An active inference account of skilled anticipation in sport

TY - CONF

T1 - An active inference account of skilled anticipation in sport

AU - Harris, David

AU - Arthur, Tom

AU - Broadbent, David P.

AU - Wilson, Mark R.

AU - Vine, Samuel J.

AU - Runswick, Oliver

PY - 2023/5/17

Y1 - 2023/5/17

N2 - Time-constrained and dynamically changing sporting environments require performers to anticipate the future states of the game; the conditions of the ground, the bounce of the ball, or the movement of opponents. To anticipate effectively, performers must identify and use relevant information from both the current environment and their prior knowledge. Important questions remain, however, about how performers select and combine different information sources to aid anticipation, and whether they do so in a statistically optimal way. Neuroscientific theories, such as predictive processing and active inference, characterise perception and action as a Bayesian inference process, and provide a principled account of how performers may enact this information integration process. We will present recent work that has outlined a model of anticipation based on active inference. This work has proposed a series of hypotheses about anticipation derived directly from Bayesian computational models, some of which are consistent with existing evidence and some point to future research questions. We will discuss how active inference may provide a promising framework for future work on anticipation.

AB - Time-constrained and dynamically changing sporting environments require performers to anticipate the future states of the game; the conditions of the ground, the bounce of the ball, or the movement of opponents. To anticipate effectively, performers must identify and use relevant information from both the current environment and their prior knowledge. Important questions remain, however, about how performers select and combine different information sources to aid anticipation, and whether they do so in a statistically optimal way. Neuroscientific theories, such as predictive processing and active inference, characterise perception and action as a Bayesian inference process, and provide a principled account of how performers may enact this information integration process. We will present recent work that has outlined a model of anticipation based on active inference. This work has proposed a series of hypotheses about anticipation derived directly from Bayesian computational models, some of which are consistent with existing evidence and some point to future research questions. We will discuss how active inference may provide a promising framework for future work on anticipation.

M3 - Abstract

T2 - Expertise and Skill Acquisition Network

Y2 - 17 May 2023 through 18 May 2023

ER -