Sensory-motor networks involved in speech production and motor control: an fMRI study

Abstract

Speaking is one of the most complex motor behaviors developed to facilitate human communication. The underlying neural mechanisms of speech involve sensory-motor interactions that incorporate feedback information for online monitoring and control of produced speech sounds. In the present study, we adopted an auditory feedback pitch perturbation paradigm and combined it with functional magnetic resonance imaging (fMRI) recordings in order to identify brain areas involved in speech production and motor control. Subjects underwent fMRI scanning while they produced a steady vowel sound /a/ (speaking) or listened to the playback of their own vowel production (playback). During each condition, the auditory feedback from vowel production was either normal (no perturbation) or perturbed by an upward (+600 cents) pitch-shift stimulus randomly. Analysis of BOLD responses during speaking (with and without shift) vs. rest revealed activation of a complex network including bilateral superior temporal gyrus (STG), Heschl's gyrus, precentral gyrus, supplementary motor area (SMA), Rolandic operculum, postcentral gyrus and right inferior frontal gyrus (IFG). Performance correlation analysis showed that the subjects produced compensatory vocal responses that significantly correlated with BOLD response increases in bilateral STG and left precentral gyrus. However, during playback, the activation network was limited to cortical auditory areas including bilateral STG and Heschl's gyrus. Moreover, the contrast between speaking vs. playback highlighted a distinct functional network that included bilateral precentral gyrus, SMA, IFG, postcentral gyrus and insula. These findings suggest that speech motor control involves feedback error detection in sensory (e.g. auditory) cortices that subsequently activate motor-related areas for the adjustment of speech parameters during speaking.

Keywords: Auditory feedback; Pitch perturbation; Sensory–motor integration; Speech motor control; fMRI.

Figure 1

Experimental design for a single trial of volume acquisition using sparse-sampling method during a) speaking and b) playback pitch shift conditions. During speaking, subjects were instructed to maintain a steady production of the vowel sound /a/ following the onset of a visual cue lasting for 5 s. In a randomized order, speech feedback across trials was either unaltered or altered by a pitch shift stimulus beginning 750 ms following the speaking onset. Each speaking trial was immediately followed by playback of the same speech produced in the previous trial. fMRI volumes were acquired 5 s after the onset of visual cues in each trial with parameters TR=10000 ms, TE=30 ms and TA=2650 ms.

Figure 2

a) Overlaid time course of the magnitude of the grand-averaged vocal responses to no shift (red line) and +600 cents pitch shift stimuli (black line). Solid lines and highlighted areas show the mean and standard error of the vocal responses, respectively. b) Bar plot representation of the absolute value of the mean of the grand-averaged vocal response magnitudes during no shift and pitch shift stimulus conditions.

Figure 3

BOLD responses for contrasts between four different experimental conditions compared with rest. Responses are shown for the group (n = 8) from a random effects analysis (p<0.05, FWE corrected). The color map corresponds to the calculated T-values.

Figure 4

BOLD responses for the positive effect of condition (speaking > playback) for shift (a and b) and no shift (c and d). Responses are shown for the group (n = 8) from a random effects analysis (p<0.05, FWE-corrected). The color map corresponds to the calculated T-values.

Figure 5

BOLD responses for the positive effect of stimulus (shift > no shift) for a) speaking and b) playback condition. Responses are shown for the patient group subjects (n = 8) from a random effects analysis (p<0.001, uncorrected). The color map corresponds to the calculated T-values.

Figure 6

Performance correlation analysis between the absolute magnitude of the behavioral vocal responses and the percentage change in BOLD activation in response to pitch shift stimulus in bilateral STG and left precentral gyrus.