Neurophysiological evidence of efference copies to inner speech

Abstract

Efference copies refer to internal duplicates of movement-producing neural signals. Their primary function is to predict, and often suppress, the sensory consequences of willed movements. Efference copies have been almost exclusively investigated in the context of overt movements. The current electrophysiological study employed a novel design to show that inner speech - the silent production of words in one's mind - is also associated with an efference copy. Participants produced an inner phoneme at a precisely specified time, at which an audible phoneme was concurrently presented. The production of the inner phoneme resulted in electrophysiological suppression, but only if the content of the inner phoneme matched the content of the audible phoneme. These results demonstrate that inner speech - a purely mental action - is associated with an efference copy with detailed auditory properties. These findings suggest that inner speech may ultimately reflect a special type of overt speech.

Keywords: corollary discharge; covert speech; efference copy; human; inner speech; neuroscience; sensory attenuation.

Figure 1.. A schematic of the experimental protocol.

Participants were instructed to fixate their eyes on the central red fixation line (Panel A). After a delay (1–2 s), the green trigger line, which was presented on the far right-hand side of the screen, and visible in participants’ peripheral vision, began to move smoothly across the screen in a leftwards direction at a speed of 6.5°/s (Panel B), such that after 3.75 s the trigger line overlapped with the fixation line. At this precise moment, dubbed the ‘sound-time’, two events occurred simultaneously (Panel C). Firstly, the participant was asked to imagine themselves producing a pre-defined phoneme in inner speech (either /ba/ or /bi/ or no inner phoneme). Secondly, an audible phoneme (either /BA/ or /BI/), produced by a male speaker, was delivered to the participant’s headphones. In Match trials (Panel D, top, blue), the inner phoneme was congruent with the audible phoneme (e.g., inner phoneme: /ba/; audible phoneme: /BA/). In Mismatch trials (Panel D, middle, red), the inner phoneme was incongruent with the audible phoneme (e.g., inner phoneme: /bi/; audible phoneme: /BA/). In Passive trials (Panel D, bottom, black), the participant did not produce an inner phoneme. Following the sound-time, the trigger line continued to move past the fixation line for an additional 1 s. The trial was then complete and the participant was asked to rate how successfully they managed to follow the instructions on that trial, on a scale from 1 (Not at all successful) to 5 (Completely successful).

Figure 2.. Inner speech experiment: N1 component analysis.

(A) Waveforms showing the auditory-evoked potentials elicited by the audible phonemes in the Match condition (blue line), Mismatch condition (red line) and Passive condition (black line). The N1-component is labelled; the waveforms were averaged across electrodes FCz, Fz, and Cz, as these were the electrodes at which the N1 component was maximal. The waveforms are shown collapsed across audible phoneme (audible /BA/ and /BI/), and the waveforms for the Match and Mismatch conditions are shown collapsed across inner phoneme (inner /ba/ and /bi/). Voltage maps are plotted separately for each condition; white dots illustrate the electrodes used in the analysis. (B) Box-and-whiskers plots showing the amplitude of the N1 component elicited by the audible phonemes in the Match, Mismatch and Passive conditions. The edges of the boxes represent the top and bottom quartiles, the horizontal stripe represents the median, the cross represents the mean, the whiskers represent the 9th and 91st percentiles, and the colored dots represent the participants whose data fell outside the range defined by the whiskers. (C) Scatterplots showing the within-subjects difference scores (in terms of N1-amplitude) for the three contrasts-of-interest in the inner speech experiment; namely Match minus Mismatch, Match minus Passive, and Mismatch minus Passive. These difference scores were approximately normally distributed with no clear outliers. Each dot represents a single participant’s difference score. The horizontal bars represent the mean, and the error bars represent the 95% confidence interval.

Figure 3.. Inner speech experiment: P2 component analysis.

(A) Waveforms showing the auditory-evoked potentials elicited by the audible phonemes in the Match condition (blue line), Mismatch condition (red line), and Passive condition (black line). The P2-component is labelled; P2 amplitude was calculated as the average voltage in the 150–190 ms time-window. The waveforms were averaged across electrodes Cz, FCz, and CPz, as these were the electrodes at which the P2 component was maximal. Voltage maps are plotted separately for each condition; white dots illustrate the electrodes used in the analysis. (B) Box-and-whiskers plots showing the amplitude of the P2 component elicited by the audible phonemes in the Match, Mismatch, and Passive conditions. The edges of the boxes represent the top and bottom quartiles, the horizontal stripe represents the median, the cross represents the mean, the whiskers represent the 9th and 91st percentiles, and the colored dots represent the participants whose raw data fell outside the range defined by the whiskers. (C) Scatterplots showing the within-subjects difference scores (in terms of P2-amplitude) for the three contrasts-of-interest in the inner speech experiment; namely Match minus Mismatch, Match minus Passive, and Mismatch minus Passive. These difference scores were approximately normally distributed with no clear outliers. Each dot represents a single participant’s difference score. The horizontal bars represent the mean, and the error bars represent the 95% confidence interval.

