Neural representation of direct-to-reverberant energy ratio in recorded and simulated binaural room auralizations

Abstract

AbstractIn a complex acoustic environment, sound localization involves the extraction and integration of numerous interrelated auditory cues. To understand how these cues are processed in the brain, studies typically isolate a single cue in an artificial experimental framework, to evaluate what brain regions process individual auditory cues. However, multivariate analyses facilitate more complex manipulations with greater ecological validity by providing a method for comparing between brain activity to a quantitative breakdown of the experimental stimuli. Concurrent advancements in virtual acoustics enable a systematic examination of spatial acoustics in complex realistic environments. Although these simulations have a high perceptual plausibility, they still alter auditory reverberation cues in a perceptible way. The impact of these subtle differences on neural processing is unclear. Auditory distance perception is a particularly challenging perceptual process to study, due to the relative nature of the sensory cues. Therefore, we conducted an imaging study to investigate the representation of auditory cues in recorded and simulated acoustic environments, while performing a distance discrimination task. We recorded the actual MRI environment to reduce room divergence, and the auditory simulations modeled reverberation with different degrees of accuracy. We used an acoustic analysis to determine the differences between the acoustic environments and used these quantitative measures to compare to the pattern of brain activity. We found that although the room auralizations were highly similar, it was possible to decode them from brain activity. The ratio of direct-to-reverberant energy level (DRR) was the only acoustic parameter that made a relevant contribution to brain activity. The locus of this activity was in the posterior auditory cortex.