High post-movement parietal low-beta power during rhythmic tapping facilitates performance in a stop task

Abstract

Voluntary movements are followed by a post-movement electroencephalography (EEG) beta rebound, which increases with practice and confidence in a task. We hypothesized that greater beta modulation reflects less load on cognitive resources and may thus be associated with faster reactions to new stimuli. EEG was recorded in 17 healthy subjects during rhythmically paced index finger tapping. In a STOP condition, participants had to interrupt the upcoming tap in response to an auditory cue, which was timed such that stopping was successful only in ~ 50% of all trials. In a second condition, participants carried on tapping twice after the stop signal (CONTINUE condition). Thus the conditions were distinct in whether abrupt stopping was required as a second task. Modulation of 12-20 Hz power over motor and parietal areas developed with time on each trial and more so in the CONTINUE condition. Reduced modulation in the STOP condition went along with reduced negative mean asynchronies suggesting less confident anticipation of the timing of the next tap. Yet participants were more likely to stop when beta modulation prior to the stop cue was more pronounced. In the STOP condition, expectancy of the stop signal may have increased cognitive load during movement execution given that the task might have to be stopped abruptly. However, within this condition, stopping ability was increased if the preceding tap was followed by a relatively larger beta increase. Significant, albeit weak, correlations confirmed that increased post-movement beta power was associated with faster reactions to new stimuli, consistent with reduced cognitive load.

Keywords: finger tapping; motor inhibition; sensorimotor synchronization; stop signal.

Figure 1

Schematic of the CONTINUE and STOP condition in the upper and lower row respectively. Black rectangles depict metronome sound cues, grey ellipses represent taps. The bolt indicates the stop signal, a sound higher in pitch than the regular tones that was delivered at an individually fixed delay relative to the previous last regular tap. The dark grey rightmost ellipse in the lower panel would represent an unsuccessfully inhibited tap.

Figure 2

Behavioural data from the first subject. The upper panel shows pressure sensor data. The stop signal was delivered relative to the time of the finger touching the sensor when the pressure signal passed a threshold (time = 0 ms). For this subject, the stop signal delay time was 550 ms relative to 0 ms, marking the last regular tap. Black lines are trials where the tapping movement after the stop signal was successfully interrupted, grey lines are trials where the tap was not inhibited before touching the sensor. The markers around 0 ms represent the temporal offset between the sound and the tap (○ = stopped, × = failed to stop). The markers ○ and × are overlapping rather than separated showing that random fluctuations of the tap‐to‐sound offset were not crucial in determining stopping performance. The lower panel displays the extent of finger flexion as recorded by the goniometer. Note that successfully stopped trials frequently contained downward movement of the finger, which was interrupted timely enough to stop touching of the sensor.

Figure 3

Scatter plot of correlations between movement extent (x‐axis) and tap‐to‐sound offset (y‐axis). Subplots show individual participants. Plot titles denote Spearman's rho followed by its 95% bootstrapped confidence interval. Participants were less likely to stop when the last regular tap was relatively late with respect to the metronome sound. 12 of 17 subjects had significant correlations.

Figure 4

Temporal development of the % change in (A) 12–20 Hz and (B) 20–30 Hz power. The left and right column depict modulation in the electrodes C3 and Pz respectively. Data are aligned to taps as denoted in the legend showing one tap‐cycle within a – 100 : 600 ms window. The left, middle and right traces depict early (2–3), middle (3–4) and late (> 5) taps respectively. Downward arrows denote the location of the average power trough, and upward arrows denote the location of the average power peak, which was used to compute the modulation. In (A) the 12–20 Hz modulation developed in the CONTINUE condition (upper row) after the third tap both in C3 and Pz. In the STOP condition, modulation was strongly attenuated, particularly in Pz. In (B) the average 20–30 Hz modulation was again stronger in C3 than in Pz, and modulation increased after the third tap. Topoplots to the right show the distribution of negative t‐scores of the condition differences in modulation, which were significant in Pz, C3 and FCz for the low‐beta band.

