Introduction
Parental depression is a robust risk factor for the emergence of depression in children. Offspring of depressed mothers are about 3 times more likely to develop major depressive disorder (MDD) than children who experienced no maternal depression (Weissman et al.,
2016). Information-processing biases may interact with genetic and environmental factors to potentiate the emergence of depression in the offspring (Gotlib et al.,
2020). The existing literature has underscored the association between exposure to maternal depression and attention bias (AB), characterized as preferential attention toward, and/or difficulty disengaging attention from, emotional stimuli (e.g., sad faces) in the offspring (Joormann,
2009). However, the existing literature leaves some important questions unanswered. Most of the studies focused on middle childhood to adolescence, when early symptoms of depression may already be present (Avenevoli et al.,
2008; Kessler et al.,
2005). Hence, the findings cannot speak to whether AB is present in early childhood before the onset of clinical-level depressive symptoms. Further, the possible role of maternal attentional behavior in modeling or directing the development of AB in children remains unclear. The present study aims to investigate (1) whether young children of mothers with a history of MDD exhibit AB to negative emotional faces compared to offspring of never depressed mothers, and (2) whether mothers with a history of MDD also show AB to negative emotional faces relative to never-depressed mothers. Additionally, we will also explore whether there is an association between mothers and their children’s attention patterns. Understanding the attention patterns associated with depression risk is particularly important for identifying potential targets for early preventative interventions (Browning et al.,
2012).
AB to depression-relevant stimuli might be an early-emerging risk factor for depression. Cognitive theories and empirical evidence in adults suggest that depression is associated with increased attention allocation to dysphoric or sad stimuli and reduced attention allocation to happy information (Armstrong & Olatunji,
2012; Suslow et al.,
2020). AB toward sad faces prospectively predicted depressive symptoms (Beevers et al.,
2011), and training attention away from dysphoric images reduced depression (Wells & Beevers,
2010). Hence, AB potentially plays a causal role in the development of depression. Extending from adult literature, maternal depression may increase psychiatric risk in offspring through shaping offspring’s attention toward emotional faces. For example, 8-month-olds exposed to prenatal or postnatal maternal depression showed greater difficulty in disengaging from fearful faces than infants of mothers with low depression levels (Kataja et al.,
2020). Hence, attentional processing of emotional faces is susceptible to the influence of maternal symptoms from early infancy (Porto et al.,
2020; Sandre et al.,
2022).
The extant studies in youths with depression diagnosis or high-risk (HR) offspring of depressed mothers have used dot-probe, passive viewing, and visual search tasks to measure AB to emotional faces. The dot-probe paradigm (MacLeod et al.,
1986) displays an emotional (e.g., sad) and a neutral face side by side, followed by a probe that replaces either the emotional (congruent trial) or neutral stimulus (incongruent trial). Faster responses on the congruent trials indicate an AB toward the emotional faces, whereas a faster response on the incongruent trials indicates AB away from the emotional stimuli. The passive viewing task (e.g., Harrison & Gibb,
2015) presents participants with a set of emotional faces simultaneously (e.g., angry, sad, happy, neutral) over a relatively long (e.g., 20s) free-viewing trial. The key dependent variables are the proportion of times engaged in looking at each emotion type over the total looking time. Lastly, The visual search task (Donnelly et al.,
2010; LoBue,
2009; LoBue & Pérez-Edgar,
2014) is designed to evaluate both facilitated detection and difficulty in disengagement from a specific emotional type (Armstrong & Olatunji,
2012). The task presents the participants with a matrix of emotional faces (e.g., 3 × 3 matrix) in each trial. Each matrix contains a single target face with one type of emotional expression (e.g., an angry face) among distracters from another emotion category (e.g., happy faces). Participants are asked to find the target as quickly as they can. For example, faster identification of the angry target among happy distracters indicates facilitated detection of the angry faces, whereas slower identification of the happy target among angry distracters indicates difficulty in disengaging from the angry faces.
Studies that measured participants manual reaction times (RTs) indicated that HR offspring (5–14 years) of depressed mothers displayed AB specifically towards sad faces (e.g., Joormann et al.,
2007; Kujawa et al.,
2011) in dot-probe tasks, comparable to youth with depression diagnoses (Hankin et al.,
2010). Thus, AB towards symptom-related emotional stimuli (e.g., sad faces) is observable in young children, and it is related to risk for depression prior to clinical diagnoses (Lau & Waters,
2017). However, AB can manifest in several components that operate on a time sequence (Cisler & Koster,
2010). RT measures cannot directly capture attention processes with sufficient temporal sensitivity (Yiend,
2010). Moreover, RT measures might not be able to reliably capture the core mechanism underlying AB, given that individual differences in multiple processes may affect RT outcomes (Mogg et al.,
2008).
