Affect and gaze responses during an Emotion-Evoking Task in infants at an increased likelihood for autism spectrum disorder

Emotional regulation (ER) begins to appear in the first year of life and refers to the ability to modulate the occurrence, intensity, and valence of emotional reactions through intrinsic (learned with experience) and extrinsic (with assistance from others) strategies [9,46,59, 26]. Depending on context, ER can be unconscious or conscious, controlled or automatic, and extrinsic (e.g., parent regulating child’s emotions) or intrinsic (child regulating own emotions) [27]. Emotional regulation is predictive of several domains of development in childhood, including behavioral problems (e.g., externalizing behaviors, [49, 63], social skills [14, 15], and academic skills [5, 65]). Difficulties in ER also show high concordance with the core features of autism spectrum disorder (ASD, [43, 50, 61]). For example, impairments in communication, affective expression, and reciprocal play are often associated with emotional dysregulation [10]. Although neurotypical children make developmental strides in learning to regulate their emotions during their early school years, many children with neurodevelopmental disorders, including those on the autism spectrum, continue to struggle with ER into adolescence and adulthood [45].

Emotional regulation can be measured during childhood using questionnaires, direct observation, and physiological measurement, such as heart rate [64]. Studies of ER in individuals with ASD suggest that they experience increased negative emotions and reduced positive emotions [3,7,56, 30]. Most previous research examining ER in very young children (2 years and under) has used parent questionnaires [44] that assess temperament, that is, individual differences in reactivity and self-regulation of emotion, attention, and activity [53], rather than direct (i.e., physiological) measures. For example, Capps et al. [7] compared ratings on the parent-rated Emotion Behavior Checklist [33] between children with ASD and neurotypical children who were matched on mental age (24 months). Parents of children with ASD rated their children as showing more sadness and fear, as well as less joy than did parents of neurotypical children. Similarly, Garon et al. [20] examined parent ratings on the Infant Behavior Questionnaire-Revised (IBQ-R, [52]) at 12 months and the Toddler Behavior Assessment Questionnaire-Revised [54] at 24 months and found that parents of infants at an increased likelihood of an ASD diagnosis (IL, younger siblings of children diagnosed with ASD) rated their children as showing higher levels of fear, sadness, and anger, and lower inhibitory control, soothability, attention focus, high pleasure, and low pleasure compared to typically developing peers. Furthermore, IL infant siblings who were later diagnosed with ASD at age 3 showed lower levels of positive affect at 12 and 24 months and lower effortful control at 24 months, compared to IL infant siblings who were not diagnosed with ASD at age 3, Garon et al. [20]. Most recently, Ersoy et al. [16] asked parents of IL and children without a family history of ASD (low likelihood, LL) children to complete the IBQ-R at 9 and 15 months of age, when no group differences emerged for the sadness scale. However, the Early Childhood Behavior Questionnaire [51] administered at 24 months yielded higher levels of sadness among the IL group than for LL children.

The earliest age at which the emotional expressivity of children with ASD has been directly observed during emotionally valanced tasks was two years. Macari et al. [39] found that children at age 2 with ASD displayed lower intensity fear, but no differences for anger or joy when compared to age-matched neurotypical children. In the only other study to look at observed emotion, videos taken at 12 months during toy play (not designed as an emotionally salient task) showed that children later diagnosed with ASD had lower rates of positive affect (i.e., smiling) compared to children who were not diagnosed with ASD [17]. Thus, further examination of positive and negative emotional responses early in life in relation to ASD is warranted.

