Social perspective taking is associated with self-reported prosocial behavior and regional cortical thickness across adolescence

Abstract

Basic perspective taking and mentalizing abilities develop in childhood, but recent studies indicate that the use of social perspective taking to guide decisions and actions has a prolonged development that continues throughout adolescence. Here, we aimed to replicate this research and investigate the hypotheses that individual differences in social perspective taking in adolescence are associated with real-life prosocial and antisocial behavior and differences in brain structure. We used an experimental approach and a large cross-sectional sample (n = 293) of participants aged 7-26 years old to assess age-related improvement in social perspective taking usage during performance of a version of the director task. In subsamples, we then tested how individual differences in social perspective taking were related to self-reported prosocial behavior and peer relationship problems on the Strengths and Difficulties Questionnaire (n = 184) and to MRI measures of regional cortical thickness and surface area (n = 226). The pattern of results in the director task replicated previous findings by demonstrating continued improvement in use of social perspective taking across adolescence. The study also showed that better social perspective taking usage is associated with more self-reported prosocial behavior, as well as to thinner cerebral cortex in regions in the left hemisphere encompassing parts of the caudal middle frontal and precentral gyri and lateral parietal regions. These associations were observed independently of age and might partly reflect individual developmental variability. The relevance of cortical development was additionally supported by indirect effects of age on social perspective taking usage via cortical thickness. (PsycINFO Database Record

Figure 1

Director task instructions. Before the experimental director condition of the task, participants were shown and explained images of their view and the corresponding director’s view of a stimulus configuration with an example of an object that both the participant and the director can see (“car”), and an example of an object that the participant, but not the director, can see (“apple”).

Figure 2

Director task. Top panel: Examples of a critical trial and a control trial in the experimental director condition. Bottom panel: Examples of a critical trial and a control trial in the control no-director condition. For illustration purposes, the target stimulus is highlighted with a green (light gray) circle, whereas the distractor/irrelevant object is highlighted with a red (dark gray) circle.

Figure 3

Director task performance: accuracy. Percentage errors (mean and standard errors) in control trials and critical trials in the director condition and the no-director condition for each age group. Children: 7.1–11.3 years (n = 60), adolescents: 11.7–17.9 years (n = 108), and adults: 18.0–26.7 years (n = 125).

Figure 4

Director task performance: response time. Median response times (mean and standard errors) from correct trials only in control trials and critical trials in the director condition and the no-director condition for each age group. Children: 7.1–11.3 years (n = 60), adolescents: 11.7–17.9 years (n = 108), and adults: 18.0–26.7 years (n = 125).

Figure 5

Associations between task performance and cortical thickness. general linear models (GLMs) were used to test the effects of the difference in percentage errors on director critical trials and on no-director critical trials on cortical thickness, while controlling for sex. The results were corrected for multiple comparisons using cluster size inference. Uncorrected p values within the corrected significant clusters are shown. All clusters showed positive effects, indicating that better performance was related to thinner cortices. No effects were seen in the opposite direction.

Figure 6

Age-independent associations between task performance and cortical thickness. General linear models were used to test the effects of the difference in percentage errors on director critical trials and on no-director critical trials on cortical thickness, while controlling for sex and age. The results were corrected for multiple comparisons using cluster size inference. Uncorrected p values within the corrected significant clusters are shown. Two clusters in the left hemisphere showed positive effects, indicating that better performance was related to thinner regional cortices. No effects were seen in the opposite direction.