Introduction

Fairness is an important concept in human social behavior and a crucial factor in social stability (Liu, Hu, Shi, & Mai, 2020; Zhou & Wu, 2011). Numerous experimental studies have demonstrated that people who tend to pursue fairness, demand a fair distribution of wealth, are willing to punish individuals who behave unfairly (Batson & Shaw, 1991; Camerer, 2011; Fehr & Schmidt, 2006; Waal, 2008) and may even loathe those who take advantage of others (de Hooge, Nelissen, Breugelmans, & Zeelenberg, 2011).

The Ultimatum Game (UG) which has been widely used to study human fairness (Ma, Hu, Jiang, & Meng, 2015; Oosterbeek, Sloof, & Kuilen, 2004; Peterburs et al., 2017; Qu, Wang, & Yunyun, 2013; Sanfey, Rilling, Aronson, Nystrom, & Cohen, 2003; Wu, Yuqin, Dijk, Leliveld, & Zhou, 2011b), confirms the hypothesis that fairness is affected by different social roles (Cote, Piff, & Willer, 2012; Wu, Leliveld, & Zhou, 2011a), because human beings are naturally with social attributes (Hirst & Woolley, 1982). A typical UG involves two players: proposer and responder. The proposer makes an offer to divide a sum of money between the two players. The responder decides to accept or reject the offer. If the offer is accepted, the money is divided according to the division suggested by the proposer, but if the responder rejects the proposal, neither party receives anything (Güth, Schmittberger, & Schwarze, 1982). Yu et al. (2015) extended such a typical UG paradigm to explore how social distance affects subjects’ perception of fairness and found that offers from strangers elicit more lengthy consideration than offers from friends (Yu, Hu, & Zhang, 2015). A similar study conducted by Wu et al. (2011) confirms that social distance affects the way fairness is perceived and that opinions about fairness are strongly socially dependent (Wu, Leliveld, et al., 2011a). However, these studies only explore the impact of two broad interaction partner identities (friend vs. stranger) on fairness decision-making. Many other characteristics, such as strangers’ level of economic neediness, may affect decision-making (Cote, House, & Willer, 2015, Liu et al., 2020). Previous studies on economic neediness are mainly related to altruistic behavior (Bereczkei, Birkás, & Kerekes, 2009; Lamm, Batson, & Decety, 2007; Liu, Hu, Shi, & Mai, 2018), whereas the perception of fairness has been shown to be associated with prosocial emotions (Reuben & Winden, 2010). Our hypothesis is that the degree of economic neediness is one of the key factors that influence fairness perception. However, to the best of our knowledge, there is no study directly focused on this issue.

Studies of people in need have reported that the perceived welfare of a person in need can elicit empathic concern, which is described as state empathy (Batson & Ahmad, 2001; Stocks, Lishner, Waits, & Downum, 2011; Woltin, Corneille, Yzerbyt, & Förster, 2011). Additionally, it has been demonstrated that empathy can lead to fairness (Page & Nowak, 2002). We suspected that the degree of neediness might influence the responders’ fairness perception by modulating the proposers’ empathic concern. Accordingly, we raised two scientific issues: 1) how the degree of neediness influences fairness perception in an asset division situation and its underlying neural basis; 2) whether and how empathic concern plays a role in the way the degree of neediness impacts fairness perception. An important academic contribution presented by the Pan team is that the neuromanagement method could be used to measure accurately the emotion in individual and group decision-making process (Pan et al., 2019; Wang et al., 2019). Thus, to address our raised questions, we employed event-related potentials (ERPs) to measured subjects’ brain responses to the unfair/fair offer and examined the change of empathic concern based on whether proposers are in need or not.