In the present study, we examined behavioral responses to emotionally salient stimuli at 12 and 18 months of age in children who were at a low likelihood (LL; no family history of ASD) and IL (infant sibling of child with ASD) for ASD. Predictions were informed by previous studies of ER in older children with ASD [3,7,56, 30]. Specifically, we predicted that (1) children in the IL group would display higher levels of negative affect and lower levels of positive affect during the Emotion-Evoking (EE) Task, which was adapted from the Laboratory Temperament Assessment Battery (Lab-TAB, Goldsmith and Rothbart 1996), compared to children in the LL group at 12 and 18 months; and (2) affect and gaze at 12 and 18 months would predict ASD symptoms at 24 months. To test the assumption that our EE task was a valid measure of ER, we predicted that affect and gaze would be associated with concurrent ratings on the IBQ-R at 12 and 18 months.

We explored behavioral responses (affect and gaze) to emotionally salient stimuli at 12 and 18 months of age by children who were at a low or increased likelihood for a later diagnosis of ASD. Parents completed the IBQ-R temperament questionnaire at 12 and 18 months, and all children received an ADOS-2 assessment for ASD symptomatology at 24 months. There were three main results. First, the IL group showed higher rates of negative affect and spent less time looking at the task objects compared to the LL group during the Emotion-Evoking Task. Second, affect and gaze showed concurrent associations with several IBQ-R subscales for both the LL and IL groups. Third, gaze at 12 months and gaze and affect at 18 months, but not IBQ-R scores, predicted ADOS-2 scores at 24 months in the IL group. These results suggest that behavioral responses to emotionally salient stimuli may provide important information to support early detection of emerging ASD symptoms, complementing parent ratings of temperament in IL children as early as 12 months of age.

A critical consideration when assessing ER is to determine whether the tasks are producing the expected result (i.e., the putatively negative tasks elicit negative responses [55]). The tasks used in this study were adapted from the Lab-TAB [24] and were designed to probe-specific emotions. Comparisons across our tasks showed increasingly negative responses following bubbles (most positive) to face washing and hair brushing (most negative). Participants also spent more time looking at the more positive tasks (bubbles and toy play) and less time looking at the negative tasks, particularly toy removal, hair brushing, and face washing. The reduced time spent looking at on-task objects during hair brushing and face washing may also be related to the difficulty of looking at a comb/brush and face cloth during these tasks, as well as attempts to avoid (move away from) the brush and face cloth. Some children in both groups responded in ways that were incongruent with the probed emotion, for example, smiling during toy removal. Despite this individual variability, we showed that the vast majority of responses aligned with the probed emotion, which may reflect the validity of the task (positive tasks were experienced as positive, and vice versa), and the placement of a neutral task between the positive and negative tasks to allow time to recover from the previous emotionally salient stimuli [55, 58]. That our tasks appear valid is important because we chose tasks that children could experience in their day-to-day life that would be emotionally valanced (positive or negative) without being too emotionally arousing for the children (evidenced by the low means [< ± 1] for affect during negative and positive tasks).

As noted, we included three baseline periods within our testing protocol. The first allowed participants to acclimate to the testing environment and provided baseline values for affect and gaze, the second allowed an opportunity to recover to minimize carry-over from positive to negative tasks, and the third provided an opportunity to recover from stress produced by the negative tasks, per methodological recommendations [55, 58]. Although we did collect heart rate data in this study, these were not examined in the current report. We did, however, follow the protocol for testing autonomic nervous system reactivity (calculating the difference between affective responses during the emotionally salient stimuli and baseline [35]). Comparisons of affect and gaze during baseline showed no differences between the LL and IL groups. Participants (collectively) had slightly more negative affective responses and spent less time looking at the screen during baselines 2 and 3 compared to baseline 1. Evaluation of ER during baseline is important as it provides a measure of the child’s ability to regulate their emotions [2]. That our participants showed more negative affect and spent less time looking at the computer screen during successive baseline periods may be the result of (1) the EE Task, highlighting the importance of baseline periods to minimize carry-over effects and reduce cumulative stress to the child caused by emotionally challenging tasks, (2) the child becoming restless or fatigued from the EE Task, and/or (3) the child becoming bored by the baseline video, which was the same across the three baseline periods.

