Predictive processing theories, especially active inference, have been proposed as a way to reconcile embodied and traditional cognitive science. This analysis argues that most active-inference accounts rely on weak or trivial conceptions of embodiment, while stronger claims do not follow from the framework itself. A more compelling version of embodied active inference is motivated by taking a diachronic view of how rhythmic physiological activity shapes neural development before birth. The visceral afferent training hypothesis proposes that early-emerging physiological processes, particularly from the cardiovascular system, are essential for configuring cognitive architecture. Three candidate mechanisms are suggested: activity-dependent neuronal development, periodic signal modeling, and oscillatory network coordination.
Motion-induced blindness (MIB) is an illusion where a peripheral target disappears from awareness when viewed against a moving background. This study measured how long a target remained visible at eight positions around the visual field. Disappearance times and frequencies varied significantly with target location: targets on the cardinal axes (horizontal and vertical) vanished less often and for shorter periods than those on oblique axes, and targets on the horizontal meridian vanished less than those on the vertical. These polar angle asymmetries in MIB suggest that conscious visual perception is not uniform across the visual field, with specific consistencies between visual field location and the timing of perceptual disappearances.