This study, presented by Takemura et al., aims to examine heterogeneity in the association between social participation and depressive symptoms and identify the characteristics that older adults who benefit the most from social participation possess. Longitudinal cohort data was taken from JAGES, a nationwide survey examining social participation and depression in adults aged 65 and over in Japan. Propensity score matching was used on 11,146 participants to create matched groups, and a machine learning (ML) causal forest algorithm was trained using half of these groups. The other half was organized into high-benefit and low-benefit subgroups based on the findings of the ML algorithm and characteristics of each group were analyzed. Social participation was associated with decreased risk of depressive symptoms in the high-benefit group but not the low-benefit group. The high-benefit group was characterized by a higher mean age, higher rate of retirement, lower rate of marriage, higher rate of living alone, lower educational level, lower household income and higher prevalence of past smoking compared to the low-benefit group. These results suggest that the oldest adults as well as adults with lower socioeconomic status receive the most mental health benefits from social participation.