As stated above, altruistic behavior can be induced when observing others in a difficult situation, and fairness perception was proved to be highly situation-dependent (Wu, Leliveld, et al., 2011a; Yu et al., 2015). As such, we supposed participants might be more tolerant of unfair offers from proposers who are economically needy than from those who are not and would be reflected in higher acceptance rates of unfair offers. Reaction time is another element in decision-making (offer acceptance or rejection). As people are naturally inclined to pursue fairness and longer reaction times were associated with higher rejection rates (Mussel, Göritz, & Hewig, 2013), we expected participants might react faster for fair offers, hence shorter reaction times. However, as to the proposer effect on reaction time, it is difficult for us to deduce whose proposal the subjects would spend longer considering. Thus, we focus on intrinsic brain responses. In previous ERPs studies of the UG (Boksem & Cremer, 2009; Hewig et al., 2011; Polezzi et al., 2008; Van der Veen & Sahibdin, 2011; Wang, Li, Li, Wei, & Li, 2016; Wu, Leliveld, et al., 2011), they found a negative peak at approximately 200 ms in the offer presentation stage. Some researchers named it MFN (medial frontal negativity) and others named it FRN (feedback related negativity), which mean the same component (Glazer, Kelley, Pornpattananangkul, Mittal, & Nusslock, 2018; Liu et al., 2018; Pornpattananangkul, Nadig, Heidinger, Walden, & Nusslock, 2017). In the current study, we use the term MFN. Besides, P300 is another commonly examined components. Below, we detail our specific hypotheses for MFN and P300.

The MFN, sometimes also called FRN, originates in the frontal-central regions and peaks at 200–350 ms following the onset of stimuli (Gehring & Willoughby, 2002; Holroyd & Coles, 2002). MFN was detected by ERP source localization and fMRI studies, indicating that the anterior cingulate cortex (ACC) is the brain region mainly involved. According to reinforcement learning theory, ACC activity is modulated by dopamine signals from the midbrain, where positive and negative prediction errors are coded. Specifically, negative prediction errors induced by unfavorable or negative outcomes initiate phasic decreases in dopamine inputs and result in increased ACC activity, which is reflected as a more negative MFN component (Gehring & Willoughby, 2002; Ma, Meng, Wang, & Shen, 2014; Wang et al., 2016; Yeung & Sanfey, 2004). MFN responses elicited by a proposer’s offer have been widely identified in UG studies, which suggests a violation of social norms (i.e., unfair offer) could generate larger negative MFN than compliance with social norms (i.e., fair offer) (Boksem & Cremer, 2009; Hewig et al., 2011; Hu & Mai, 2021; Polezzi et al., 2008; Van der Veen & Sahibdin, 2011). Furthermore, multiple researchers (Campanhã, Minati, Fregni, & Boggio, 2011; Ma et al., 2015; Qu et al., 2013) also have found that the proposer condition (in-need or not-in-need) can modulate individuals’ fairness perception, causing differentiated MFN (or FRN) responses to unfair/fair offers. For instance, Qu et al. (2013) adapted a UG task to examine how social exclusion affects responders’ fairness perception and found that, compared with the excluders’ offers, includers’ offers weaken responders’ fairness perception, reflected in a smaller d-FRN effect (FRN difference between unfair and fair offers) (Qu et al., 2013). In this experiment, taking into account the impact of economic neediness, responders’ fairness perception toward offers from proposers in need may be attenuated, leading to a smaller MFN difference effect (MFN difference between unfair and fair offers) than with offers from proposers not in need.

