(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Endogenous auditory and motor brain rhythms predict individual speech tracking [1]

['Christina Lubinus', 'Department Of Cognitive Neuropsychology', 'Max-Planck-Institute For Empirical Aesthetics', 'Frankfurt Am Main', 'Cooperative Brain Imaging Center', 'Goethe University Frankfurt', 'Anne Keitel', 'Psychology', 'University Of Dundee', 'Dundee']

Date: 2026-08 

Slow, endogenous brain rhythms in the auditory cortex are hypothesized to track acoustic amplitude modulations during speech comprehension. Temporal predictions from the motor system are thought to enhance this tracking. However, direct evidence for the involvement of endogenous auditory and motor brain rhythms is lacking. Combining magnetoencephalographic recordings with behavioral data, we here show that endogenous peak frequencies of individuals’ resting-state theta rhythm in superior temporal gyrus predict speech tracking during comprehension. Importantly, endogenous rates of speech motor areas predicted auditory-cortical speech tracking only in individuals with high auditory–motor synchronization profiles. Higher rates in the supplementary motor area and lower rates in inferior frontal gyrus predicted stronger tracking. These findings provide support for oscillatory accounts of auditory–motor interactions during speech perception. Behaviorally, higher auditory–motor synchronization was related to higher comprehension, with effects of the spontaneous speech motor production rate only in high synchronizers. Working memory capacity predicted speech comprehension performance only in individuals with low auditory–motor synchronization profiles. No significant relationship between the neural data and behavioral readouts was observed. The findings highlight differential speech processing preferences across individuals, with an auditory–motor route related to enhanced comprehension performance.

Funding: We thank the Max Planck Institute for Empirical Aesthetics for funding this project (C.L., J.M.R.) ( https://www.aesthetics.mpg.de/ ). A.K. is supported by the Medical Research Council (grant number MR/W02912X/1) ( https://www.ukri.org/councils/mrc/ ). A.K. is supported by the Royal Society of Edinburgh (RSE Saltire Facilitation Network Award, Reference Number 1963) ( https://rse.org.uk/ ). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Introduction

Verbal communication, by requiring the matching of acoustic and articulatory representations, exemplifies action-perception interactions in humans essential for everyday behavior [1]. In particular, speech production and perception are intricately interwoven. The perception of speech may rely on the identification of intended vocal tract gestures [2], allowing for the prediction of the speech signal, as considered crucial in vocal learning [3]. Others assume a more moderate role of the motor system, such that it is aiding speech perception particularly in demanding listening situations [4,5]. During speech production, the motor system engages in operations requiring precise timing [6,7], with these motor areas possibly being similarly recruited during speech perception [6,8,9]. Specifically, the supplementary motor area (SMA) and (parts of the) inferior frontal gyrus (IFG) are assumed to generate temporal predictions about upcoming sensory events [10–15]. The SMA and IFG take different roles during speech perception, with the SMA receiving input from subcortical loops serving mostly inhibitory input and the IFG more closely linked to auditory cortex processing in the temporal lobe [9,16–18]. Based on the literature, on the one hand, and on our goal to focus on higher-order speech processing and sequencing [19–21], on the other, we operationalized speech motor areas as comprising the pars triangularis of the IFG as well as SMA, while not including for these analyses the region of precentral gyrus implicated in speech motor control [21,22].

A prominent neural oscillatory account of speech perception proposes that slow endogenous brain rhythms in auditory cortex allow for speech segmentation by aligning their neural excitability phase to the speech acoustics (speech tracking; [23–28]). Additionally, neural oscillations from cortical motor areas may be involved in speech perception to various degrees. Slow and fast neural oscillations are argued to aid temporal predictions from motor areas through coupling with auditory areas [10,11,29,30]. The mechanisms of auditory–motor interactions during speech (and auditory) perception, however, are not fully understood, and the involvement of neural oscillations is controversially discussed [31–33]. A crucial characteristic of neural oscillations is that they reflect endogenous brain rhythms observed in the absence of external stimulation. Here, we put a neural oscillatory framework of auditory–motor interactions to a rigorous test by investigating whether individuals’ peak frequencies of endogenous rhythms of auditory and motor cortical areas (observed during resting-state) and their coupling strength predict auditory cortical tracking during speech comprehension.

Oscillatory speech perception models propose that endogenous theta rhythms in auditory cortex synchronize to temporal fluctuations in the speech signal (i.e., the amplitude envelope) and thereby segment it into syllable-sized chunks [24–27,34,35]. This brain-to-speech alignment is most pronounced in the theta range (~5 Hz), declining at higher syllabic rates [36], as speech comprehension also decreases (for non-speech see: [37–41]). Given the observation of endogenous theta brain rhythms in auditory cortex [42–44] and the optimal speech processing in this range, the preferred frequencies of neuronal populations in auditory cortex in the theta range have been proposed to constrain the temporal granularity of perception [26,34,44–48]. The hypothesized connection between endogenous theta rhythms and speech processing, a fundamental aspect of oscillatory theories, however, has been rarely investigated directly, i.e., by relating endogenous and functional processing within individuals [49,50]. Such research may be hindered by the difficulty of quantifying individual endogenous brain rhythms in auditory cortex in the theta range [46]. While it is relatively straightforward to identify individual peak frequencies in some cases (e.g., posterior alpha peak frequency) because those can be detected in average power spectra, this is typically more challenging in most other frequency bands and areas. We here use spectral-fingerprinting of resting-state brain activity [46] to comprehensively identify individual auditory spectral peaks in a completely data-driven way. Specifically, we use time-resolved clustering procedures of normalized data that result in several frequency peaks per individual, which reflect endogenous rhythmic activity.

