The popularity and promise of mixed reality (MR) technologies, such as virtual (VR) and augmented (AR) reality, have advanced many applications focusing on training and skill development. If successfully deployed, these technologies could prove extremely valuable for high-impact professional training like medical operations where the physical resources are limited or inaccessible. Despite MR’s potential, the effectiveness of skill transfer from the virtual to the physical world remains inconclusive. It is still unclear whether repeatedly performing a task in MR would affect the performance in the same or related tasks in the physical environment. To investigate this issue, participants executed a series of visually-guided manual pointing movements in the physical world before and after spending one hour in VR or AR (n = 20 per group) where they performed the same movements. Results showed that, due to the intrinsic perceptual geometry of the headsets, movements executed in VR were shorter and movements executed in AR were longer than the desired Euclidean distance. More crucially, the perceptuomotor bias in MR also manifested in the subsequent post-test pointing task; participants transferring from VR initially undershoot whereas those transferring from AR initially overshoot the target in the physical environment. These findings call for careful consideration of MR-based training because the exposure to MR may perturb the perceptuomotor processes in the physical environment and negatively impact performance accuracy and transfer of training from MR to UR.