The LL and IL groups showed differential responding during the emotionally salient tasks, as predicted. The IL group displayed higher rates of negative affect and spent less time looking at the task objects compared to the LL group, in accordance with previous research on parent ratings of temperament in children diagnosed with ASD [7,36,57, 21]. Although there is a paucity of research on observed ER in children under 2 who are at increased likelihood of/diagnosed with ASD, a few studies have explored ER in children between ages 2 and 5 years. Jahromi et al. [34] assessed facial affect in 4-year-old children with and without ASD during two frustration tasks (toy in a locked box and unsolvable puzzle) and found no differences between the two groups. Similarly, Zantinge et al. [67] presented 5-year-old children with and without ASD with an unpredictable toy robot and recorded facial affect; again, the researchers did not find group differences. Hirschler Guttenberg et al. [30] measured affect and gaze during tasks designed to elicit fear (experimenter wears masks) and joy (child and parent play with hand puppets) in 5-year-old children with and without ASD. Although no differences were found for gaze, positive emotions were reduced and fear was increased during the fearful task in children with ASD, but only when fathers rather than mothers were present. The protocol that most closely resembled ours was carried out by Macari et al. [39]. Two-year old-children with ASD and neurotypical children participated in tasks designed to elicit anger, fear, and joy using tasks from the Lab-TAB. The researchers found that children with ASD displayed lower intensity fear, but no differences for anger or joy when compared to neurotypical peers.

Our findings of differences between the LL and IL groups may be explained by differences in methodology relative to previous studies. First, our participants were tested at younger ages [12 and 18 months vs. ~20 [39] or ~50 months [30, 34, 67]], and as such, may be more reactive because ER systems are still developing. Second, the previous studies included smaller samples and selected children with higher cognitive and language functioning [34]. Our relatively large sample of IL children was tested at two time-points, and we did not select participants based on level of cognitive or language ability. Third, we employed shorter intervals for coding affect (5 s) compared to the 10-s (or longer) intervals used by Jahromi et al. [34], Macari et al. [39], and Zantinge et al. [67], which may have allowed us to capture more nuanced changes in affect.

As predicted, the validity of our Emotion-Evoking Task relative to assessing emotion regulation was supported by concurrent relations with temperament on the parent-reported IBQ-R. Interestingly, when both the IL and LL groups were combined, significant relationships were not found at 12 months of age, but were found for both affect (mask 1, mask 2, and hair brushing) and gaze (baseline 1) at 18 months. When separated out, the IL group did show significant relationships between three subscales on the IBQ-R at 12 months and affective responses on the mask 1 phase. At 18 months, affect and gaze during baseline phase 1 (before any EE Task phase) was associated with low pleasure and perceptual sensitivity, and affective responses during mask 2 and hair brushing were associated with sadness and distress to limitation, respectively. For the LL group, affect during mask 2 was associated with recovery rate and gaze during face washing was associated with cuddliness. There were no relationships for the LL group at 18 months. These finding are important because they suggest our EE Task shows convergent validity with parent-reported temperament, specifically the affective responses during negative tasks for mask 1, mask 2, hair brushing, and face washing and gaze durations for baseline 1). These results are in line with a recent review by Sacrey et al. [55], which reviewed physiological and affective responses during emotionally salient tasks and found that the overwhelming majority of studies used negatively salient tasks to elicit responses. That there were relationships between EE Task and the IBQ-R at 18 months for the combined and IL group, but not the LL group may be due to the age parameters of the IBQ-R. The IBQ-R had suggested use for infants between 6 and 12 months of age. We included it here at 18 months both for consistency between time points for the parent-reported measure and the EE Task, but also due to variability in the developmental age of the IL group (which was confirmed by the Mullen subscales at 18 months). Nevertheless, all subscales of the IBQ-R were significantly correlational with each other at 12 and 18 months. Temperament is viewed as the biologically based disposition to express certain emotions when challenged, and with development we learn to regulate our expressed emotions with respect to our inherent disposition using a variety of ER strategies [18].