The P300 component has positive peaks in the 300–600-ms period after the presentation of stimuli. Its amplitude is commonly maximal at the centroparietal electrodes and is larger for outcomes with high arousal levels. Some researchers have posited that the P300 component is related to motivational/affective salience (Nieuwenhuis, Aston-Jones, & Cohen, 2005), the allocation of cognitive resources and the processing of attentional distribution (Hu, Xu, & Mai, 2017; Polich, 1987, 2007; Yang, Tang, Gu, Luo, & Luo, 2014). Research shows that P300 amplitude can be affected by social roles in decision-making. For instance, observing one’s own reward outcome would induce larger P300 than observing that of strangers (Jin, Wang, Liu, Pan, & DongLyu, 2020; Ma et al., 2011). These researchers suggest that high motivational stimuli (i.e., one’s own reward) might evoke more positive P300 than low motivational stimuli (i.e., strangers’ reward). According to our behavioral hypothesis, the perceived neediness might decrease the fairness perception, making subjects insensitive to their proposals. Then, the motivational/affective salience of the proposals might be reduced, which is reflected in the smaller P300. As such, we hypothesized that a smaller P300 might be observed in the proposer-in-need condition, compared with the proposer-not-in-need condition.

As to the cognitive empathy effect, we expected it might be reflected in the correlation with P300 amplitude, because P300 reflects the top-down control system regulation of cognitive resources in the late stage (Polich, 2007). More deeply, as stated above, the induced empathic concern would be larger in the proposer-in-need condition than in the proposer-not-in-need condition. Thus, three possible situations would cause the difference in empathic concern between these two conditions. First, perceived neediness would increase empathic concern in the proposer-in-need condition, while remaining unchanged in its opposite condition, which can be reflected in a negative P300 correlation only in the proposer-in-need condition. The second possible situation might be that individual’s empathic concern remained unchanged in the proposer-in-need condition but decreased in the proposer-not-in-need condition, which can be reflected in a positive P300 correlation only in the proposer-not-in-need condition. Third, empathic concern might increase the proposer-in-need condition and decrease the proposer-not-in-need condition, which could be reflected in a reverse correlation with P300 in both conditions.

To sum up, our primary goal was to examine why and how the degree of neediness influences the perception of fairness. We hypothesized that the perceived neediness might influence fairness perception and be reflected in the behavioral and ERPs results of (un)fair offers in the UG. Furthermore, concerning the internal mechanism, we supposed that empathic concern might play a central role in this phenomenon.

Materials and Methods

Participants

To estimate the sample size, we conducted a priori power analysis for 2 (proposer type: in need vs. not-in-need) × 2 (offer type: fair vs. unfair) within-subjects repeated-measures analysis of variance (ANOVA), by using G*Power 3.1 software (Faul, Erdfelder, Lang, & Buchner, 2007). Vazire (2016) advises researchers to ensure that the power value does not fall below 80% (Vazire, 2016). Thus, a sample of 14 participants was required to achieve a power of 0.80, with parameters that included an expected effect size of 0.25(f), an alpha value of 0.05, a default within-subjects measurement correlation of 0.5, and a non-sphericity correlation value (ƹ) of 1. This result means that the number of subjects in this experiment must be greater than 14, and the more the better.

In total, due to cost and time restrictions, we finally recruited 24 (12 females) healthy graduate or undergraduate students, with a mean age (18-25 years) of 20.79 years, S.D. = 2.00 from Ningbo University to participate in the current study. All of them were native Chinese speakers, right-handed, had normal or corrected-to-normal vision, and did not have any history of neurological or mental disease. Written, informed consent was provided before the formal experiment. The experiment was approved by the school internal review committee. It was in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards (Association, 2014).

Materials

A modified version of UG was employed in the current study. In this study, participants played the responder role, deciding whether to accept or reject the offer from proposers who came from a middle school. Proposers gave unfair offers (U1:1, 9 and U2:2, 8) and fair offers (U4:4, 6; and U5:5, 5). The U3 offer (3,7) was an additional distribution to induce variance in the set of offers so that participants were not faced only with unfair and fair splits on all trials. The U3 offer was not included in the data analysis. This experimental design is adapted from previous research using ERPs to study the UG (Hu et al., 2015; Ma, Qian, Hu, & Wang, 2017; Qu et al., 2013; Radke & de Bruijn, 2012). To reduce the duration of the ERP experiment, an imbalance was included in the trial design. The offers U1, U2, U4, and U5 were repeated 48 times respectively, whereas U3 was repeated 24 times, resulting in a total of 216 trials. Participants were paid 30 yuan plus the income of one randomly selected trial.

