Results are reported in four sections. In the first, we report differences related to parent (mother, father) and group (ASD, TD) in the effects of social rupture (SF) on child and parent’s social behavior. In the second, we test differences in regulatory behavior of children and parents. The third section presents data on mothers’, fathers’, and children’s cortisol reactivity. Finally, Pearson’s correlations assess the inter-relatedness between child factors (symptom severity, IQ, baseline CT) and parent and child’s regulatory behavior. Prior to data analyses, we examined child gender effects on all study variables and none was found, and thus, analyses are reported across gender.
Still-face effect on parent and child’s social behavior
Child social behavior
Four repeated-measure ANOVAs were computed to assess change in each social behavior (negative emotionality, withdrawal, positive vocalizations/laughter, and social gaze) as a function of episode (FP, SF, RE) and parent (mother, father) with group as the between-subject factor. Findings for the four social behaviors according to parent and group are presented in Figure 1A,B,C,D.
Negative emotionality
An overall effect was found for episode: F(4,74) = 3.66, P = .009, effect size (ES - partial Eta squared) = .16, which was significant during interactions with both mother, F(1,74) = 4.272 = .016, ES = .053, and father, F(1,304) = 9.13, P < .001, ES = .10. Children expressed more negative emotionality/anger during the SF episode compared to FP, t(1,78) = 2.65, 2.87, P < .01, for mother and father respectively, and negative emotionality returned to baseline at RE. No group or group-by-episode interactions were found, indicating a similar pattern in TD and ASD.
Withdrawal
Similar findings emerged for withdrawal. Significant overall effects were found for episode: F(4,73) = 4.009, P < .01, ES = .05: mother, F(4,74) = 6.05, P = .003, ES = .07; father: F(4,74) = 5.68, P = .004, ES = .07, with no group or interaction effect. Children expressed more withdrawal during the SF episode compared to FP, t(1,78) = 2.81, 2.87, P < .01, for mother and father respectively, and returned to baseline at RE with both parents.
Positive vocalizations/laughter
Significant effect for episode emerged only during interaction with mother; F(2, 76) = 3.53, P = .034, ES = .04. Positive vocalizations/laughter was low during FP and SF, but both TD and ASD children increased positive vocalizations when mother resumed play after rupture, t(1,78) = 2.03, P < .05. No effects were found for father.
Social gaze
An overall effect was found for episode: F(4,74) = 8.83, P < .001, ES = .32: mother, F(1, 74) = 40.27, P < .001, ES = .34; father: F(4,74) = 28.18, P < .001, ES = .27, indicating that child’s social gaze changed across the three episodes. A significant group-by-episode effect for fathers was found; F(4,74) = 5.14, P = .001, ES = .22. During mother-child FP, children mainly engaged in joint attention to toys, but gaze to the unresponsive mother increased during SF, t(1,78) = 4.15, and during RE children maintained high vigilance and social gaze did not return to baseline and was higher than during FP, t(1,78) = −1.98, P = .05. For fathers, the SF effect was found only in the ASD group. Among TD children, no differences were found between social gaze to father during FP, SF, and RE; however, for the ASD group, social gaze to the unresponsive father increased from FP to the SF episode, t(1,38) = 3.11, P = .004, and remained high at RE. During the SF episode, children with ASD showed more social gaze to their unresponsive parents compared to TD children t(1,78) = 1.99, P = .05.
Parent social behavior
Parent’s social behavior (negative emotionality/anger, withdrawal, social gaze, positive affect/laughter) was compared for the two episodes where the parent participated actively - free play and reunion - and change in the parent’s social behavior from before to after social rupture was tested using repeated-measure ANOVAs (FP, RE) with group as the between-subject factor.
Mothers’ social behavior before (FP) and after (RE) social rupture
No episode or group effects emerged for negative-angry and withdrawn affect. However, for positive affect, there were both effects for episode, F(2,75) = 26.23, P < .001, ES = .22, and a group-by-episode interaction, F(2,75) = 51.11, P < .001, ES = .40, emerged. Mothers of TD children expressed no positive affect/laughter during FP and play was mainly characterized by mother and child playing with toys and expressing neutral affect rather than high positive arousal. After social rupture, mothers increased the expression of positive affect (M = 8.20%, SD = 10) possibly to restore social communication. On the other hand, mothers of children with ASD expressed higher levels of positive affect initially (M = 4.11%, SD = 5.1) and those remained unchanged at reunion (M = 5.32%, SD = 6.62), indicating that mothers of ASD children color their typical interactions with more positive arousal and focus on maintaining sameness of communication.
For social gaze, significant effects of episode, F(1,77) = 14.94, P < .001, ES = .16, and group, F(1,77) = 4.67, P < .05, ES = .04, were found. All mothers increased social gaze after rupture at reunion. However, mothers of children with ASD showed significantly more social gaze during both FP (TD: M = 15.87%, SD = 17.31; ASD = 22.64%, SD = 21.96) and RE (TD: M = 22.83%, SD = 23.16; ASD = 33.52%, SD = 27.02). These findings demonstrate the great effort mothers of children with ASD recruit in order to provide positive social environment and maintain a person-focus, particularly after rupture, possibly sensing the child’s difficulty in handling moments of maternal unavailability.
Fathers’ social behavior before (FP) and after (RE) social rupture
No effect emerged for fathers’ negative and withdrawn affect. With regard to positive affect, fathers in both groups expressed more positive affect after social rupture at reunion, F(1, 76) = 26.67, P < .001, ES = .26. During free play, no positive affect was expressed; however, during reunion, all fathers increased the expression of positive affect to resume play after social rupture (M = 5.43%, SD = .4.56).