Associations between affect, gaze, and IBQ-R scales and ADOS-2 Total scores supported our prediction that affect and gaze would predict ASD symptoms at age 2. The discriminatory ability of affect and gaze was important for the IL group. Gaze at 12 months and both affect and gaze at 18 months were predictive of 24-month ADOS scores in the IL group. Differences in ER have been associated with later mental health disorders [32], as maladaptive ER strategies tax our cognitive capacity and increase autonomic arousal, resulting in long-term ER dysregulation [4, 25]. As such, our results are in accordance with studies that report a higher prevalence of emotional difficulties in children with ASD compared to neurotypical children [12] and children with intellectual disability [6]. Rates of emotional difficulties in children with ASD are reported to range from 71 to 86% [50, 61], with over 50% reporting four or more internalizing or externalizing problems [43]. Because emotional difficulties can have negative effects on a child’s academic ability and quality of life, as well as on their families [19, 66, 60], the earlier ER difficulties can be identified, the earlier interventions can be implemented. For example, the Attachment and Biobehavioral Catch-Up intervention has been shown to improve emotional dysregulation through mother-oriented strategies in emotionally dysregulated infants as young 12 months [28], although long-term effects will be important to demonstrate.

Our study has several strengths; we measured behavioral responses to positively and negatively valanced tasks twice prior to age 2, we included three baseline periods to minimize carry-over effects between positive and negative tasks, our effect sizes were within the medium range, and our sample of IL infant siblings was relatively large.

Limitations include first that there may be a difference in ER between IL siblings and children with non-familial ASD; as such, our results may not be generalizable to non-IL samples. Second, due in part to the age of participants, we did not identify outcomes based on clinical best estimate diagnosis (ASD versus no ASD), but rather compared LL and IL groupings and used ADOS-2 scores as an index of ASD symptoms. Third, the lesser percentages of time spent looking at the on-task object for the IL group may have impacted the affect results. That the IL group spent less time looking at the mask 1, mask 2, and toy removal phases, but did not differ in affective responding from the LL group, may suggest that the IL group was gazing away from the on-task object as a means of regulating their affective response [62]. Further examination in the different types of gaze used during the phases of the EE Task (e.g., looking at parent), is warranted. However, there is value in examining early ASD symptoms on a continuum, especially in relation to emotion regulation in siblings of children with ASD, for whom a higher prevalence of mental health difficulties is an additional concern beyond increased likelihood of ASD [31].

Future work will include comparison of affect and gaze between IL siblings stratified by ASD diagnosis at age 3. Nevertheless, the current study contributes to the growing evidence that ER difficulties are one of the earliest expressions of ASD vulnerability and manifest as early as 12 months of age. These results have the potential to inform ASD surveillance efforts as well as novel treatment strategies to interrupt pathways between emotional dysregulation and academic, behavioral, and social impairments [5,14,15,49,63,65].

Our study is the first to show that children with increased familial likelihood of an ASD diagnosis have differences from children at community-level risk in directly observed behavioral responses to emotionally evocative stimuli by as young as 12 months. These findings add to the cumulative evidence that children at IL for ASD have very early ER difficulties. Observed behavioral responses in the IL sample, but not parent ratings on the IBQ-R, were associated with later ASD symptoms, highlighting the importance of directly observing behavioral responses in emotionally salient situations. The associations between increased negative affect for participants on the mask 1, mask 2, hair brushing, and face washing phases and parent endorsement of more problematic scores on the scales that measure distress or sadness when placed in a confined position, when barred from performing a desired activity, or when engaged in caretaking activities may help focus future work on ER in children with ASD to those tasks and scales that show the highest concordance. Further, these more negatively salient tasks were those that predicted ASD symptomology at 24 months. These observations may provide nuanced differences that can complement standard parent-reported temperament questionnaires.