| Functional brain imaging studies using electroencephalography (EEG) and magnetoencephalography (MEG) have shown that neural populations in human auditory cortex entrain (or "phase-lock") to the low-frequency amplitude modulations in speech utterances (i.e. its envelope). While this activity was originally believed to represent low-level responses to the speech's acoustics, recent studies have found that phase-locking is sensitive to attention, intelligibility, and the presence of complementary visual information. Thus, phase-locking is more accurately described as an interaction between top-down influences and bottom-up acoustic features. The following studies examined the role of phase-locking in speech perception by analyzing EEG recorded from young adults engaged in speech perception tasks.;Experiment 1 had subjects engage in a "cocktail party" task, wherein they had to attend to one speaker while ignoring a competitor. We demonstrated that phase-locking could be used both to enhance the attended speaker, as well as to suppress the competing speaker, by aligning periods of high (or low) neuronal excitability with the arrival of syllables from the attended (or unattended) speaker. Furthermore, we show that these differences in phase-locking between attended and unattended speech are sufficiently robust within single-trial EEG to form the basis of a novel brain-computer interface (BCI).;In experiment 2, subjects viewed audio-visual speech stimuli containing brief narrative passages. The relative onsets of the audio and video stimuli varied, so that only a subset of the trials appeared to be synchronized. Additionally, on some trials the audio and video stimuli did not match. We found that phase-locking was enhanced when the audio and video components of the speech matched. However, phase-locking did not show the expected enhancements for synchronous conditions, arguing against the prevailing "predictive" theory of how visual information modulates phase-locking to the acoustic envelope.;Together, our findings support the view that phase-locking to speech is both subject to and a mechanism of exerting top-down influences, but further work is clearly needed to determine the full extent of its functions. |
