Contextual Modulation of Primary Visual Cortex by Temporal Predictability During Motion Extrapolation

ABSTRACT Background Predicting future events is a fundamental cognitive ability that often depends on the volatility of the environment. Previous studies on apparent motion reported that when the brain is confronted with low levels of predictability, activity in low‐level sensory areas is enhanced, including the primary visual cortex, while others suggest that the enhanced activity in this area is independent of the predictability level. Consequentially, it remains unclear how temporal predictability modulates brain responses in continuous, thus more ecologically valid, motion paradigms. Purpose Our study investigated whether motion extrapolation in high and low predictable contexts would differently modulate fMRI responses in subject‐specific primary visual cortex during visible and partially occluded stimulation. Methods and Materials Eighteen participants performed a modified version of the interception paradigm in visible and occluded phases, in which they observed a stimulus moving horizontally, then vertically at two different velocities, while fMRI data was acquired. They judged when and where the stimulus would reach a given point of contact. In a high predictable context, the velocity was identical during horizontal and vertical (occluded) movement; whereas, in a low predictable context, the velocity could change during the vertical trajectory, introducing a temporal incongruence to the task with a predictive role of velocity on the trajectory estimation. Results Univariate results indicated that on average both low and high predictable contexts similarly modulated activity in primary visual areas. On the other hand, trial‐history analysis showed that a change in trial type (constant velocity after change in velocity and vice versa) increased BOLD responses in V1. Conclusion This pattern of results suggests that motion extrapolation can modulate activity in the primary visual cortex regardless of average predictability but is influenced by recent trial history. These results were further supported by multivariate pattern analysis, which revealed different patterns when comparing congruent and incongruent trials in the context of lower predictability.