Adaptive and Selective Time Averaging of Auditory Scenes.

Area of Science:

• Auditory perception
• Psychoacoustics
• Computational neuroscience

Background:

• Perceptual systems summarize sensory input using statistical summaries to manage variability and estimate scene characteristics.
• While statistical representations in perception are increasingly recognized, the underlying mechanisms, particularly in auditory scenes, remain unclear.
• Auditory background texture is thought to be represented by time-averaged sound statistics.

Purpose of the Study:

• To investigate the temporal averaging mechanisms in auditory perception.
• To determine the duration of the averaging window for auditory statistical summaries.
• To explore how stimulus characteristics and source attribution influence auditory integration.

Main Methods:

• Utilized "texture steps"—subtle shifts in stimulus statistics—to probe auditory averaging.
• Measured listener judgments of texture following imperceptible statistical shifts.
• Analyzed the influence of texture variability and attributed sound sources on integration times.

Main Results:

• A multi-second averaging window was identified, with shifts in statistics biasing judgments even when occurring seconds prior.
• Listeners could not consciously control the averaging window duration.
• Temporally variable textures elicited longer auditory integration times.
• Integration times were significantly longer than previously reported auditory system integration times.
• Auditory integration appeared restricted to sound elements perceived as originating from a common source.

Conclusions:

• Auditory integration is a dynamic process influenced by stimulus properties, extending longer for variable signals to improve statistical estimation.
• The auditory system selectively integrates sounds likely originating from a common source, reflecting real-world causal relationships.
• The developed methodology can be applied to study statistical representations in other sensory modalities.