Selective perceptual phase entrainment to speech rhythm in the absence of spectral energy fluctuations

Abstract

Perceptual phase entrainment improves speech intelligibility by phase-locking the brain's high-excitability and low-excitability phases to relevant or irrelevant events in the speech input. However, it remains unclear whether phase entrainment to speech can be explained by a passive "following" of rhythmic changes in sound amplitude and spectral content or whether entrainment entails an active tracking of higher-level cues: in everyday speech, rhythmic fluctuations in low-level and high-level features always covary. Here, we resolve this issue by constructing novel speech/noise stimuli with intelligible speech but without systematic changes in sound amplitude and spectral content. The probability of detecting a tone pip, presented to human listeners at random moments during our speech/noise stimuli, was significantly modulated by the rhythmic changes in high-level information. Thus, perception can entrain to the speech rhythm even without concurrent fluctuations in sound amplitude or spectral content. Strikingly, the actual entrainment phase depended on the tone-pip frequency, with tone pips within and beyond the principal frequency range of the speech sound modulated in opposite fashion. This result suggests that only those neural populations processing the actually presented frequencies are set to their high-excitability phase, whereas other populations are entrained to the opposite, low-excitability phase. Furthermore, we show that the perceptual entrainment is strongly reduced when speech intelligibility is abolished by presenting speech/noise stimuli in reverse, indicating that linguistic information plays an important role for the observed perceptual entrainment.

Keywords: entrainment; envelope; high-level; oscillation; phase; speech.