Robust generalization of tuning to self-induced sensation

Abstract

Perceptual and sensorimotor learning is often specific to the trained stimuli and movement parameters. This specificity also applies to recalibrating sensory and motor maps, such as saccadic eye movements in response to systematic visual errors. Here, we show that the perceptual recalibration of stationarity during smooth pursuit eye movements generalizes to untrained eye movement speeds. During smooth pursuit, the retinal image motion of the stationary surround (reafference) must be compensated to maintain perceptual stability. Prior research revealed that the predicted reafference signal is continuously updated through interactions between the motor command and experienced retinal motion and is specific to movement direction and visual field location. Here, we show that stationarity recalibration transfers across pursuit speeds. The generalization pattern reveals two distinct mechanisms: a multiplicative gain for decreasing predicted reafference signals and a constant shift for increasing signals. The former is consistent with a gain control model of smooth pursuit.