Estimation of the Neural Sources of the EEG-Measured Speech-FFR Using a Phenomenological Model of Auditory-Nerve Fiber Responses

The frequency-following-response to continuous speech (speech-FFR) is a characteristic neural activity that emerges at the fundamental frequency of a speaker's voice and can be measured by electroencephalography (EEG) and magnetoencephalography (MEG). Its spectrotemporal dynamics and neural sources encode important pitch information critical for reliable speech processing. EEG studies have found a peak latency of the speech-FFR at around 10 ms with putative subcortical origin while recent MEG studies identified additional cortical contributions to the speech-FFR driven by the carrier and the envelope modulations at higher frequencies. In this study, we examined the spatiotemporal dynamics of the speech-FFR at the fundamental frequency using linear modelling and inverse source localization. We found that the response mainly originates from subcortical sources with major contributions from the midbrain for the two implemented acoustic features, the fundamental waveform and higher frequency envelope modulations. Interestingly, the latter evoked some additional brainstem activity at latencies significantly later than the typical subcortical latency range. Our results confirm the subcortically-dominated nature of the EEG-measured speech-FFR while additional top-down modulation might be evident by recurring brainstem activity.