Figure 1. Illustration of the Inclusion of Other in the Self (IOS) Scale
Words of forgiveness and complaint. To examine an individual’s reactive attitude toward the wrongdoers, twelve words with forgiveness or complaint attribution (six words for each) were selected before the study, the details are as follows. First, 100 two-character Chinese words (40/40 related to forgiveness/complaint, 20 neutral filling stimuli) were collected from several thesauruses. Second, another thirty-one homogeneous college students were recruited to evaluate the forgiveness degree and the familiarity on a 9-point Likert scale (9 being very similar to forgiveness, very familiar; 1 being very opposite of forgiveness, very unfamiliar, respectively) of the all random-order words. Finally, 6/6 words representing forgiveness/complaint were reserved according to the ranking of the above indicators. Independent t -test results of the words showed that the forgive-label words were significantly higher than the complain-label words in forgiveness degree (t (10) = 27.441, p < 0.001; forgive- vs. complain-label = 7.45 ± 0.13 vs. 2.15 ± 0.15). And there was no significant difference between the forgive- and complain-label words in familiarity (t (10) = 1.376, p = 0.199; forgive- vs. complain-label = 8.02 ± 0.11 vs. 7.83 ± 0.07). Detailed results are listed in Table 1.
Noise for inducing the feeling of offended . Unpleasant noise stimuli were edited by version 3.0.0 of Audacity(R)11Audacity® software is copyright © 1999-2021 Audacity Team. Web site: https://audacityteam.org/. and delivered via stereo. High-intensity noise was classified as 80 dB, low-intensity noise was classified as 50 dB, and no noise indicates a relatively quiet state. These noise stimuli were proved to induce strong emotional responses and physical instincts without permanent damage to hearing or nervous systems (Chester & Lasko, 2019; Lasko & Chester, 2022).

Experiment procedures

The experiment consisted of the following three phases.
Phase 1: Measurement of interpersonal distance . Participants were required to take part in our experiment accompanied by their best friend (same gender). On the day of the experiment, the two real participants together with one sham stranger participant from our lab came and were first informed to complete IOS scales aimed at each other. They received introductions together and were then assigned to three separate rooms.
Phase 2: Offended experience induction . In the cover story, all three participants were instructed that this study was designed to investigate the processing of tacit understanding between peers via an online interactive task (an adapted version of the Taylor Aggression Paradigm). In brief, they would choose one from two poker cards respectively, and the congruous choice with the peer meant a tacit understanding; vice versa (Li et al., 2021; Zhu et al., 2019).
To induce offended experiences without suspicion, two roles were set in theonline interactive task , i.e., the passive and the active. The former would receive noise for the incongruous trials between peers, and the latter controlled the noise intensity (high/low). A rock-paper-scissors intranet program was used to assign roles, with the loser acting the passive role. In reality, the real participant would always lose and then acted as the passive one deservedly. Besides, the participant believed that both the friend and the stranger were successful in getting the active role, and soon afterward was punished with the noise intensity selected by the friend or the stranger in all incongruous trials. Nobody doubted the rule and no one reported unbearable discomfort with the noise.
Specifically, as shown in Figure 2A, each trial began with a fixation (0.2 s), followed by the presentation of a name-photo pairing of the friend or the stranger for 2.5 s to indicate who would be interacting with in this trial. At the same time, the opposite player would select the noise intensity for the passive player. Names and photos of all participants were collected beforehand. The photo size was 163 × 210 pixels, blue background (RGB: 35, 35, 255), adjusted by Adobe Photoshop CS6. Next, the participant was required to select one from two poker cards by pressing the “F” or “J” key to assess the tacit understanding. A “waiting” interval then lasted for 1, 1.5, or 2 s to create a more authentic sense of interaction. The noise intensity (high/low) selected by the opponent player was presented for 1 s subsequently. Finally, the preset outcome feedback and corresponding noise were displayed for 1 s. Eighty trials in total were included (40/40 trials: friend/stranger). The probability of incongruous feedback was 60% (48 trials in total), and the ratio of high-intensity noise chosen by the opponent player was 80% in both friend and stranger trials. The whole procedure would take about 10 min. To avoid any interference with the subsequent reactive attitude task,offended experience, and explicit forgiveness were measured at the end of the experiment, with a 9-point Likert scale (9 being highly offended, highly forgive; 1 being no offended, cannot forgive, respectively).
Phase 3: Measurement of reactive attitude. The formal task, i.e.,word identification test , was used to evaluate the reactive attitude toward the friend- and the stranger-wrongdoer. Four types of trials, i.e., Forg-Fri, Comp-Fri, Forg-Str, and Comp-Str, were randomly presented in this task (48 trials for each). As shown in Figure 2B, each trial began with a fixation (0.2~0.5 s), followed by a name-word combination for a maximum of 2 s, during which participants were instructed to judge whether the word is a synonym for forgiveness (e.g., “原谅”, “谅解”) or an antonym (e.g., “埋怨”, “指责”) by pressing the “F/J” keyboard. The button assignment was counterbalanced across participants. A blank screen appeared for 1 s at the end of each trial. The whole procedure would take about 10 min. In the preprocessing analysis, data with response accuracy higher than 80% were reserved for further analysis. Then the trials with reacting time of less than 450 ms were excluded from the subsequent analyses and reacting time for each error trial was replaced with the mean reacting time for that block plus a 400 ms penalty (Karpinski & Steinman, 2006). All stimuli were presented in the center of a 17-inch LCD screen with black backgrounds via E-prime 3.0 software (Psychology Software Tools, Sharpsburg, USA). The viewing angle was 3.0 × 3.5° with a distance of 60~70 cm from the participants’ eyes.