Is emotional support the key to improving emotion regulation? A machine learning approach

According to the emotion regulation process, situation selection comprises actions that increase or decrease the likelihood of being in contexts that foster a certain type of emotion, positive or negative. This concept is complemented by the social basis theory, which starts with the assumption that the primary ecology of humans is characterized by its social components. Thus, reduced access to social relationships increases cognitive and physiological effort, which leads to a decrease in well-being. In order to make a joint assessment of both concepts, the study used supervised machine learning models to analyze the associations between selected variables of social support, emotion regulation, coping, and several psychological symptoms (somatization, obsession-compulsion, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism). For this purpose, an Argentine sample ( = 812, age = 44.35, female = 435) was collected through the Internet, nested cross-validations were performed with 8 different learning algorithms and Shapley values were computed for the predictive models that minimized the test errors. The results showed that adaptive strategies have considerable effects on maladaptive strategies, but they do not have significant effects on symptoms. Contrariwise, social support variables have significant effects on symptoms, while they do not have major effects on maladaptive strategies. It is concluded that the main function of regulatory flexibility does not appear to be a better adaptation to situations, but rather the maintenance of adequate levels of social support, i.e. emotional support received, perception of available emotional support, and perceived comprehension. Further implications are discussed, and a hypothetical model proposed.