Noise schemas aid hearing in noise

Area of Science:

• Auditory Neuroscience
• Psychoacoustics
• Cognitive Science

Background:

• Human hearing exhibits remarkable robustness to background noise, a phenomenon not fully explained by current models.
• Existing theories suggest adaptation to stable sound components, potentially suppressing noise, but this conflicts with noise's informational value.

Purpose of the Study:

• To investigate whether internal models of noise structure, rather than simple adaptation, mediate auditory noise robustness.
• To explore how the auditory system separates relevant sounds from background noise.

Main Methods:

• Participants performed auditory tasks (detection, recognition, localization) with foreground sounds embedded in real-world background noise.
• Noise characteristics were manipulated, including timing of foreground sounds, noise interruptions, and recurring noise patterns.

Main Results:

• Auditory performance improved for sounds presented later in a noise excerpt, with gains over the first second of exposure.
• Performance remained robust to noise interruptions and was enhanced by intermittently recurring backgrounds.
• A reduced benefit for later sounds in recurring noises suggested a maintained internal noise representation.

Conclusions:

• Auditory noise robustness is supported by rapidly estimated, stored, and updated internal models of noise structure, termed "noise schemas."
• These "noise schemas" facilitate the separation of concurrent sounds by predicting and modeling background noise.
• Findings challenge purely adaptation-based models and highlight the role of predictive internal representations in auditory perception.