Learning to localise weakly-informative sound spectra with and without feedback

Area of Science:

• Auditory Neuroscience
• Psychoacoustics
• Perceptual Learning

Background:

• The mechanisms by which the human auditory system learns to interpret pinna-induced spectral cues for sound-source elevation remain unclear.
• Previous research indicates significant plasticity in sound localization following alterations to the pinna or visual input.

Purpose of the Study:

• To investigate perceptual learning of sound-source elevation using spectrally degraded sounds with weak elevation cues.
• To determine the role of visual feedback in auditory spatial learning.

Main Methods:

• Listeners were trained on sounds with weak, elevation-dependent spectral cues, with visual feedback provided for a subset of targets.
• Stimulus-response relations were analyzed to assess learning.
• Performance was evaluated with and without visual feedback, and generalization to untrained conditions was tested.

Main Results:

• Listeners showed gradual improvement in response accuracy for trained targets, indicating perceptual learning.
• Performance improved even without visual feedback, though to a lesser extent.
• Post-training, listeners demonstrated generalized improvements in sound-source elevation estimation across various locations and acoustic spectra.

Conclusions:

• The auditory system can learn to re-weight spectral information from different frequency bands to update prior elevation estimates.
• Auditory spatial learning for elevation is influenced by, but not solely dependent on, visual feedback.
• A neuro-computational model can explain the observed perceptual learning and generalization.