Participants were told that all the proposers are real whose information and offers were collected in the previous experiment. The proposers were divided into two types: proposer-in-need versus proposer-not-in-need. Proposer-in-need condition consisted of economically underprivileged students from a school in a remote, impoverished region. Proposer-not-in-need condition consisted of general students studying in a normal urban school. A simple 5-point scale was provided to rate participants’ cognitive empathy for the two kinds of proposers and was based on responses to the following statements: “I recognize the proposer’s situation”; “I understand the proposer’s behavior”; “I can imagine what the proposer is going through”; “The proposer’s reaction to the situation is understandable.” An internal consistency test was performed on the four measurement items for proposer type (in-need vs. not-in-need). Results showed that Cronbach’s α coefficient was 0.717 (proposer-in-need) and 0.813 (proposer-not-in-need).

Procedure

Participants were seated comfortably in front of a computer screen in an electrically isolated room. The experimental stimuli were presented in the center of a computer screen at a distance of 100 cm, with a visual angle of 8.69° × 6.52° (15.2 cm × 11.4 cm, width × height). We used the E-prime 3.0 software package (Psychology Software Tools, Pittsburgh, PA) for the stimuli presentation, triggers, and response recording. A keypad was provided for subjects to make choices. The experiment consisted of six blocks, and each block contained 36 trials. Counterbalancing of blocks was manipulated among the subjects. Practice trials were administered before the formal experiment.

A single trial is illustrated in Fig. 1. A fixation appeared at the beginning of each trial for 600-800 ms on a black screen. The proposer’s information was subsequently presented for 3,000 ms followed by the proposal of how to split 10 Yuan between herself/himself and the responder. The unfair/fair offer was shown in 1,500 ms, which is marked for ERP analysis. Then, participants decided to accept or reject this offer, and the decision result is presented for 1,000 ms as confirmation. After finishing the experiment, participants were asked to complete a simple 5-point scale to rate their empathy for the two kinds of proposers.

Fig. 1
figure 1

A single trial in the adapted Ultimatum Game. Each trial began with a fixation cross. Participants viewed the proposer information for 3,000 ms. After a blank screen of 1,000 ms, the unfair/fair offer was presented for 1,500 ms, which is marked for ERP analysis. Then, they were asked to choose to accept or reject this proposal by pressing the corresponding button. EEGs were recorded from the participants throughout the experiment

Full size image

EEG data recording

Electroencephalograms (EEGs) were recorded (bandpass 0.05-100 Hz, sampling rate 500 Hz) using a Neuroscan Synamp2 Amplifier (curry8, Neurosoft Labs, Inc., Virginia, USA) with Ag/AgCl electrodes placed at 64 scalp sites according to the extended international 10-20 system. An electrode between PFz and Fz on the forehead was connected as the ground, and the left mastoid was selected as an online reference. Two pairs of electrodes recorded vertical and horizontal electrooculograms (EOGs), one pair placed above and below the left eye in parallel with the pupil and the other pair placed 10 mm from the lateral canthi. All interelectrode impedances were less than 5 kΩ.

EEG recordings were digitally filtered with a low-pass, 30 Hz filter (24 dB/octave). EOG artifacts were corrected using the method proposed by Semlitsch et al. (1986). EEG data was time-locked to the onset of offer, processed and analyzed. The signal was segmented to analyze the epoch from 200 ms before the onset of offer to 800 ms after the onset, with the first 200-ms prestimulus used as a baseline. Trials containing amplifier clipping, bursts of electromyography activity, or peak-to-peak deflection exceeding ±100 μV were excluded. The EEG epochs were averaged for four conditions (proposer-in-need unfair, proposer-in-need fair, proposer-not-in-need unfair, proposer-not-in-need fair) for each participant. Then, the difference wave also was generated by subtracting the ERPs elicited by the fair trials from the ERPs elicited by unfair trials for each participant.

