Among animals, natural selection has resulted in a broad array of behavioural strategies to maintain core body temperature in a relatively narrow range. One important temperature regulation strategy is social thermoregulation, which is often done by warming the body together with conspecifics. The literature suggests that the same selection pressures that apply to other animals also apply to humans, producing individual differences in the tendency to socially thermoregulate. We wanted to investigate whether differences in social thermoregulation desires extend to other personality factors in a sample of French students. We conducted an exploratory, hypothesis-generating cross-sectional project to examine associations between thermoregulation and personality. We used conditional random forests in a training segment of our dataset to identify clusters of variables most likely to be shaped by individual differences to thermoregulate. We used the resulting clusters to fit hypothesis-generating mediation models. After we replicated the relationships in two datasets, personality was not related to social thermoregulation desires, with the exception of attachment avoidance. Attachment avoidance in turn predicted loneliness. This mediation proved robust across all three datasets. As our cross-sectional studies allow limited causal inferences, we suggest investing into prospective studies to understand whether and how social thermoregulation shapes attachment avoidance early in life and loneliness later in life. We also recommend replication of the current relationships in other climates, countries, and age groups.
Keywords: attachment; machine learning; personality; social thermoregulation.
© 2022 The Authors.