Attention capture by action effects are driven by prediction error minimization

Abstract

Previously, we showed that stimuli that are considered as action-effects are processed with priority (Kumar, Manjaly & Sunny, 2015). We considered the role of common-coding mechanisms in driving this effect. However, recent studies suggest that these effects may follow the principles laid out by the predictive coding framework. In the present study, we explore how predictions about various parameters involved in action-effect monitoring (location, SOA, contingency) modulate attention capture by an action effect. In five experiments we investigate the effects of uncertainty associated with these characteristics in driving attention capture. Specifically, a reliable prediction error driven by controlled irregularity seems critical for automatic processing of action effects. Moreover, in case of high regularity or high irregularity the visual system uses minimal information from the environment and hence attention control appears to be top-down. The findings present robust evidence supporting the role of predictive coding mechanisms in driving automatic processing of visual information.