Data analysis

We chose the time windows of MFN and P300 on the basis of the visual observation and the guidelines proposed by the Picton team. According to Picton et al. (2000), we determine the peak of the grand average waveforms based on visual observation, and then select a certain time around the peak to determine the time window of this component. Thus, we analyzed the mean amplitude of the MFN in the time window 290-320 ms after the onset of offer and the mean amplitude of P300 in the 330–400-ms time window after the onset of offer. We selected Fz in the frontal area for the MFN and CPz in the central-parietal area for P300 in the statistical analysis. Within-subjects repeated-measures ANOVA for the MFN was performed with proposer (proposer-in-need and proposer-not-in-need) × fairness (fair and unfair offers). ANOVA for P300 was performed with proposer (proposer-in-need and proposer-not-in-need) × fairness (fair and unfair offers). Then, paired t-test for the MFN difference was performed with proposer (proposer-in-need and proposer-not-in-need). The Greenhouse-Geisser correction was applied when the assumption of sphericity was violated (Greenhouse & Geisser, 1959). Effect sizes in all ANOVA analyses were reported by partial eta-squared (η2p), with 0.05 representing a small effect, 0.10 representing a medium effect, and 0.20 representing a large effect (Cohen, 1973). If there was an interaction effect between factors, a simple effect analysis was conducted. Finally, a correlation analysis between the empathy rating and P300 amplitude was performed.

Also, we conducted a generalized linear mixed-effects model (GLMM) using the nlme package (R Core Team, 2014) with R studio (https://www.r-project.org/) software to check the robustness of MFN and P300 results. We defined subjects as random intercepts, MFN and P300 amplitude as the response variable respectively, and proposer type, fairness level, and empathy rating as the fixed effects.

Results

Behavioral results

Data for 1 subject was not recorded, leaving a total of 23 valid behavioral data. We conducted a two-way 2 (proposer) × 2 (fairness) repeated measure ANOVA for reaction time (RT) and offer accept frequency (AF), respectively. It showed that there was a significantly [F (1, 22) = 11.126, p = 0.002, η2p = 0.348] longer RT for unfair offers (M = 471.440, S.E. = 31.763) than that of fair offers (M = 424.949, S.E. = 26.101). However, the main effect of proposer [F (1, 22) = 0.616, p = 0.441, η2p = 0.027], and the interaction effect of these two factors were not significant [F (1, 22) = 1.040, p = 0.319, η2p = 0.045]. For the analysis of AF, there was a significant effect for the proposer [F (1, 22) = 11.447, p = 0.003, η2p = 0.342], fairness [F (1, 22) = 241.048, p < 0.001, η2p = 0.916], and their interaction effect [F (1, 22) = 10.233, p = 0.004, η2p = 0.317]. The proposer-in-need condition (M = 32.478, S.E. = 1.824) received more accept offers than the proposer-not-in-need condition (M = 26.891, S.E. = 0.880). The fair offers (M = 47.457, S.E. = 0.215) also were accepted more than the unfair offers (M = 11.913, S.E. = 2.305). For the significant interaction effect, a simple effect analysis also was conducted, which showed that the acceptance frequency for unfair offers from proposer-in-need (M = 17.347, S.E. = 3.633) was significantly larger than for proposer-not-in-need [M = 6.478, S.E. = 1.684; F (1, 22) = 10.900, p = 0.003, η2p = 0.331], but there was no difference in the fair offers acceptance frequency [F (1, 22) = 1.344, p = 0.259, η2p = 0.058].

We also conducted a paired t-test for participants’ empathy ratings of proposers-in-need and proposers-not-in-need. The result suggested that participants showed higher empathy toward proposers-in-need (M = 3.594, S.E. = 0.138) than proposers-not-in-need [M = 3.083, S.E. = 0.158; t (23) = 2.822, p = 0.010]. All of the behavioral results can be seen in Fig. 2.

