Perceptual categorization of and adaptation to human voice and musical instruments: A passive-listening study

The human voice is a highly social and relevant auditory stimulus. Previous studies have observed a wide range of voice-sensitive effects, but less is known about the role of attention and context when categorizing sounds as either voice or non-voice. To address this gap, the current study adopted electroencephalography (EEG) passive listening tasks. In Experiment 1, vowel utterances (/a/, /o/, /u/, and /i/) and instrumental tones (bassoon, horn, saxophone, and viola) were presented with equal probability in a random sequence, and different brain responses to the two categories were observed. In Experiment 2, mismatch negativity was observed for rare instrumental tones (viola) embedded in a random sequence of four different vowels, but not vice versa, suggesting that categorization of voice and non-voice could require little to no attention but may be modulated by stimulus familiarity. In Experiment 3, ambiguous voice-instrument morphs were presented in either vocal or instrumental contexts. Logistic regression models performed above chance in predicting the type of context (voice or instrument) from the responses to ambiguous morphs. The results suggest that neural signatures of both perceptual categorization (voice/non-voice) and context effects can be observed in EEG responses under passive listening conditions. [Work supported by NIH grant R01 DC012262.]