative variations in early parent–child interactions influence child brain structure in the early postnatal period.

This is an important omission, considering the compelling evidence that an early sensitive caregiving environment likely provides an optimal emotional context for children’s early brain maturation and subsequent cognitive abilities (Bernier et al. 2010). Furthermore, the postnatal period is characterized by rapid brain development. Specifically, the first year of life is the period of greatest brain volume growth in typical children—total brain volume at 2–4 weeks of age is approximately 36% of adult volume, and by 1 year it is approximately 72% of adult volume (Knickmeyer et al. 2008). Brain plasticity during this period makes the infant brain particularly sensitive to environmental influence, especially the social-affective environment (Schore 2001). Variations in maternal care are thought to help shape neural structures and circuits, and subsequently psychological outcomes (Roth and Sweatt 2011); and there is reasonable consensus that maternal sensitivity in the first year of life has a key impact on development (de Wolff and van Ijzendoorn 1997). Defined as the timely and accurate response to the infant’s communicative cues, maternal sensitivity predicts positive social relationships and enhanced cognitive abilities in the infant (Wade et al. 2015); and sensitive caregiving during the first year is critical for the maturation of the infant’s stress response system (Gunnar and Cheatham 2003; Hane and Fox 2006).

Therefore, the relationship between normative variations in parenting and brain structure in children has now started to be examined (Kok et al. 2015; Moutsiana et al. 2015; Rao et al. 2010; Rifkin-Graboi, et al. 2015; Whittle et al. 2014). For example, higher levels of parental sensitivity in early childhood have been linked with larger total brain and grey matter volumes in children at 8 years of age (Kok, et al. 2015). In another study, insecure attachment at 18 months was associated with greater amygdala volumes at 22 years (Moutsiana et al. 2015). In contrast, a study of structural MRI data from twenty 6-month-old infants, has demonstrated a link between maternal sensitivity and hippocampus volume (Rifkin-Graboi et al. 2015)—specifically, reduced maternal sensitivity was associated with larger volumes. While these studies are important first steps, some had a lengthy period between caregiving measures and brain MRI acquisition (Kok et al. 2015; Moutsiana et al. 2015), and others used adolescent samples (Whittle et al. 2014). In infancy and childhood, the changes in brain volume over time occur in parallel to maturation of cognitive, motor and socio-emotional processes (Shulman 2016; van Soelen et al. 2012). By examining the brain and the factors that influence it at the same time, we can begin to identify possible causes of altered brain growth and behaviour, as well as potential treatment targets and biomarkers that are predictive of outcomes.

In the current study, we used magnetic resonance imaging (MRI) to investigate whether mother–infant interactions observed in a community sample of mothers and their 3- to 6-month-old infants, are related to variations in regional brain volumes. In addition to studying an association between maternal behaviours and infant brain volume, it is important to know whether—or not—infant behaviours are related to brain volumes as this may help us understand what brain systems drive infant behaviour and/or respond to infant behaviour changes. Such information may eventually help us develop objective predictive tools to identify infants who might benefit from early intervention to improve outcomes. Moreover, since there is a bidirectional link between maternal sensitivity and infant behaviours (Beebe et al. 2016; Feldman 2007; MacLean et al. 2014), it is also possible that infant behaviours relate to brain development indices. Therefore, we predicted that there would be a relationship between both maternal and infant behaviours and regional brain volumes. Owing to limited prior information in infancy, with both larger and smaller regional brain volumes reported in relation to early caregiving, and no previous evidence in relation to infant behaviours, the direction of this relationship was not a priori predicted.

Finally, where an association between brain regions and mother–infant interactions was observed, we conducted an exploratory examination of potential sex differences in these relationships. As preclinical and clinical studies of adverse rearing conditions (i.e., exposure to childhood maltreatment) indicate that the male brain is influenced more by the early environment (Belsky and de Haan 2011; Glaser 2000), we predicted that any relationship between maternal and/or infant behaviour and brain would be stronger in males.

Table 2 shows the means, standard deviations, and sex differences for mother–infant interaction dimensions and brain volumes. Significant sex differences were found in the raw measures of the subcortical grey and intracranial volumes; in both instances, male infants had larger volumes than females—subcortical grey: (males: mean = 36.17, SD = 4.16; females: mean = 33.05, SD = 2.36; t (37) = 2.89, p = 0.007), and intracranial volume (males: mean = 888.56, SD = 88.54; females: mean = 825.20, SD = 102.68; t (37) = 2.07, p = 0.046).

In this cross-sectional exploratory study, we show that variations in typical mother–infant interactions are associated with differences in infant brain volumes. Specifically, we found that lower maternal sensitivity was correlated with smaller subcortical grey matter volumes in both sexes. In contrast, male infants with higher levels of communication during early interactions had smaller cerebellar volumes.

Prior studies of extreme neglect, leading to paediatric post-traumatic stress disorder, have reported that childhood maltreatment is associated with smaller total grey and white matter volumes, and larger frontal lobe CSF volumes, especially in males (De Bellis and Keshavan 2003; De Bellis et al. 2002). A more recent investigation of normal variations in parental care and brain structure (at 8 years of age) has revealed a similar relationship between early childhood parental sensitivity and total brain and grey matter volumes (Kok et al. 2015). More specifically, and when compared to other brain regions, the subcortical grey matter appears to be particularly ‘responsive’ to early environmental influences. For example, the basal ganglia and thalami (which comprise the subcortical grey) are very sensitive to hypoxic events in utero (Okereafor et al. 2008; du Plessis and Volpe 2002; Shalak and Perlman 2004); and infants so exposed, tend to have poor neurodevelopmental outcomes.

