Figure 2. Experimental procedures and behavioral results. A,
Illustration of one trial in the procedure of offended experience
induction (i.e., an adapted version of the Taylor Aggression Paradigm).
Due to copyright, the name and photo in the sample image are replaced by
one of the authors of this study (S. Li). B, Illustration of one trial
in the word identification test. The critical events for EEG analysis
are marked with a dashed line. C, Correlation between offended
experience and explicit forgiveness in the friend and the stranger
condition. Results showed significant negative correlations in the
stranger (but not in the friend) condition. D, Behavioral results in the
reactive attitude task. Forg-Fri, forgive friend condition; Comp-Fri,
complain friend condition; Forg-Str, forgive stranger condition;
Comp-Str, complain stranger condition. *p < .05.
EEG recording
The EEG signal was recorded from 64 Ag/AgCI electrodes mounted in an
elastic cap (NeuroScan Inc., EI Paso, Texas, USA) sampled at 1000 Hz,
with a bandpass filter of 0.01~100 Hz online. Impedance
was maintained below 5 kΩ. Recording electrodes were referenced to the
left mastoid (M1). The vertical electrooculogram (VEOG) signal was
measured both above and below the left eye. The horizontal
electrooculogram (HEOG) signal was measured on both sides at the
external canthi.
EEG pre-processing
Raw EEG data were preprocessed and analyzed in Python using MNE tools
(mne.tools/dev/index.html): data were first downsampled to 500 Hz, and
were then re-referenced by whole-brain average reference (the common
average). All channels were filtered at 0.1~30 Hz
(zero-phase, 24 dB/Oct) with a 50-Hz notch filter offline. An
independent component analysis (ICA) procedure was then applied to the
EEG signals to identify components related to eye blinks and horizontal
eye movements automatically by a python script (based on
mne.preprocessing.ICA). Trials
contaminated with artifacts greater than ± 150 μV were rejected. For the
stimuli (marked with dash line in Figure 2B), EEG epochs were extracted
using a time window of 200 ms before the stimuli onset to 800 ms post
after, and all Epochs were baseline corrected using the pre-stimulus
time interval. EEG segments were then averaged for further ERP analysis.
In this study, the ERP components were interested in the frontal cortex.
Channels in the frontal area were picked for ERP analysis (F1, F2, Fz,
FC1, FC2, FCz). We selected the N2 peak in the window of
300~350 ms (Nieuwenhuis et al., 2004; Suwazono et al.,
2000) and the P3 component between 380~420 ms (Mulert et
al., 2004).
The event-related oscillations (ERO) were analyzed for comparing the
different averaged spectral power under two variables
(interpersonal distance and attitude ) for alpha bands
(8~13 Hz). As suggested, the ERO was averaged at the
individual level (Herrmann et al., 2014). The ERPs in the last step for
each participant were used to calculate the ERO. The time-frequency
decomposition of ERO used the Morlet wavelet transform (based on
mne.time_frequency.Morlet; Bajaj & Pachori, 2014). The ERO was
computed for all data channels between 4 and 20 Hz and Morlet wavelet
analysis was applied to each epoch to measure the amount and phase of
the data in each successive time window, beginning with a 2-cycle
wavelet (Schneider & Maguire, 2018). The power spectral were normalized
(i.e., with a mean of zero). We also calculated the value of FAA, i.e.,
the difference between the counterparts of right- and left-hemispheric
electrodes (AF4-AF3, F2-F1, F4-F3, F6-F5, F8-F7, FP2-FP1, FC2-FC1,
FC4-FC3, and FC6-FC5; see Boksem et al., 2012; Galang & Obhi, 2019).
Statistics
Statistical analysis was performed using SPSS Statistics 26.0 (IBM,
Somers, USA). Descriptive data are presented as mean ± standard error
(M ± SE ) unless otherwise mentioned. In the reactive attitude
task, repeated-measures ANOVAs were performed on the reaction time and
EEG/ERP indexes, with interpersonal distance (friend or stranger)
and attitude (forgive or complain) as the within-subject factors.
Pairedt -test was used to examine the difference between the friend and
the stranger in the interpersonal distance , offended
experience , and explicit forgiveness. The Greenhouse-Geisser
correction for the ANOVA tests was used whenever appropriate.
Results
Manipulation check: interpersonal distance and offended
experience
The IOS scores were used to evaluate the interpersonal distance. A
paired t -test showed that the IOS scores of friends were
significantly higher than strangers (t (36) = 21.182, p< 0.001, Cohen’s d = 3.57; friend vs. stranger = 4.89 ±
0.17 vs. 1.16 ± 0.07).
