An active inference account of skilled anticipation in sport

Abstract

Optimal performance in time-constrained and dynamically changing environments depends on making reliable predictions about future outcomes. In sporting tasks, performers have been found to employ multiple information sources to maximize the accuracy of their predictions, but questions remain about how different information sources are weighted and integrated to guide anticipation. In this paper, we outline how active inference, a unifying account of perception and action, explains many of the prominent findings in the sports anticipation literature. Active inference proposes that perception and action are underpinned by the need to minimize prediction errors and optimise a predictive model of the world. To this end, decision making approximates Bayesian inference and actions are used to minimize future prediction errors. Using a series of Bayesian neurocomputational models based on a partially observable Markov process, we demonstrate that key findings from the literature can be recreated from the first principles of active inference. In doing so, we formulate a number of novel, empirically falsifiable hypotheses about human anticipation capabilities which could guide future investigations in the field.

A Bayesian computational model to investigate expert anticipation of a seemingly unpredictable ball bounce

“…Recent work has outlined how applying Bayesian brain perspectives (see Friston et al, 2017;Knill & Pouget, 2004;Rao & Ballard, 1999) to the study of sporting anticipation can help to explain how performers integrate and weight multiple sources of information to make predictions during time constrained tasks (see Gredin et al, 2018Gredin et al, , 2020Harris et al, 2021). In the present study, we used a Bayesian active inference model of perception to examine observers' contextual priors regarding likely ball bounce trajectories and their sensitivity to early postural cues from the kicker and ball flight trajectory.…”

“…The findings from the current study extend knowledge of anticipation in sports by developing an understanding of precision-based weighting of information sources. Previous models of anticipation in sport have pointed to the use of contextual priors and online sensory input (Müller & Abernethy, 2012), discussed issues related to how these information sources are weighted based on reliability (Gray & Cañal-Bruland, 2018;Runswick, Roca, et al, 2020), and suggested a Bayesian approach to tackle this question (Gredin et al, 2018;Harris et al, 2021). However, the work presented in this study is the first to combine these narratives in the literature to investigate information integration processes in expert performers using a combination of empirical data and computational modelling.…”

“…In addition to revising their generative model, a Bayesian agent can minimize prediction errors through movements (e.g., see the use of vision to minimize surprisal; Parr et al, 2021) or can actively change the world into the predicted state (Adams et al, 2013;Sarpeshkar et al, 2017). Harris et al (2021) have recently suggested that active inference can enhance our understanding of skilled anticipation by providing a principled account of how actions are used to optimize predictions, as well as accounting for decision making via Bayesian inference. However, such theorizing about the use of action to optimize predictions is largely missing from current accounts of anticipation, as identified in framework.…”

“…(2020) reported that both novice and expert groups anticipated ball bounce more accurately when more sensory cues were available (the ball flight and postural cues from the kicker), and that expert differences were most pronounced for early occlusion (postural cues only). The active inference account of anticipation (Harris et al, 2021) simply explains the underlying mechanism of these effects based on the developed generative model of the expert performer and the relative contribution of less certain sensory cues (i.e., during visual occlusion). However, active inference models are yet to be explicitly applied to empirical data during anticipation in sport.…”

“…However, active inference models are yet to be explicitly applied to empirical data during anticipation in sport. Therefore, we aimed to extend the active inference approach to anticipation outlined in Harris et al (2021) by scrutinizing the relative contributions of online sensory information and contextual priors during expert anticipation. To this end, we performed a reanalysis of the data reported in by fitting a Bayesian computational model to the anticipatory behaviour of novice and expert players to determine the role of contextual priors (in the form of prior knowledge about the behaviours of a bouncing oval ball) and sensitivity to sensory information (in the form of the opponent's kinematics during the kick and the kinematics of the ball during flight) in their anticipation performance.…”