Compared to manual reaction times (RTs), eye tracking (ET) provides a continuous and temporally sensitive measure of eye movements and is well suited to depict the time course and components of AB (Fu & Perez-Edgar,
2019). In a passive viewing task, HR children (8 to 14 years) with depressed mothers showed greater attention maintenance specifically to sad faces, indexed by a greater proportion of time spent fixating on the sad face, compared to children of never-depressed mothers (Owens et al.,
2016). Furthermore, HR children’s attention maintenance toward sad faces in an ET dot-probe task increased longitudinally from 8 to 14 years (Gibb et al.,
2023). Together, these findings suggest that HR children exhibit increased attention maintenance toward depression-relevant stimuli, an AB pattern comparable with individuals with depression (Mogg & Bradley,
2005). However, existing ET studies that used the visual search task with HR (Platt et al.,
2022) and depressed youth (9 to 14 years; Buhl et al.,
2021) did not find depression-related group difference in attention disengagement from sad or angry relative to happy distracters.
While ET enables direct assessment of the microstructure of the AB mechanism compared to RT measures, there is a limited understanding of the association between familial risk of depression and attention detection of symptom-relevant emotional stimuli. Behavioral evidence suggests that adulthood depression was associated with a longer time to detection of happy targets (Suslow et al.,
2001). Moreover, sensitivity in detecting sad faces may differentiate between a subset of offspring (8 to 14 years) of depressed mothers and those of never-depressed mothers (Burkhouse et al.,
2016). However, there is a lack of ET evidence on whether HR children show altered detection of negative and/or positive emotional stimuli. Additionally, the existing RT and ET studies of AB implicated in depression risk have largely been conducted in middle childhood and adolescence. There is a need to study AB patterns in younger HR children before the risk mechanisms are confounded by pubertal changes and the development of compensatory mechanisms.
The emergence of AB to depression-relevant stimuli in HR offspring may be explained by the continuous exposure to maternal depression from early life. Maternal depression provides developmental experiences that fine-tune children’s attention orienting to environmental stimuli related to mothers’ depressive state (e.g., mothers’ sad faces). This may gradually heighten children’s perceptual and neurological sensitivity to detect sad facial expressions and depression-related information in the environment that might not be as salient to others (Smith & Pollak,
2020). The AB may initially be adaptive for the offspring to be affectively attuned to their mothers. However, it can become maladaptive when the AB to depression-relevant stimuli becomes more canalized through development and is applied to a broader range of socioemotional contexts (Burkhouse et al.,
2016; Leppanen & Nelson,
2009; Pollak,
2003).
Parent-child association or concordance of AB is another possible mechanism underlying emotion modeling and family aggregation of depression and anxiety (Aktar et al.,
2019; Creswell et al.,
2010). The genetic and shared environmental factors may contribute to the parent-child concordance of AB (Colich et al.,
2017; Ethridge et al.,
2022). Parental depressive symptoms and parental AB may sensitize their child to negative stimuli in the environment, inducing AB in the child (Field & Lester,
2010). Supporting evidence in normative samples indicates that both parents and their infants displayed AB towards fearful versus happy faces, measured using ET (Aktar et al.,
2021). However, existing studies (Aktar et al.,
2019; Mogg et al.,
2012; Waters et al.,
2015,
2018) that assessed AB using RT measures in children and their mothers with or without a lifetime diagnosis of anxiety and depressive disorders have predominately showed no significant concurrent or prospective associations between parent and child AB (Aktar,
2022 for review). Only one study showed that among HR children, child AB to angry faces was associated with reduced AB to happy faces in mothers when controlling for maternal symptoms (Waters et al.,
2015). Given the heterogeneity in maternal diagnoses, we do not yet know the mother-child association of AB as a function of maternal depression diagnostic status. Due to the limitations of manual RT measures of attention, the mother-child association of the specific component of AB also remains unclear.
The present study extends the existing literature by implementing ET to assess AB in mothers with and without a history of MDD and their 4-year-old offspring, an age precedes the onset of clinical-level anxiety and depression symptoms in the offspring. The visual search task was implemented to measure AB in the initial detection and maintenance of the emotional (e.g., sad and angry) target faces in the presence of competing emotional (e.g., happy) distracters. We also explored the association between mothers’ and their offspring’s AB patterns. Based on the extant literature, we hypothesized that HR children of depressed mothers will exhibit AB to sad faces, characterized by faster latency of initial fixation to sad than happy targets and longer dwell time on the sad than happy targets, and LR children with never-depressed mothers will not show AB to sad faces. We also hypothesized that mothers with a history of MDD will show AB to sad faces, comparable to their HR children. Given the mixed findings in the literature about the parent-child association of AB in mothers with depressive symptoms, we did not have a prediction on whether there will be a significant mother-child association of the ET measures, and whether maternal depression history and/or face emotion will moderate the association.