According to recent work, not only auditory processing but also auditory–motor coupling has an optimal range (~4.5 Hz [29,51]), hinting at the involvement of motor cortex oscillations in the auditory–motor interaction. Computational modeling supports an oscillatory account of auditory–motor coupling during speech perception, assuming oscillators with slightly higher preferred frequencies in the auditory than the motor system [3,29]. Behavioral and computational studies have shown a relation between preferred (spontaneous) motor production rates and the ability to synchronize to sound at different rates in music and sound sequences (for a neural approach see: [52–55]). In the context of dynamical systems theory, the Arnold tongue phenomenon describes how an oscillator’s ability to track a stimulus depends on its preferred frequency and the frequency and intensity of an external stimulation [56]. However, how this phenomenon transfers to the preferred frequencies of two oscillatory systems (auditory and motor) and their coupling strength to shape our ability to track speech acoustics is unknown. In a recent behavioral study [41], we approached this question, demonstrating superior speech comprehension in individuals with higher auditory–motor synchronization and higher preferred motor rates. The spontaneous speech synchronization (SSS) test was used to behaviorally estimate auditory–motor cortex coupling strength [57]. The preferred endogenous motor rates were quantified by the spontaneous rhythmic speech production rates of an individual, like the spontaneous tapping measure typically used in the field of dynamic attending [53,58–62].

Here, we use a behavioral and MEG approach in a relatively large sample, combining spectral fingerprinting of individuals’ MEG resting-state brain activity with MEG data recorded during a speech intelligibility task (see Fig 3A). As a speech intelligibility task, we used a sentence repetition task to estimate individual’s speech comprehension performance. Additional behavioral tasks (Fig 3A, 3C, and 3E; and the digit-span to access working memory capacity) were conducted. First, we aim to test the neural oscillatory approach of speech segmentation by quantifying predictive effects of individual endogenous theta frequencies in auditory cortex on speech tracking and comprehension. Second, to characterize the conjectured neural oscillatory approach of auditory–motor interactions, we expect predictive effects of individual theta frequencies in motor cortices and of auditory–motor coupling strength. Our behavioral findings replicate and extend our previous behavioral study [41], by showing that higher preferred auditory rates and stronger auditory–motor synchronization predict speech comprehension. In contrast, higher spontaneous speech motor production rates only predicted comprehension in parts of the population (high audio-motor synchronizers), with low synchronizers instead showing predictive effects of their working memory capacity. No detectable relationship between the neural data and behavioral readouts was observed. Importantly, the neural results show that only in individuals with high auditory–motor synchronization profiles, the endogenous frequencies of cortical speech motor areas in SMA and IFG predicted speech tracking in auditory cortices, STG and HG, respectively. In contrast, endogenous auditory frequencies of STG (but not HG) were predictive of speech tracking across the population. Our study provides evidence for a neural oscillatory account that highlights alternative speech processing routes across individuals.

PNG larger image

TIFF original image Download: Fig 1. Experimental paradigms and behavioral data. A. We measured auditory–motor synchronization using the (explicit) SSS-test [63], wherein participants whisper the syllable/te/ (green speech signal) while listening to a syllable train (purple speech signal) embedded in noise (grey signal). They are instructed to synchronize their motor output to the auditory input. B. The histogram illustrates audio-motor synchronization (measured as PLV). Colored lines represent fitted normal distribution, obtained by a Gaussian mixture model. The black box indicates participants that were excluded from the group analysis, as no unique group affiliation could be determined (i.e., probability of 0.5). C. To quantify the motor rate, we conducted a speech motor production task. Upon being prompted with a question/statement, participants spoke freely (green speech signal) for 30 s. To control for auditory feedback, noise was presented while participants spoke (grey signal). E. In the preferred auditory rate preference task, participants indicated which speech rates they liked better. To this end, two versions of a sentence, differing in their syllabic rates, were presented on each trial and participants chose the sentence (i.e., syllabic rate) they preferred. D, F. Density and dot plots visualize the motor production rate (D) and preferred auditory rate (F) of participants. All icons, the “Headphones” by Dong Gyu Yang, “speaking” by Ainul Muttaqin and the “Click” by Wahyu Prihantoro, are taken from The Noun Project used under the Creative Commons Attribution 3.0 license https://creativecommons.org/licenses/by/3.0/. The underlying numerical data shown in B, D, F are available at https://doi.org/10.17605/OSF.IO/SNDPE. https://doi.org/10.1371/journal.pbio.3003924.g001

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3003924

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/