Figure 4.. Inner speech experiment: P3 component analysis.

(A) Waveforms showing the auditory-evoked potentials elicited by the audible phonemes in the Match condition (blue line), Mismatch condition (red line), and Passive condition (black line). The P3-component is labelled; P3 amplitude was calculated as the average voltage in the 250–310 ms time-window. The waveforms were averaged across electrodes CPz, Cz, and Pz, as these were the electrodes at which the P3 component was maximal. Voltage maps are plotted separately for each condition; white dots illustrate the electrodes used in the analysis. (B) Box-and-whiskers plots showing the amplitude of the P3 component elicited by the audible phonemes in the Match, Mismatch, and Passive conditions. The edges of the boxes represent the top and bottom quartiles, the horizontal stripe represents the median, the cross represents the mean, the whiskers represent the 9th and 91st percentiles, and the colored dots represent the participants whose raw data fell outside the range defined by the whiskers. (C) Scatterplots showing the within-subjects difference scores (in terms of P3-amplitude) for the three contrasts-of-interest in the inner speech experiment; namely Match minus Mismatch, Match minus Passive, and Mismatch minus Passive. These difference scores were approximately normally distributed with no clear outliers. Each dot represents a single participant’s difference score. The horizontal bars represent the mean, and the error bars represent the 95% confidence interval.

Figure 5.. Overt speech experiment: N1 component analysis.

The experimental protocol for the overt speech experiment was identical to the inner speech experiment except that participants were required to overtly (as opposed to covertly) vocalize the phoneme at the sound time. (A) Uncorrected waveforms showing the auditory-evoked potentials elicited by the audible phonemes in the Match condition (blue line), Mismatch condition (red line), and Passive condition (black line) in the overt speech experiment. The waveform for the motor-control condition is also shown (green line: in this condition participants overtly vocalized a phoneme at the sound-time, but no audible phoneme was delivered). The N1-component is labelled; the waveforms were averaged across electrodes FCz, Fz, and Cz, as these were the electrodes at which the N1 component was maximal. The waveforms are shown collapsed across audible phoneme (audible /BA/ and /BI/), and the waveforms for the Match, Mismatch, and Motor Control conditions are shown collapsed across vocalized phoneme (overt /ba/ and /bi/). (B) Motor-corrected waveforms showing the auditory-evoked potentials elicited by the audible phonemes in the Match condition (blue line), Mismatch condition (red line), and Passive condition (black line) in the overt speech experiment. The motor-corrected waveforms were generated by subtracting the activity generated in the motor-control condition from each participant’s Match, Mismatch, and Passive waveforms. Voltage maps are plotted separately for each condition; white dots illustrate the electrodes used in the analysis. (C) Box-and-whiskers plots showing the amplitude of the N1 component elicited by the audible phonemes in the Match, Mismatch, and Passive conditions in the overt speech experiment, using motor-corrected data for the Match and Mismatch conditions. The edges of the boxes represent the top and bottom quartiles, the horizontal stripe represents the median, the cross represents the mean, the whiskers represent the 9th and 91st percentiles, and the colored dots represent the participants whose raw data fell outside the range defined by the whiskers. (D) Scatterplots showing the within-subject difference scores (in terms of N1-amplitude) for the three contrasts-of-interest in the overt speech experiment; namely Match minus Mismatch, Match minus Passive, and Mismatch minus Passive. These difference scores were approximately normally distributed with no clear outliers. Each dot represents a single participant’s difference score. The horizontal bars represent the mean, and the error bars represent the 95% confidence interval.

Figure 6.. Auditory-evoked potentials elicited by nine different phonemes; namely: /BA/, /BI/, /DA/, /DI/, /GA/, /KI/, /PA/, /PI/, and/TI/.

Each phoneme was ~200 ms in duration, presented at ~70 dB SPL, and was produced by the same male speaker. Each phoneme was presented 90 times; the presentation order was randomized. Participants were instructed to simply sit quietly and listen to the phonemes. Of the nine different phonemes, /BA/ and /BI/ were judged to be most similar in terms of their amplitude and overall shape, and hence these phonemes were chosen to be used as the audible phonemes in both the inner and overt speech experiments.

Author response image 1.

Author response image 2.

Author response image 3.