Bayes-Like Integration of a New Sensory Skill with Vision

Negen, James; Wen, Lisa; Thaler, Lore; Nardini, Marko

Abstract

Humans are effective at dealing with noisy, probabilistic information in familiar settings. One hallmark of this is Bayesian Cue Combination: combining multiple noisy estimates to increase precision beyond the best single estimate, taking into account their reliabilities. Here we show that adults also combine a novel audio cue to distance, akin to human echolocation, with a visual cue. Following two hours of training, subjects were more precise given both cues together versus the best single cue. This persisted when we changed the novel cue’s auditory frequency. Reliability changes also led to a re-weighting of cues without feedback, showing that they learned something more flexible than a rote decision rule for specific stimuli. The main findings replicated with a vibrotactile cue. These results show that the mature sensory apparatus can learn to flexibly integrate new sensory skills. The findings are unexpected considering previous empirical results and current models of multisensory learning.

Citation

Negen, J., Wen, L., Thaler, L., & Nardini, M. (2018). Bayes-Like Integration of a New Sensory Skill with Vision. Scientific Reports, 8, Article 16880. https://doi.org/10.1038/s41598-018-35046-7

• Journal Article Type: Article
• Acceptance Date: Oct 22, 2018
• Online Publication Date: Nov 15, 2018
• Publication Date: Nov 15, 2018
• Deposit Date: Oct 29, 2018
• Publicly Available Date: Nov 16, 2018
• Journal: Scientific Reports
• Electronic ISSN: 2045-2322
• Publisher: Nature Research
• Peer Reviewed: Peer Reviewed
• Volume: 8
• Article Number: 16880
• DOI: https://doi.org/10.1038/s41598-018-35046-7
• Public URL: https://durham-repository.worktribe.com/output/1310204