Multisensory integration is essential for learning and sensorimotor coding, facilitating learners' adaptation to environmental changes. Recent findings confirm that introducing unreliability into visual feedback enhances the use of motor coding, probably because proprioceptive cues are given greater weight. The present study was designed to test this hypothesis and, more generally, to explore the impact of visual versus proprioceptive cue reliability on learning processes. Participants performed a 12-target pointing sequence 100 times with different combinations of visual and proprioceptive feedback: reliable versus unreliable. Retention tests and intermanual transfer tests were administered 24 hours later. Results showed that learning and sensorimotor coding were both affected by the different combinations of visual and proprioceptive cue reliability. Fully reliable feedback allowed for the best retention, while fully unreliable feedback resulted in the worst retention. Visual reliability alone mediated the level of visuospatial coding performance in visuospatial transfer, regardless of the level of proprioceptive reliability, and conversely, reliable proprioception combined with unreliable vision provided the optimum sensory environment for motor coding in the motor transfer test.
Overall, our study highlighted the essential role of both visual cue reliability and proprioceptive cue reliability -and their interactions-in motor learning and its generalization.