Interpersonal interactions improve the representativeness of embodied decision-making behaviour in football

Decision making is a crucial skill in team sports, yet remains challenging to measure in controlled settings. For instance, video-based testing of decision making has long neglected the interpersonal interactions between attackers and defenders, therefore it remains unclear to what degree embodied decision-making tested in those conditions reflects that observed in the natural environment. The aim of this study was to examine the extent to which interpersonal interactions influence behaviour in an embodied decision-making task. We designed an interactive avatar defender in a virtual-reality setting and conducted an empirical study to evaluate the impact of the interactions. Attackers performed a 1v1 task, and we manipulated the defender's level of interactivity to compare the attacker's behaviour to that observed when facing an opponent in-situ. Decision-making behaviour was compared across four conditions of increasing interactivity (virtual no-opponent, virtual non-interactive opponent, virtual interactive opponent, and in-situ opponent). Results strongly support the idea that adding interactions makes the task more representative of in-situ behaviour. Specifically, key measures of behaviour (e.g., trial duration and number of touches) measured with the virtual-interactive opponent (β_Touches = 4.3, β_Duration = 2.7) were much more representative of the in-situ behaviour (β_Touches = 5.4, β_Duration = 3.6) than they were against the virtual non-interactive opponent (β_Touches = 2.9, β_Duration = 1.84) and in the virtual no-opponent condition (β_Touches = 2.5 β_Duration = 1.4). The results validate the call to incorporate interpersonal interactions as a key component of decision-making tasks, and to support this we introduce a novel method for assessing interactive decision-making skills that we make available open source.