Adaptation to mirror-reversed vision is based on directionally tuned modules

Sensorimotor adaptation to rotated visual feedback is thought to be achieved by directionally tuned modules. Here we scrutinize whether adaptation to reversed vision utilizes similar mechanisms. Specifically, we hypothesize that adaptive transfer to unpracticed target directions is determined by the superposition of neighboring modules. One group of subjects adapted to a left–right reversal of visual feedback, which requires a 180°, ±90°, or no change of response direction, depending on target position. Two groups of control subjects adapted to a 180° and to a 90° rotation of visual feedback. We found that adaptation to a left–right reversal is less efficient than adaptation to rotations requiring the same adaptive change, and attribute this decrement to an overlap of neighboring modules. We further found that transfer to unpracticed targets is well predicted by a simple Gaussian model. From this we conclude that adaptation to a left–right reversal emerges in a regional and interdependent fashion, and can be modeled as overlapping Gaussian tuned processes.