The effects of action-based predictions in early visual cortex

Voluntary action typically results in reduced sensitivity to sensory action outcomes. The forward model theory proposes that this is due to neural suppression. However, this theory has recently been challenged by the pre-activation account, sharpening, and opposing process theory. In this fMRI study, we compared these theories by using univariate and multivariate analyses. Participants performed a visual orientation discrimination task on two sequential gratings, which were presented automatically (passive condition) or triggered by button press (active condition). Decoding of predicted stimulus orientation from early visual cortex activity prior to stimulus presentation was significantly above chance in both active and passive conditions. During stimulus presentation, actively generated stimuli elicited larger blood-oxygen-level-dependent (BOLD) responses. However, both decoding accuracy and the timing of the BOLD responses did not differ between conditions. These results differ from the forward model's predicted sensory attenuation, the pre-activation account's earlier BOLD response, and the enhanced precision of the sharpening hypothesis. Instead, they mostly fit the opposing process theory, which posits pre-activation in both conditions. However, the stronger BOLD response for actively generated stimuli is not predicted by any existing theory, implicating additional mechanisms, such as heightened attention or motor-related enhancement.