Participants
Twenty-six young adults participated in the study (15 male, 11 female; mean age 23.2 ± 1.25 years). All participants were right-handed, as confirmed by an Edinburgh Handedness inventory (min–max 70–100; median 84.12). Participants were naïve as to the purpose of the experiment and were informed before the study that their privacy would be secured. Written informed consent was provided by the participants prior to the commencement of the experiment, and participants were debriefed following experiment completion.
The procedure of the study was approved by the Ethics Committee of Kyushu University. The study was conducted according to the principles of the Declaration of Helsinki.
Equipment
The study was conducted in an acoustically and electrically sealed room located in Kyushu University. A 64-channnel EEG was recorded using an EEG amplifier (Net Amps 200, EGI) with sensor-net (HCGSN-64, EGI). Electrode impedance was maintained to remain under 50 kΩ as suggested by the manufacturer. EEG data were filtered in real time by the amplifier hardware, with the filter set to 0.01 Hz high-pass and 200 Hz low-pass, and digitized at 500 Hz. Stimuli were delivered with Presentation Ver. 20.0 (NBS Inc.) and an LCD display (E2351VR-BN, LG Electronics) refreshing at 60 Hz. Participants reaction (finger lifting) was acquired by capacitance sensor (AD00019, Bit Trade One, LTD.) connected to the Presentation software.
Imitation-inhibition task
The task utilized in the current study was based on a task developed and used in previous research [42]. Participants were required to respond as soon as possible according to simultaneously displayed finger movement and instructions (congruent or incongruent; Fig. 1). The presented stimuli consisted of a short movie of a hand quickly lifting either an index or middle finger, and a surrounding colored frame (red or green) indicating a response by the congruent or incongruent finger. The assignment of colors to the tasks (congruent or incongruent) was counter balanced among the participants. For example, if the index finger was lifted in the stimuli and the colored frame indicated an incongruent response, the correct response for the participant was lifting their middle finger.
In addition to the imitation condition described above, a spatial condition, in which one of two black-dots moved upwards, was tested to take the spatial compatibility aspect into account. The order of the conditions was counter balanced.
The task was divided into three blocks, including a randomization of trials, in which half were congruent and the other half were incongruent. This was done so that participants could not predict the next task instruction (total 300 trials per condition; 75 trials each for congruent index finger, congruent-middle finger, incongruent index finger, and incongruent middle finger).
Behavioral measurements and analysis
Reaction time (RT) was calculated for each task instruction (congruent or incongruent) of each condition (imitation or spatial). In order to highlight the congruency-effect, ΔRT was calculated by subtracting the mean RT of congruent trials from the mean RT of incongruent trials for each condition.
Based on signal detection theory (SDT), task sensitivity index (d’), and response bias (C) were calculated as performance indices. SDT assumes that, in an uncertain condition, participants are active decision makers who make difficult perceptual judgements. The SDT index, rather than the percent of correct responses, provides effective solutions to differentiate participants’ response tendencies (bias) from the ability to detect and discriminate information (task sensitivity) [39,40,41]. We calculated sensitivity and bias by categorizing congruent success, congruent failure, incongruent failure, and incongruent success as hit, miss, false alarm, and correct rejection, respectively. A higher d’ indicates higher task performance. The response bias index (C) is zero when there is no bias at all. This variable would take a positive value when responses are biased toward incongruent responses, and vice versa. Formulas for d’ and C are the following:
$$ {d}^{\prime }={Z}_{\mathrm{hit}}-{Z}_{\mathrm{false}\ \mathrm{alarm}} $$
$$ C=-0.5\times \left[{Z}_{\mathrm{hit}}+{Z}_{\mathrm{false}\ \mathrm{alarm}}\right] $$
where, Zhit is the z-transformed hit rate [hit count/(hit count + miss count)] and Zfalse alarm is the z transformed false alarm rate [false alarm count/(correct rejection count + false alarm count)].
EEG measurements and analysis
As an index of automatic imitation, mu wave event-related desynchronization (ERD) measured around central sulcus was acquired. In addition to mu ERD, alpha ERD possibly reflecting the attention level was calculated from the occipital sites, in order to take alpha wave contamination to the mu ERD into account. All EEG preprocessing and analysis was carried out using EEGLAB 14.1.1b [43], which is an open source toolbox of MATLAB (MathWorks Inc.). Raw EEG data, following manual rejection of bad channels, were filtered by FIR band-pass filter (0.5–40 Hz; transition band width 1 Hz) and epoched according to stimuli onset. Furthermore, bad epochs found in the data were automatically rejected according to joint probability of the data (both single-channel and all-channel threshold were set to 3 S.D). Following this, all data were re-referenced to the average. While data were re-referenced, outer electrodes (e.g., facial electrodes) were excluded from calculations (channels 23, 55, and 61 ~ 64). Two participants were excluded from further analysis due to an insufficient number of epochs. Preprocessed data were then sent to infomax independent component analysis (ICA). Independent components representing eye-blinks or eye-movement were manually rejected based on the topographical map, frequency spectrum, and activation synchrony with electrooculography (EOG).
ERD was calculated by EEGLAB’s time-frequency analysis function. First, event-related spectrum perturbations were calculated by wavelet analysis, starting with 2 cycles and increasing by 0.5 cycles toward higher frequencies. Second, mu wave (8–13 Hz) power from 300 ms to 600 ms after the event onset was calculated for each participant in decibel, according to the baseline period, which was 200 ms to 0 ms prior to the event onset. Finally, ERDs from each channel were averaged across the regions of interest (ROIs) to improve the reliability of the data. The left central (LC) region, representing mu ERD, was covered by C3 and its neighboring channels (i.e., 15, 16, 20, 21, and 22). The mid occipital (MO) region, representing alpha ERD, was covered by Oz and its neighboring channels (i.e., 35, 37, and 39; Fig. 2). After calculating ERDs, a congruency effect for mu and alpha ERDs was calculated in the same manner as ΔRT.
Statistical analyses
In order to test the relationship between the task performance and the EQ-SQ scores, Spearman’s rank correlation was calculated. A Bonferroni correction was used to adjust the p value for multiple tests. To highlight the effect of task and personal cognitive traits, an additional mixed-design analysis of variance (ANOVA) was conducted on data that showed significant correlations. The factors included in ANOVAs were high and low groups of EQ or SQ, and indicated tasks (congruent and incongruent). High EQ or SQ groups contained the upper 33% of participants and low EQ or SQ groups consisted of the lower 33% of participants. An independent t test was conducted to test the shift of response bias. All results are reported using an α of p < 0.05. All statistical tests were conducted using R version 3.5.1 [44].