No conclusive evidence for an association between differential observational threat learning and physical activity in everyday life

Background Exercise is well known to generally improve health status in humans and seems to be beneficial not only for physical, but also learning processes. The evidence for the impact of general physical activity on emotional learning, is however scarce. Here, we test the pre-registered hypothesis that the individual physical activity level of the past seven days is positively associated with observational threat learning, indicated by the differentiation between threat and safety cues.

Methods We conducted a two day online study. 90 healthy participants (mean age = 27.82 years) engaged by completing questionnaires (State-Trait Anxiety Inventory (STAI-S/STAI-T and International Physical Activity Questionnaire (IPAQ)) on day one, followed by an observational learning task. Participants were asked to rate their discomfort, fear, and physiological response towards the CS+ and CS-both before and after the learning phase using a visual analogue scale. On day two (approximately 24 hours after day one), participants completed the STAI-S again, followed by a direct generalization task. Similar to before, participants were asked to report their discomfort, fear, and physiological response both before and after the generalization task.

To quantify the level of physical activity (PA) of the past seven days a sum score of the IPAQ was calculated. The pre-registered primary endpoint was to test for apositive association between PA and the ability to discriminate the conditioned stimuli.

Results Pearson’s correlation analyses revealed no significant correlations between the combined total physical activity (PA) score and differential ratings of subjective discomfort (r = 0.11, p_corr = 0.150), fear (r = 0.203, p_corr = 0.081) or physiological responses (r = 0.145, p_corr = 0.17) on day 1. The same analysis revealed no significant correlations on day 2, either (differential ratings of subjective discomfort, r = 0.053, p_corr = 0.93, fear, r = -0.068, p_corr = 0.99, and physiological responses, r = -0.072, p_corr = 0.751 on day 2). However, we also found no sufficient evidence supporting the null hypothesis (i.e. no correlation) for the association between differential learning and PA on day 1, when applying bayesian statistics. Instead, we found a covariation between the ratings of discomfort and physical activity, as well as between ratings of fear and physical activity on day 1 and 2 within a repeated measurement ANOVA. This was supported by bayesian statistics.

Conclusion Our results provided no convincing evidence for a correlation between differential observational threat learning and physical activity (as measured by the IPAQ). Future studies that provide a better control for individual physical activity are warranted.