Exploring the social-behavioral-cognitive pathways to English learning self-efficacy in college students

English learning self-efficacy is a critical psychological factor associated with college students' engagement and persistence in English learning, yet its relationships with the social environment, health behaviors (e.g., physical activity), and cognitive resources have rarely been examined within an integrated framework. Based on social cognitive theory, this study constructed a structural model to examine the direct and indirect associations of social support, physical activity, and a cognitive construct operationalized as self-reported cognitive functioning with college students' English learning self-efficacy. A cross-sectional survey was conducted among 510 Chinese college students, with model comparison and path analysis performed by categorizing social support (overall/family/friend/other) and physical activity intensity (total/moderate-to-high-intensity/walking). Social support and self-reported cognitive functioning were positively associated with English learning self-efficacy, whereas an adjusted direct structural association between physical activity and English learning self-efficacy was not supported. Family support showed a relatively stronger direct association (β = 0.321, P < 0.001), and the association of self-reported cognitive functioning was stable across models (β = 0.361-0.470, P < 0.001). Moderate-to-high-intensity physical activity showed the most prominent association with self-reported cognitive functioning (β = 0.008-0.009, P < 0.001). Bootstrap analyses indicated that the indirect association between social support and self-efficacy through self-reported cognitive functioning was stable across models, whereas the serial indirect association through physical activity and self-reported cognitive functioning was significant only for moderate-to-high-intensity activity (P < 0.05). This indicator-dependent pattern suggests that physical activity intensity may be a boundary condition of the observed serial association. Because the data are cross-sectional and cognitive functioning was assessed by self-report, these findings do not establish temporal ordering, objective cognitive mediation, or causal mechanisms.