Play in Predictive Minds: A Cognitive Theory of Play

Abstract

Play occurs universally in human beings, and it has been a subject of considerable academic scrutiny for over 100 years. In this article, we propose a cognitive theory of play building on recent advances in cognitive and computational neuroscience that portray the human brain as an advanced prediction machine. Central to the theory is the idea that when an agent is free from the demands of certain competing cognitive systems, it may deliberately seek out and create surprising situations that gravitate towards sweet-spots of relative complexity. We argue that this framework can explain why humans play and why playing is so fun and rewarding.

Mastering uncertainty: A predictive processing account of enjoying uncertain success in video game play

“…Paradoxically, this means that agents that care about nothing more than prediction error reduction will search for surprise and error all the time and even sometimes deliberately create it. Uncertainty-reducing agents, in other words, will be intrinsically motivated to engage in exploratory play, and experientially enjoy play if and when the thus-generated uncertainty is then reduced surprisingly quickly (Andersen et al, 2022).…”

“…In natural environments, this drive to reduce error at a betterthan-expected rate lets predictive agents drift towards niches that are replete with consumable errors-that is, situations that agents expect and observe to be neither too complex to manage, nor too predictable and devoid of new information (Andersen et al, 2022;Miller et al, forthcoming). For instance, to most people, a high, straight, and smooth pole in the middle of an empty parking lot offers little consumable error: we expect it to be unlikely that we can climb to the top of it, and there is no new information to gain from the view atop.…”

“…We first set out current research on the appeal of uncertain success in games, how ACE theories account for it, and where they fall short. We then give a brief introduction into PP, which has seen a rapid ascent in the cognitive sciences, but also more recently in play research and neuroaesthetics (Frascarolli et al, 2021;Andersen et al, 2022). PP sees living beings as aiming to minimise prediction error -that is, mismatches between predicted and actual states.…”

Preferences need inferences: Learning, valuation, and curiosity in aesthetic experience

“…A relaxed, playful dog picks out the most unwieldy stick, the one that fights back most. A child chooses the most unfamiliar, inappropriate object to play with (Andersen & Roepstorff, 2021;Bonawitz et al, 2012). At least in origin, this is not driven by self-conscious showmanship, let alone a display of prowess or the "costly signaling" of one's fitness, an evolutionary rationale that has been invoked to explain art as well (Dissanayake, 2007;Helmut Leder & Nadal, 2014).…”

Great expectations: The construct validity of the violation-of-expectation method for studying infant cognition

“…Many dominant theories of learning are based on learners making predictions and subsequent revisions after observing evidence. Classic models of associative learning (e.g., Friston, 2005;Fletcher et al, 2001;Hayden et al, 2011;Pearce & Hall, 1980;Rescorla & Wagner, 1972;Schultz & Dickinson, 2000), Bayesian models of cognitive development (e.g., Perfors et al, 2017;Schulz, 2012;Tenenbaum et al, 2006), and predictive processing models of early learning (e.g., Andersen et al, 2022;Köster et al, 2020) all argue that prediction errors drive learning. While these theories differ in many respects, they all have at their cores the assumption that when there is a mismatch between what a learner expects and what they observe, they must update their behavior or knowledge accordingly.…”