Discussion
The present study aimed to examine (1) whether children with familial risk of depression (HR) displayed different AB patterns relative to their LR peers, and (2) whether mothers with a history of MDD exhibit comparable AB patterns as their children. Additionally, we directly examined the mother-child association in the attention measures. We implemented an ET visual search task (Armstrong & Olatunji,
2012; Donnelly et al.,
2010; LoBue,
2009; LoBue & Pérez-Edgar,
2014) to examine AB in children at 4 years of age, a developmental period that precedes the onset of clinically significant levels of psychopathology but is marked by increased incidence of socioemotional problems (Avenevoli et al.,
2008; Egger & Angold,
2006).
Our primary findings revealed both normative AB patterns across all children as well as AB patterns specific to children at familial risk for MDD. In the visual search task, AB is manifested as faster detection (i.e., shorter latency) and/or longer maintenance (i.e., greater dwell time) of the negative emotional targets relative to happy targets. We found that children across both groups displayed AB to angry targets relative to happy targets indexed by both measures. Children from both groups exhibited longer attention maintenance on the sad relative to happy targets after the targets were located. However, there was a significant interaction effect, reflected by the findings that the difference in the predicted values between mean dwell time on the sad versus happy targets was greater in the HR than LR group. Furthermore, the HR children were faster in detecting the sad than happy targets, whereas the LR children did not show such pattern of sad bias. We did not find significant group-by-stimulus interaction effect on mothers’ ET measures. Although there was an indication that mothers with a history of MDD had longer attention maintenance to sad relative to happy targets, the non-significant interaction effect suggests that AB to sad faces in mothers with an MDD history is not significantly greater than in never-depressed mothers. Our supplementary results (see SI) revealed that there were no mother-child associations in AB to angry or sad targets. Maternal MDD history did not moderate the mother-child associations in AB patterns.
Our findings indicate potential normative AB towards threat-related (i.e., angry faces) in early development. Both LR and HR in the current study showed longer attention maintenance to sad relative to happy targets. Children across both groups also displayed angry bias. This is consistent with findings that HR and LR children did not differ in AB patterns (Gibb et al.,
2009; Waters et al.,
2015). AB to threat-related faces might be normative and early emerging in development (e.g., Leppanen & Nelson,
2009; LoBue & DeLoache,
2010; LoBue & Pérez-Edgar,
2014; Nakagawa & Sukigara,
2019). However, maladaptive AB may stem from the normative AB patterns for a subgroup of children who are at risk for internalizing problems (Morales et al.,
2016).
Indeed, children with familial risk of MDD also displayed a distinct pattern of sad bias compared to their LR peers. Cognitive theories based on adult research suggest that depression is associated with increased attention allocation to dysphoric or sad stimuli and reduced attention allocation to anhedonic or rewarding stimuli (Armstrong & Olatunji,
2012; Suslow et al.,
2020). Offspring of mothers with an MDD history displayed facilitated detection to the sad than happy targets, whereas this sad bias was not found in their LR peers. The findings were consistent with the existing evidence that risk for depression in childhood is linked to heightened sensitivity in detecting sad faces (Burkhouse et al.,
2016) or potentially blunted sensitivity in detecting socially rewarding information as shown in adults with depression (Suslow et al.,
2001). Moreover, HR children exhibited greater attention maintenance to sad than happy targets, and this sad bias pattern was greater in HR than LR children. However, different from existing evidence (Owens et al.,
2016), HR children did not show longer attention maintenance specifically on the sad targets compared to LR children. One possibility is that LR healthy controls also show sad AB that takes place in the later stage of information processing in tasks that require goal-directed attention orienting based on task demands (Buhl et al.,
2021; Platt et al.,
2022; Sylvester et al.,
2016). Thus, the difference in sad bias between LR and HR children might be diminished in these tasks compared to passive viewing tasks (Owens et al.,
2016).
The current findings do not support a significant mother-child association of AB. Mothers with an MDD history exhibited longer attention maintenance to sad than happy targets. Hence, there is a similarity of the sad AB pattern towards symptom-relevant stimuli between mothers with MDD history and their HR children. However, the sad bias in mothers with an MDD history was not significantly greater than in those never-depressed mothers. Furthermore, we did not find a positive relation between fixation latency and dwell time on the target faces. Additionally, our findings do not indicate that the mother-child association of AB differs depending on maternal MDD history. Our finding is consistent with Aktar et al. (
2021) who did not find a significant moderating effect of parental symptoms on the parent-infant association in AB to emotional faces. Hence, we cannot conclude that offspring who display AB to sad faces modeled their attention patterns from their mothers. It is possible that additional mechanisms contributed to the emergence of AB in the HR offspring.