Our work extends these findings to show that a relationship between maternal sensitivity and infant brain development is present from as early as 3 months. However, these findings are correlational and do not necessarily indicate a causative link between early care and infant brain structure. Also, we cannot say firmly whether this relationship has ‘positive’ or ‘negative’ developmental implications. Neither can we be certain whether smaller regional brain volumes are a consequence of poorer parenting quality, or whether infants with smaller regional brain volumes influence their mothers’ interactions. It is also possible that since infant and mother are closely genetically related, the associations observed could be mediated through shared genetic variants, including an inherited brain volume and behavioural style.

We do suggest, however, that the infant stress response system, which undergoes rapid development in the first year of life, is likely to be involved. For example, in the early postnatal period when the hypothalamic–pituitary–adrenal (HPA) axis of infants is labile, sensitive parenting is associated with either smaller increases or less prolonged activations of the infant HPA axis, when subjected to mild stress (Albers et al. 2008). Therefore, exposure to negative (for example, insensitive or intrusive) parental behaviours may constitute a source of stress for the infant, and activate the infant’s adrenocortical axis (Atkinson et al. 2013). The subsequent elevation in cortisol may influence brain volume and ‘connectivity’ in the growing child (Sarkar et al. 2014). Furthermore, mothers who are more sensitive in the postnatal period have been reported to demonstrate secure mental representations of attachment during pregnancy, which in turn may impact upon the HPA axis and the intra-uterine environment (Kinsella and Monk 2009). Hence, associations between maternal behaviours and infant brain volume may have their origins even earlier in development, but future studies including objective measures of the HPA axis and a comprehensive characterization of maternal psychopathology during pregnancy are needed to better understand the mechanisms involved. In addition, as maternal sensitivity is thought to be a stable trait over time (Feldman 2010), follow-up of these dyads would help to determine whether the relationship we observed between maternal sensitivity and the infant brain persists or shifts as children grow.

An important aspect of our study design was that it also permitted examination of a possible link between infant behaviour (for example, communication and fretfulness) and brain volume. We found that a smaller cerebellum volume is associated with better infant communication, and this relationship was particularly evident in male infants. Again, we emphasize that the causal direction of this relationship is not known—i.e., does cerebellar development drive communication, or vice versa? Regardless, a link between cerebellar development and communication is not surprising given its key role in emotion processing and executive functioning (Schmahmann et al. 2007). For example, the cerebellum has been proposed to have a key role in the temporal processing of events and in allocating attentional resources in ‘real-time’ to guide or prepare behaviour (Schwartze and Kotz 2016). In addition, the cerebellum responds to auditory stimulation including spoken language (Buckner 2013). Together, these attributes likely make a key contribution to organizing effective communication during face-to-face interaction, and our data suggest that the link between cerebellum and communication is present from early infancy. Furthermore, that our results reveal a relationship primarily in males might also have been expected, as the developmental trajectory of the cerebellum is sexually dimorphic. The male cerebellum develops more slowly than the female (Tiemeier et al. 2010), potentially making the former more vulnerable to early adverse environments. Consistent with this, cerebellar pathology is a hallmark of neurodevelopmental disorders, such as ASD, which also shows marked sex differences (Wang et al. 2014). Finally, our sample size and current study design precludes an in-depth analysis of potential processes which might explain the link between infant behaviour and brain volume; including, for example, the role of maternal sensitivity which could be considered in future research.

Our study has a number of limitations. First, as noted above, our results are correlational and causality cannot be inferred. Second, although in line with the current literature (Rifkin-Graboi et al. 2015), our sample size was modest and replication in larger samples will be necessary. Third, the infants in our study were mainly from white European ‘middle-class’ families, educated to degree level, and therefore, we cannot be certain that these results generalize to families of different ethnicities and educational backgrounds. Fourth, we did not define a priori regions of interest since we do not yet have extensive knowledge of all brain areas affected by normative variations of parenting in infancy. Fifth, our primary goal in this initial study was to establish if there were brain regions linked to mother-infant interactions across the group, and then, having done that, to explore if there were sex differences in those specific regions. This approach helped us avoid type 1 error when running multiple tests. However, it risked generating type 2 errors of incorrectly retaining a false-negative finding. Therefore, we cannot exclude the possibility that there are sex differences in regions without main effects; and in our ongoing studies we are recruiting much larger cohorts in order to look at each sex separately across multiple brain regions. We hope this will provide adequate power to explore regional associations with sex and parent–child interactions in detail. Finally, there were also technical constraints to our study. The scanning of very young infants is challenging and the structural sequences used were of relatively low resolution. Hence, our overall volumetric measurement may miss the fine-grained structural differences that might be detectable in larger samples, or through higher resolution scanning protocols. Furthermore, and in line with previous studies of this age range (Hazlett et al. 2012), another limitation was the inability to differentiate between grey and white matter volume, due to ongoing myelination in these young infants.

Nonetheless, the current analyses provide a Proof of Principle that early mother–infant interactions are associated with variations in infant brain development. If correct, our finding that early sensitivity (a modifiable factor) is linked to the development of brain regions (known to impact upon emotional and cognitive development), opens up the potential to influence infant developmental trajectories.