Decreasing predictability of visual motion enhances feed-forward processing in visual cortex when stimuli are behaviorally relevant

摘要

Recent views of information processing in the (human) brain emphasize the hierarchical structure of the central nervous system, which is assumed to form the basis of a functional hierarchy. Hierarchical predictive processing refers to the notion that higher levels try to predict activity in lower areas, while lower levels transmit a prediction error up the hierarchy whenever the predictions fail. The present study aims at testing hypothetical modulatory effects of unpredictable visual motion on forward connectivities within the visual cortex. Functional magnetic resonance imaging was acquired from 35 healthy volunteers while viewing a moving ball under three different levels of predictability. In two different runs subjects were asked to attend to direction changes in the ball’s motion, where a button-press was required in one of these runs only. Dynamic causal modeling was applied to a network comprising V1, V5 and posterior parietal cortex in the right hemisphere. The winning model of a Bayesian model selection indicated an enhanced strength in the forward connection from V1 to V5 with decreasing predictability for the run requiring motor response. These results support the notion of hierarchical predictive processing in the sense of an augmented bottom-up transmission of prediction error with increasing uncertainty about motion direction. This finding may be of importance for promoting our understanding of trait characteristics in psychiatric disorders, as an increased forward propagation of prediction error is assumed to underlie schizophrenia and may be observable at early stages of the disease.