Variability in the Estimated Amplitude of Vowel-Evoked Envelope Following Responses Caused by Assumed Neurophysiologic Processing Delays.

Abstract

Vowel-evoked envelope following responses (EFRs) reflect neural encoding of the fundamental frequency of voice (f₀). Accurate analysis of EFRs elicited by natural vowels requires the use of methods like the Fourier analyzer (FA) to consider the production-related f₀ changes. The FA's accuracy in estimating EFRs is, however, dependent on the assumed neurophysiological processing delay needed to time-align the f₀ time course and the recorded electroencephalogram (EEG). For male-spoken vowels (f₀ ~ 100 Hz), a constant 10-ms delay correction is often assumed. Since processing delays vary with stimulus and physiological factors, we quantified (i) the delay-related variability that would occur in EFR estimation, and (ii) the influence of stimulus frequency, non-f₀ related neural activity, and the listener's age on such variability. EFRs were elicited by the low-frequency first formant, and mid-frequency second and higher formants of /u/, /a/, and /i/ in young adults and 6- to 17-year-old children. To time-align with the f₀ time course, EEG was shifted by delays between 5 and 25 ms to encompass plausible response latencies. The delay-dependent range in EFR amplitude did not vary by stimulus frequency or age and was significantly smaller when interference from low-frequency activity was reduced. On average, the delay-dependent range was < 22% of the maximum variability in EFR amplitude that could be expected by noise. Results suggest that using a constant EEG delay correction in FA analysis does not substantially alter EFR amplitude estimation. In the present study, the lack of substantial variability was likely facilitated by using vowels with small f₀ ranges.