Multisensory Perception and Decision-Making With a New Sensory Skill

Abstract

It is clear that people can learn a new sensory skill—a new way ofmapping sensory inputs ontoworld states. It remains unclear how flexibly a new sensory skill can become embedded in multisensory perception and decision-making. To address this,we trained typically sighted participants (N= 12) to use a newecho-like auditory cue to distance in a virtual world, together with a noisy visual cue. Using model-based analyses, we tested for key markers of efficient multisensory perception and decision-making with the new skill.We found that 12 of 14 participants learned to judge distance using the novel auditory cue. Their use of this new sensory skill showed three key features: (a) It enhanced the speed of timed decisions; (b) it largely resisted interference from a simultaneous digit span task; and (c) it integrated with vision in a Bayes-like manner to improve precision. We also show some limits following this relatively short training: Precision benefits were lower than the Bayes-optimal prediction, and there was no forced fusion of signals. We conclude that people already embed new sensory skills in flexible multisensory perception and decision-making after a short training period. A key application of these insights is to the development of sensory augmentation systems that can enhance human perceptual abilities in novel ways. The limitations we reveal (sub-optimality, lack of fusion) provide a foundation for further investigations of the limits of these abilities and their brain basis.