The self-reported offended experience was used to evaluate the offended
feelings induced by the two wrongdoers (i.e., friends and strangers). A
paired t -test showed no significant difference, that is,
participants felt similarly offended experience by the friend and the
stranger (t (36) = -1.79, p = 0.082, Cohen’s d =
0.30; friend vs. stranger = 4.68 ± 0.31 vs. 5.30 ± 0.23).
Subjective rating: explicit
forgiveness
A
paired t -test was performed to examine the differences in
explicit forgiveness towards the two wrongdoers. The scores for friends
(8.54 ± 0.14) were significantly higher than those for strangers (7.70 ±
0.23; t (36) = 3.68, p = 0.001, Cohen’s d = 0.62).
A two-tailed Pearson correlation was performed between the offended
experience and explicit forgiveness in the friend and the stranger
condition, respectively. Results (Figure 2C) showed negative
correlations in the stranger condition (r = -0.35, p =
0.037), that is, the more offended by the stranger, the less to forgive.
However, this correlation was not significant in the friend condition
(r = -0.20, p = 0.243).
Reaction time: attitude for
wrongdoers
Reaction times of the word identification were examined utilizing
repeated-measures ANOVA (interpersonal distance ×attitude , Figure 2D). The most important finding was the two-way
interaction between interpersonal distance × attitude ,F (1,36) = 4.17, p = .049, \(\eta_{P}^{2}\) = 0.104. Simple
effect analysis showed that while in the friend condition, responses to
forgive-label words were faster than to complain-label words
(F (1,36) = 10.63, p = .002, \(\eta_{P}^{2}\) = 0.228;
789.70 ± 18.74 ms and 825.13 ± 18.76 ms for the forgive- and
complain-label words, respectively), no such significant difference was
observed in the stranger condition (F (1,36) = 0.52, p =
.475, \(\eta_{P}^{2}\) = 0.014; 825.39 ± 19.54 ms and 832.73 ± 17.47 ms
for the forgive- and complain-label words, respectively). The main
effect of interpersonal distance was significant (F (1,36)
= 14.99, p < .001, \(\eta_{P}^{2}\) = 0.294), with
participants reacting faster in the friend (807.42 ± 17.94 ms) than in
the stranger condition (829.06 ± 18.83 ms). The main effect ofattitude was significant (F (1,36) = 7.22, p = .011,\(\eta_{P}^{2}\) = 0.167), with participants reacting faster to forgive-
(807.55 ± 18.58 ms) than to complain-label words (828.93 ± 18.65 ms).
Detailed statistics of simple effect analysis are listed in Table 2.
ERP results
N2 amplitude. Repeated-measures ANOVA showed a significant main
effect of attitude (F (1,36) = 9.79, p = .003,\(\eta_{P}^{2}\) = 0.214, Figure 3A, B, C), with increased N2 amplitude
when participants react to forgive- (-2.17 ± 0.27 μ V) than to
complain-label words (-1.75 ± 0.28 μ V). No significant main
effect of interpersonal distance (F (1,36) = 3.11, p= .086, \(\eta_{P}^{2}\) = 0.080) and interaction effect (F (1,36)
= 1.11, p = .299, \(\eta_{P}^{2}\) = 0.030) were observed.
P3 amplitude. Repeated-measures ANOVA showed a significant
two-way interaction between interpersonal distance ×attitude (F (1,36) = 4.35, p = .044,\(\eta_{P}^{2}\) = 0.108; Figure 3A, B, D). Simple effect analysis
revealed that while no significant differences were found between
forgive- and complain-label words in the friend condition
(F (1,36) = 0.58, p = .451, \(\eta_{P}^{2}\) = 0.016;
forgive- and complain-label words: -0.01 ± 0.14 μ V and 0.09 ±
0.14 μ V), participants had decreased P3 amplitude when reacting
to forgive- than complain-label words in the stranger condition
(F (1,36) = 14.13, p = .001, \(\eta_{P}^{2}\) = 0.282;
forgive- and complain-label words: -0.18 ± 0.15 μ V and 0.20 ±
0.13 μ V). The main effect of attitude was significant
(F (1,36) = 5.92, p = .020, \(\eta_{P}^{2}\) = 0.141),
participants had increased P3 amplitude when reacting to complain (0.14
± 0.13 μ V) than to forgive (-0.10 ± 0.13 μ V). No
significant main effect of interpersonal distance was observed
(F (1,36) = 0.06, p = .802, \(\eta_{P}^{2}\) = 0.002).
Detailed statistics of simple effect analysis are listed in Table 2.