Future studies are needed to examine parental affective expressions and parenting behavior in mothers with MDD history. Depressed mothers tended to provide less sensitive parenting (Murray et al.,
2010) and display less positive and more neutral or negative emotions (Aktar et al.,
2017). The early-life experience of symptom-relevant stimulus exemplars and caregiving behaviors may serve to canalize the offspring’s attention to the types of stimuli that they are most likely to encounter (Leppanen & Nelson,
2009; Pollak,
2003). Environmental exposure to mothers’ negative emotional expressions and parenting behavior may gradually heighten children’s perceptual and neurological sensitivity to detect sad faces and depression-related information in the environment, thus developing “experience-specific information-processing biases” over time (Burkhouse et al.,
2016; Leppanen & Nelson,
2009; Pollak,
2003). Additionally, the neural systems underlying attention to facial emotion are particularly malleable during the first several years of life (Blair & Raver,
2012; Fox et al.,
2010). A future direction is to examine whether parental affect and parenting behavior influence the link between maternal MDD and offspring’s AB patterns.
Limitations need to be considered when interpreting current findings. First, our analyses using multiple indices of AB (ET dwell time, ET latency, and manual RTs) in a relatively small sample size were exploratory. Our HR group contained a larger proportion of females than males, although there was no significant group difference in sex. Future studies with larger sample size can benefit from testing whether there are sex differences in the effect of maternal depression on the offspring’s AB patterns (e.g., Kujawa et al.,
2011). Moreover, there are between-subjects variations in the time interval between data collections of questionnaires and ET measures. This methodological issue complicates interpretations about the impacts of child internalizing symptoms (measured by CBCL) or maternal depressive levels (measured by Beck Depression Inventory described in SI) on ET measures of AB patterns.
Second, the ET measure of latency to initial target fixation had unsatisfactory reliability. Future studies can benefit from implementing multiple ET task paradigms to capture AB operating at different stages of information processing (Fu & Perez-Edgar,
2019). The rich set of ET measures obtained from multiple tasks can be used to characterize AB profiles associated with familial risk for depression (variable-centered analysis) or characterize sub-groups of HR children (person-centered analysis) (Vallorani et al.,
2021).
Third, the present study did not examine parenting styles or parent-child interaction dynamics directly. Evidence supports the association between child AB and negative parenting (Gulley et al.,
2014) or parent-child conflicts (Connell et al.,
2013). Future studies are needed to characterize the environmental factors that mediate the impact of maternal depression on the emergence of AB in the offspring.
Fourth, the current study cannot speak to the impacts of timing and chronicity of exposure to maternal depression on the development of AB. Postnatal exposure to maternal depression may disrupt the development of neural networks underlying attention control and emotion regulation that are plastic from the prenatal period to childhood (Loman & Gunnar,
2010). Further evidence is needed to assess whether offspring’s AB patterns and mother-child association of AB may vary with the age of maternal depression exposure.
Fifth, we do not provide longitudinal evidence that examines whether AB to sad faces predicts the emergence of socioemotional problems and psychopathology in children. We also did not examine the directionality of parent-child relations in AB or depression levels. While Waters et al. (
2018) did not find a cross-sectional relation between mothers’ and children’s AB, they revealed children’s AB at the first assessment was associated with their mothers’ AB at the second assessment, indicating a child-to-parent transmission of AB. Longitudinal evidence has also pointed to the bidirectional links between parent-adolescent depressive symptoms (Johnco et al.,
2021). Future longitudinal studies are needed to examine the complex bidirectional influences between AB and psychopathology in parent-child dyads (Aktar,
2022; Pérez-Edgar et al.,
2021).
Lastly, future investigations are needed to investigate separate contributions of mothers’ and fathers’ depression on children’s AB. It is possible that depression and anxiety in mothers may have a distinct impact on the offspring’s AB patterns (Aktar et al.,
2016,
2019). Studies that evaluate the impact of depression in both parents may be able to better account for the variability in the offspring’s AB.
In conclusion, the present study examined the impact of maternal MDD history on AB towards emotional faces in four-year-olds. We also examined AB patterns in mothers to investigate whether there is a concordance in AB patterns between mothers with a history of MDD and their offspring. We found both normative AB patterns across all children as well as AB patterns specifically associated with familial risk of depression. AB to threat-related (e.g., angry) faces might be normative and emerge early in development. Maternal MDD history was associated with children’s facilitated detection of sad relative to happy faces. The facilitated detection of sad faces was not found in children with never-depressed mothers. Children across both groups exhibited greater attention maintenance to sad than happy faces, although the sad bias indexed by dwell time was greater in children of depressed mothers than those with never-depressed mothers. The present study did not find mother-child association in AB patterns. Future research needs to directly examine the possibility that maternal MDD may also have created a psychosocial environment that gradually shaped their children’s attention towards symptom-congruent socioemotional information.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.