Slide Slam F2
Pre-target alpha power predicts comprehension and subjective clarity of Digits-in-Noise
Thomas Houweling1,2, Robert Becker1, Alexis Hervais-Adelman1,2; 1Neurolinguistics, Department of Psychology, University of Zürich, Zürich, Switzerland, 2Neuroscience Center Zürich (ZNZ), Zürich, Switzerland
The resting brain is never static and the ongoing changes in its underlying activity are thought to drive variability in cerebral and behavioural responses to stimuli. It has been shown that intra-individual differences in performance on a variety of auditory tasks, such as near-threshold stimulus detection, pitch discrimination and syllable identification are subject to influence from pre-stimulus brain states. Here we examine the relevance of pre-stimulus brain activity in noisy speech comprehension. Thirty-two normally-hearing participants (mean age = 24y, 26f) took part in an EEG recording (128-channels) during which they performed a digit in noise recognition task. Stimuli were monosyllabic German digits embedded in speech-shaped noise (SSN). Participants rated their subjective impression of clarity after responding to each digit. Digits were preceded by a 3.4 to 6 second-long segment of SSN. They were presented at three signal-to-noise ratios (SNR), individually calibrated using an adaptive staircase procedure achieving 50% accuracy. An easier and a harder condition were created by respectively increasing and decreasing the SNF by 1.5dB. The pre-stimulus (noisy) time window was analysed to determine whether there is a relationship between pre-stimulus activity and subsequent performance, which would be potentially indicative of the engagement of an auditory noise-suppression mechanism. EEG signals were converted to frequency-specific, time-resolved power using Morse wavelets with frequencies sampled logarithmically between 1 and 48Hz. Preliminary analyses (differences in pre-stimulus power averaged over correct and incorrect responses) indicated that a power difference in the α band (8-12Hz) was associated with accuracy at multiple sensors. A univariate mixed-effect logistic regression at the sensor level was used to test the hypothesis that pre-target α power, averaged in 50ms time bins between -500ms and digit onset, and across channels, predicts comprehension or subjective clarity on a trial-by-trial basis. We found that enhanced pre-stimulus α activity is associated with a higher probability of correctly recognising the target digit and associated with a higher subjective clarity rating. Specifically, comprehension is significantly related to α power within a frontal left-lateralised cluster of sensors between -450 and -300ms, whilst subjective clarity is associated with α activity within a left-lateralised temporo-parietal cluster of electrodes between -350 and -150ms. Thus, the ongoing pre-stimulus EEG activity in the face of ongoing noise can predict variation in SiN recognition. The topography of the results suggests that subjective clarity might relate to trial-to-trial variability in auditory attention, whilst comprehension might reflect fluctuations of top-down control on downstream auditory areas. Existing data suggest that reduced pre-stimulus α activity indexes enhanced excitability and results in the cortex being generally more susceptible to input, thus determining more frequent reports of stimulus presence (including false alarms) and higher levels of confidence in the perceptual decision. The association between increased α activity within the noisy pre-stimulus period and performance observed here might be understood in terms of a reduction in cortical excitability, effectively acting as a filter to avoid perceptual over-representation of noise, while leading to enhanced representation or detectability of the signal of interest.