Behavioral and neural sound classification: Insights from natural and synthetic sounds

Abstract

Throughout the course of a day, human listeners encounter an immense variety of sounds in their environment. These are quickly transformed into mental representations of objects and events that guide subsequent cognition and behavior in the world. Previous studies using behavioral and temporally resolved neuroimaging methods have demonstrated the importance of certain acoustic qualities for distinguishing among different classes of sounds during the early time period following sound onset (noisiness, spectral envelope, spectrotemporal change over time, and change in fundamental frequency over time). However, this evidence is largely based on correlational studies of natural sounds. Thus, two additional behavioral (Experiment 1) and EEG (Experiment 2) studies further tested these results using a set of synthesized stimuli (interspersed among a new set of natural sounds) that explicitly manipulated previously identified acoustic dimensions. In addition to finding similar correlational results as previous work (using new natural stimuli and tasks), classification results for the synthesized acoustic feature manipulations reinforced the importance of aperiodicity, spectral envelope, spectral variability and fundamental frequency change for representations of superordinate sound-categories. Analyses of the synthesized stimuli suggest that aperiodicity is a particularly robust cue in distinguishing some categories and that speech is difficult to characterize within this framework (i.e., using these acoustic dimensions and synthetic feature manipulations). These results provide a deeper understanding of the neural and perceptual dynamics that support the recognition of behaviorally important categories of sound in the time windows soon after sound onset.