Controllability changes pain perception by increasing the precision of expectations.

The ability to exert control over an intensely unpleasant experience, such as pain, can modulate its perception. It is often assumed that control exerts this modulatory effect through a specific control mechanism. We revisit this issue using a task that allowed participants to either control or predict the intensity of a painful stimulus. By approximating Bayesian perceptual integration with computational models, our data show that acute pain modulation by control can be parsimoniously explained by an increase in expectation precision. Importantly, this effect is present in contrast to a condition in which pain is equally predictable, but not controllable. The control-induced increase in expectation precision leads to activation changes in the periaqueductal gray, the supplementary motor area and the rostral anterior cingulate cortex, regions that mediate the interplay between threat uncertainty, motor-control and descending pain modulation.