Left motor cortex contributes to auditory phonological discrimination

Abstract

Evidence suggests that the articulatory motor system contributes to speech perception in a context-dependent manner. This study tested 2 hypotheses using magnetoencephalography: (i) the motor cortex is involved in phonological processing, and (ii) it aids in compensating for speech-in-noise challenges. A total of 32 young adults performed a phonological discrimination task under 3 noise conditions while their brain activity was recorded using magnetoencephalography. We observed simultaneous activation in the left ventral primary motor cortex and bilateral posterior-superior temporal gyrus when participants correctly identified pairs of syllables. This activation was significantly more pronounced for phonologically different than identical syllable pairs. Notably, phonological differences were resolved more quickly in the left ventral primary motor cortex than in the left posterior-superior temporal gyrus. Conversely, the noise level did not modulate the activity in frontal motor regions and the involvement of the left ventral primary motor cortex in phonological discrimination was comparable across all noise conditions. Our results show that the ventral primary motor cortex is crucial for phonological processing but not for compensation in challenging listening conditions. Simultaneous activation of left ventral primary motor cortex and bilateral posterior-superior temporal gyrus supports an interactive model of speech perception, where auditory and motor regions shape perception. The ventral primary motor cortex may be involved in a predictive coding mechanism that influences auditory-phonetic processing.

Keywords: magnetoencephalography; speech motor; speech perception; speech production; speech-in-noise.

Fig. 1

Results of linear mixed models for A) accuracy and B) reaction time. The error bars represent standard errors of the mean. C) Box plot of overall sensitivity (d’). The higher scores indicate better discrimination ability. D) Box plot for overall response bias (c). A score of 0 indicates no bias. The negative values indicate a tendency to indicate identical syllable pairs, whereas the positive values indicate a tendency to indicate different syllable pairs.

Fig. 2

Sensor-level and source-level grand averages of the event-related fields. A) Butterfly plot of all sensors. Epochs are time-locked to the second syllable (S2) at t = 0 ms, with the first syllable (S1) onset at approximately t = 750 ms. B) Topographies at peak activity for the first (t = −250 ms) and second syllables (t = 400 ms). C) sLORETA maps depicting cortical source localization at peak activity for the first (t = −250 ms) and second (t = 400 ms) syllables.

Fig. 3

Differences in absolute source activity between the 2 extreme SNR conditions (SNR −3 dB vs. SNR +3 dB), focusing on the time window following the second syllable. The maps display peak differences at their respective time points, with t-test results overlaid on the source activity difference maps, highlighting only the significant differences. The results were adjusted for an FDR (P = 0.05) and a minimum duration threshold of 10 ms. In the maps, blue areas indicate a decrease in source activity as noise levels increase, while red areas show an increase in source activity with increasing noise levels.

Fig. 4

Differences in absolute source activity over time between different and identical syllable pairs, focusing on the time window following the second syllable. The statistical t-test results were overlaid on the source activity maps, highlighting only the significant differences. The results were corrected for an FDR (P = 0.05) with a minimum duration threshold of 10 ms. In the maps, blue areas indicate a decrease in source activity for different syllable pairs, while red areas show an increase in source activity for different syllable pairs.

Fig. 5

A) Time course of source activity by trial type for bilateral ROIs in the pSTG, M1, IFG. B) Difference in source activity between trial types for each ROI. Positive values indicate higher activity for different syllable pairs than for identical pairs.

Fig. 6

A) Results from coordinate-based ALE meta-analysis of speech production studies. The map shows areas of increased activation associated with speech production (left hemisphere only). The dot located more dorsally marks the peak activity in the precentral gyrus with MNI coordinates: x = −48, y = −10, and z = 40. The more ventrally positioned dot represents the peak activity related to phonological discrimination in M1 from the MEG analysis, with MNI coordinates: x = −51, y = −9, and z = 31. B) This panel illustrates source activity associated with button presses, regardless of the trial type and response accuracy. The peak activity was observed in the left dorsal precentral gyrus.

Fig. 7

Brain–behavior relationships (t-maps) between the difference in source activity between different and identical syllable pairs and A) sensitivity (d’) and B) response bias (c). Maps were thresholded at t = 1.7 (uncorrected at P = 0.05), with a cluster-extent threshold set at 10 vertices. The results did not survive FWE rate correction.