In our study, mothers enrolled in the PT program exhibited less parenting stress, as measured in both the child- and parent-domains of the PSI. PS scores revealed that mothers in the PT group adopted fewer inappropriate overreactive parenting practices.
Our findings were consistent with those of previous studies using PT for mothers with ADHD children, showing decreased maternal stress and enhanced positive parenting practices after PT (Chronis-Tuscano et al., 2013; Heath et al., 2015; Shimabukuro et al., 2017; Treacy et al., 2005). Presumably, these results indicate that PT promotes mothers’ understanding of the behavior of their children with ADHD, thus reducing their stress and improving parenting disciplines. Considering the content of PT, reduced stress might suggest that, by participating in PT, mothers increased their ADHD knowledge and their understanding of the cause-and-effect relationships between their own emotional responses to the child’s misbehaviors, which changed their perception of these behaviors, and thus reduced stress. For improved parenting practices, mothers might well learn the content and purpose of our PT; particularly training on how to provide clear explanations and rules for their children, and to respond appropriately to the children’s ADHD-related non-adaptive behaviors.
Additionally, we conducted PT as a group-based program. Previous studies have shown that group-based programs improved parenting stress and practices in mothers of children with ADHD (Shimabukuro et al., 2017; Treacy et al., 2005). The group-based program provides opportunities for mothers to discuss and share their concerns with peers having similar difficulties (in particular, ADHD), which may further motivate them to participate in the program and enhance their mental health. However, since the non-PT group was not given this opportunity, the effects may include not only those of PT itself, but also the effects of the group meetings.
For the behavioral performance of RTs in the RMET under the ToM condition, the main effect of Time and interaction of Time × Group showed a trend toward significance. The post-hoc test showed significantly reduced RTs at Time 2 only for the PT group. The results of the interaction of Time × Group and the post-hoc test suggest that the PT group tends to improve more with respect to RT compared to non-PT group. However, this should be interpreted with caution, as the main effect of Time showed that improvement tends to be time dependent regardless of the group. In additional analyses, the interaction effect on correct responses remained a trend toward significance, but the main effect of Time was (more strongly) significant, suggesting that the effect of time might be stronger than the effect of PT on the decrease in RTs to correct responses. For the behavioral performance of accuracy in the RMET, a trend toward significance was observed for the main effect of Time, showing somewhat worse performance at Time 2 regardless of the group. It is unclear why RTs and accuracy tended to decrease regardless of the group, but one possible mechanism may be reduced engagement with the task at Time 2 due to boredom and/or habituation (Orr & Stern, 1970). However, this study did not assess the degree of engagement with the task, and this interpretation remains a speculation. Regarding RMET scores, another possibility is the ceiling effect in neurotypical adults, and thus, this tool may be insensitive to slight differences (Black, 2019). Figure 4 shows the distribution of accuracy on ToM condition.
No significant changes in regional brain activation was found in the PT group as compared to the non-PT group under either condition. The current results may underestimate relevant differential effects of PT due to the lack of power caused by the small sample size. However, using a more lenient threshold, we did observe increased activation in the calcarine cortex, inferior temporal gyrus, and inferior occipital gyrus at Time 2 in the PT group, as compared with the nonPT group. These regions are known to be involved in processing facial information, including emotion recognition (Collins et al., 2012; Pitcher et al., 2011). Indeed, previous neuroimaging studies using RMET also reported involvement of these regions during this task (Adams et al., 2010; Castelli et al., 2010b; Schmidt et al., 2020), which agrees with our findings. Furthermore, additional ROI analysis showed activation changes in overlapping region in the left visual cortex (including occipital fusiform gyrus). However, these regions did not survive correction for multiple testing, and the findings should be interpreted cautiously. Planned within-group comparisons showed significantly increased activation in the occipital part of the fusiform gyrus while judging others’ affective states from facial features only in the PT group under the ToM condition. The fusiform gyrus has been reported to be involved in face and object perception (Fusar-Poli et al., 2009), understanding of socioemotional meanings (Gallagher et al., 2000), and in the perception of emotions in facial expressions (Baron-Cohen et al., 1999; Greimel et al., 2010; Li et al., 2019). In one study, using the RMET, Batista et al. (2017) found a positive correlation between ToM scores and cortical thickness in the fusiform gyrus. In another study, Li et al. (2019) used intracranial electrodes and suggested that the fusiform gyrus participates in a relatively early stage of facial expression processing. Thus, we presumed that the increased brain activity and reduced reaction times seen in the PT group might reflect their improved sensitivity/efficiency to others’ affective states.
Taken together, we presumed that these changes might reflect that PT could reduce stress, and involvement of the fusiform gyrus. However, the causal relationship was not clarified in this study since this could not be proven by correlation analysis.
This study has some limitations. First, as mentioned above, it included a relatively small sample size and lacked a control group comprising mothers of typically developed children. Second, as all psychometric measures were self-reported, subjective bias might be present. Lastly, because of the single-center study design, the participants’ demographic variability might be limited. Future research should address these limitations.