Figure 5

EEG preceding the stop signal. (A) Power aligned to regular taps (time = 0) in the STOP condition. The black dashed line denotes the finger contact with the pressure sensor. Power was z‐transformed for each frequency within the time window displayed before being averaged across subjects for better visual display. (B) t‐Scores of power differences between successful and failed stops prior to the stop signal. Clusters surrounded by black outlines denote that power was significantly higher when participants interrupted their movement more successfully (movement extent threshold < 40%). (C) Topoplots show the distribution of t‐scores and mean differences. In locations marked with a star, 12–20 Hz beta was significantly higher prior to successful stops averaged within the window outlined by the dashed rectangle in (B) in C3. The channel location of C3 in the topoplots is highlighted with a black circle surrounding the star.

Figure 6

(A) 12–20 Hz beta power time course following the stop signal (vertical dashed line). Data were subdivided according to stopping performance (movement extent threshold < 40%: solid curve; > 40%: dashed curve). (B) Time courses of individual differences between the power of successful and unsuccessful stops subdivided as in (A). Data are aligned to the last regular tap (vertical dashed line) preceding the stop signal. The bold line denotes the mean difference. The bottom stepped line shows for each point in time the fraction of participants who had higher beta power prior to successful stops. (C) Data are aligned to the last regular tap and median split according to the amount of upward movement upMvmt (smaller upMvmt = solid curve; bigger upMvmt = dashed curve). (D) Data are aligned to the last regular tap and median split according to soundOffset (more negative, i.e., anticipatory soundOffset = solid curve; more positive, i.e., delayed soundOffset = dashed curve). Filled areas between lines indicate significant differences, which were cluster‐based multiple comparison corrected. Shaded error bars around curves denote standard errors. Note that stopping the finger taps was associated with a pattern of beta power modulation that was more similar to the CONTINUE condition; compare the solid trace of the successful stop beta profile in both C3 and Pz in A) with the CONTINUE beta profile in Figure 4A (right‐hand upper panels for Taps > 5).

Figure 7

Scatter plot of correlations between movement extent (x‐axis) and beta relative to baseline (y‐axis). Subplots show individual participants. For each participant, beta power yielding the maximum correlation (detected anywhere between 12–30 Hz and 200–500 ms after the last regular tap considering that optimal frequencies and time points may differ across subjects) is shown. Plot titles denote Spearman's rho followed by its 95% bootstrapped confidence interval. The second line denotes the correlation coefficient resulting from the partial correlation controlling for the first two components obtained by principal component analysis of the behavioural variables. 16 of 17 subjects (10 if partial correlations were considered) had significant correlations.

Figure 8

(A) t‐Scores of the contrast between power aligned to the stop signal (vertical dashed line) averaged across all STOP trials irrespective of stopping performance and the regular tap done before. (B) as (A) but computed on all CONTINUE trials of the first block instead. In clusters surrounded by black outlines, power differed significantly from power observed during the regular tap done before. Cluster‐based statistics were computed within 0 : 500 ms after the stop signal. Straight after the stop signal, power increased in low frequencies in all channels in both tasks. The cluster in the beta range denotes that beta power was lower relative to regular tapping. Topoplots show t‐scores from the average within the dashed rectangular windows: The low‐frequency power increase peaked fronto‐centrally, whereas the difference in beta was most pronounced over parietal and ipsilateral sites. (C) t‐Scores of power differences between more successful and less successful stops (movement extent threshold = 40%). Clusters denote that low‐frequency power was significantly higher when participants interrupted their tap before touching the pressure sensor. The corresponding topoplot shows that the t‐score was highest over contralateral M1. (D) Time courses of individual theta (3–8 Hz) power differences following the stop signal (vertical dashed line) corresponding to the difference depicted in (C). The bold line denotes the mean difference. The bottom stepped line shows for each point in time the fraction of participants who had higher theta power during successful stops.