Fathers’ social gaze showed a significant episode effect, F(1, 76) = 13.38, P < .001, ES = .14. Fathers in both groups expressed more social gaze at reunion (M = 23.41%, SD = 18.22) compared to FP (M = 16.23%, SD = 18.07). These data suggest that both mothers and fathers of TD and ASD children increase social involvement following rupture in order to resume positive communication.
Regulatory behavior during still-face and reunion in TD and ASD children
Child regulatory behaviors were tested during (SF) and following (RE) social rupture with repeated-measure ANOVA with group as between-subject factor. Differences in complex regulatory behaviors were measured only during SF, as it is difficult to judge whether behaviors which are not inherently regulatory and are expressed in non-stressful contexts (for example, symbolic play) serve a regulatory function. Child and parent’s regulatory behaviors are presented in Figure 2.
Child regulatory behavior
Putative regulatory strategies
Child with mother
Significant effect emerged for episode, F(1, 76) = 38. 26.67, P < .001, ES = .33, and group, F(1, 76) = 6.48, P = .013, ES = .07. All children decreased their putative regulatory behaviors from SF to RE. Children with ASD used such behaviors significantly more during the SF episode compared to TD children, F(1,77) = 4.44, P < .05, but not during RE.
Child with father
Similar effect emerged for episode, F(1,76) = 12.37, P < .001, ES = .14, and group, F(1, 76) = 5.76, P = .019, ES = .07. Children decreased the use of putative regulatory behavior from SF to RE, but children with ASD used putative regulatory behavior significantly more during both SF and RE: F(1,77) = 9.73, P < .01 (Figure 2A).
Complex regulatory strategies
ANOVA assessing children’s use of complex regulatory strategies during the SF episode showed a parent by group interaction, F(1, 76) = 4.21, P = .041, ES = .05. During maternal SF, TD children employed more complex regulatory behavior than children with ASD, F = 5.63, P = .026; but no differences were found during paternal SF (Figure 2A).
Parent regulation-facilitation following social rupture
Simple regulation-facilitation
Repeated measure ANOVA assessing maternal and paternal simple regulation-facilitation strategies at RE showed an overall effect for parent, F(1,76) = 7.38, P = .008, ES = .09, and group F(1,76) = 4.45, P = .042, ES = .05. Mothers used more simple regulatory behaviors than fathers and more simple strategies were used by parents of ASD children as compared to parents of TD children.
Complex regulation-facilitation
A similar repeated-measure ANOVA showed a parent-by-group interaction effect, F(1,76) = 5.13, P = .026, ES = .03. This effect indicated that whereas no group differences were found between mothers of TD and ASD children in use of complex regulation-facilitation strategies, fathers of ASD children used more complex regulation-facilitation tactics than fathers of TD children (Figure 2B).
Cortisol reactivity in TD and ASD children with mother and father
Child and mother
Repeated-measure ANOVA of the three child cortisol assessments with mother revealed a main effect for assessment, F(2, 75) = 3.63, P = .031, ES = .14, indicating that cortisol changed over time, and an assessment-by-group interaction, F(2,75) = 3.42, P = .38, ES = .12, demonstrating that change over time differed among groups. Among TD children, CT was high at baseline and declined over time. On the other hand, children with ASD did not show the initial stress response when mother was present and no difference was found in CT levels between the three assessments (Figure 3A). Mothers’ cortisol showed decline over time, F(2, 75) = 4.33, P = .032, ES = .11, with no group or interaction effect (Figure 3B).
Child with father
Repeated-measure ANOVA of child cortisol with father, showed decline over time from initial high levels, F(2,73) = 10.10, P < .001, ES = .29, for both groups with no group or interaction effect (Figure 3C). Fathers’ cortisol similarly showed decline over assessment, F(2,73) = 6.67, P = .002, ES = .19, with no group or interaction effects (Figure 3D). Fathers’ CT also declined over time F(2, 75) = 4.65, P = .027, ES = .13, with no group or interaction effect. No differences were found between maternal and paternal CT. These findings may suggest that the human stress response reacts to social novelty (strangers entering the house with covered boxes), which declines over time, and that by 3 years of age TD children present the adult profile.
Correlations between parent and child’s cortisol
Medium to high correlations (r = .35 to .91) were found between each individual’s three cortisol assessments. Child baseline cortisol with mother and father showed cross-time stability, r = .43, P < .001. At each cortisol assessment (baseline, reactivity, recover), there were significant correlations between parent and child’s cortisol levels, with magnitude of correlations ranging from, r = .36, P < .01 to r = .62, P < .001. These correlations point to ‘endocrine fit’ [56] or ‘biological synchrony’ [57] between parent and child’s hormonal levels.
Correlations between child factors and parent and child’s social and regulatory behavior
Pearson’s correlations between child factors - including symptom severity on the ADOS, IQ, and baseline CT with mother and father - with parent and child’s regulatory behavior appear in Additional file 2. As seen, child symptom severity score correlated with lower IQ. Children with more severe ASD symptoms and with lower IQ tended to use more simple and less complex ER behavior during both SF and RE with mother and father. Child IQ and symptom severity were generally not related to measures of parent regulation-facilitation, apart from a negative correlation between child IQ and maternal simple strategies. Baseline cortisol with mother and father showed medium-level correlation but CT was unrelated to parent or child’s regulatory behavior. Child simple strategies showed significant correlations between the SF and RE episodes with mother and father, between SF with both parents, but not between RE with the two parents. Child complex strategies showed no stability between episodes or parents, indicating that the use of complex strategies may be more context-bound. The use of simple and complex strategies mainly showed negative correlations, and this probably resulted from our coding scheme that defined each regulatory behavior as either simple or complex. Finally, mothers’ and fathers’ simple and complex regulation-facilitation strategies were unrelated.