Fig. 2
figure 2

Behavioral results. (A) Reaction time of unfair/fair offers. (B) Accept frequency of unfair/fair offers for proposer-in-need/proposer-not-in-need condition. (C) Empathy rating for proposer-in-need/proposer-not-in-need condition. Error bars indicate standard error of the mean. ∗∗∗p < 0.001; ∗∗p < 0.01; ∗p < 0.05

Full size image

EEG results

MFN results

We conducted a three-way 2 (proposer) × 2 (fairness) repeated measure ANOVA for mean MFN amplitudes. There was no significant main effect for fairness [F (1, 23) = 0.028, p = 0.869, η2p = 0.001]. However, the main effect of proposer [F (1, 23) = 15.503, p < 0.001, η2p = 0.403] and interaction effect between proposer and fairness were obvious [F (1, 23) = 10.776, p = 0.003, η2p = 0.319]. The MFN is a negative polarity component, in which low voltage means larger amplitude. As a result, the overall MFN amplitude was smaller for the proposer-not-in-need condition (M = −1.196 μV, S.E. = 1.068) than the proposer-in-need condition (M = −2.601 μV, S.E. = 1.164). Further simple effect analysis indicated that the difference between fair and unfair offers was marginally significant under the proposer-not-in-need condition [F (1, 23) = 3.841, p = 0.062, η2p = 0.143], which indicated that the unfair condition (M = −1.708 μV, S.E. = 1.113) elicited significantly larger MFN amplitudes than the fair condition (M = −0.684 μV, S.E. = 1.086). It also was marginally significant under the proposer-in-need condition [F (1, 23) = 3.900, p = 0.060, η2p = 0.145], but the MFN amplitude of unfair condition (M = −2.014 μV, S.E. = 1.091) was smaller than that of the fair condition (M = −3.188 μV, S.E. = 1.303).

In terms of MFN difference, the paired t-test was conducted. It showed that the proposer-not-in-need’s unfair-fair condition (M = −1.024 μV, S.E. = 0.522) evoked an obviously larger deflection compared with the proposer-in-need’s unfair-fair condition (M = 1.173 μV, S.E. = 0.594; t (23) = 3.283, p = 0.003). Figure 3A shows grand-average waveforms of the MFN at Fz site. Figure 3B depicts topographic maps in which we can more intuitively observe the difference of the MFN valence effect among the four conditions. Moreover, Fig. 3C shows the grand-average waveforms of the MFN difference at the Fz site, and Fig. 4D and E contain topographic maps and a bar graph for the MFN difference results.

Fig. 3
figure 3

MFN and MFN difference results. (A) Grand-average ERP waveforms at channel Fz as a function of proposer (proposer-in-need/proposer-not-in-need) and fairness (unfair/fair) for offer. (B) Topographic maps for MFN. (C) MFN difference (unfair MFN minus fair MFN) at channels Fz based on proposer (proposer-in-need/proposer-not-in-need). (D) Topographic maps for MFN difference. (E) Bar graph for MFN difference. Shaded areas indicate the time window of the MFN (290–320 ms) used for statistical analysis. ∗∗p < 0.01

Full size image
Fig. 4
figure 4

P300 results. (A) Grand-average ERP waveform at channels CPz as a function of proposer (proposer-in-need/proposer-not-in-need) and fairness (unfair/fair) for offer. (B) Average P300 (the average of unfair P300 and fair P300) amplitude at channels CPz based on proposer (proposer-in-need/proposer-not-in-need). Shaded areas indicate the time window of the P300 (330–400 ms) used for statistical analysis. (C) Topographic maps of Average P300. (D) Bar graph of Average P300. ∗∗∗p < 0.001

Full size image

P300 results

Additionally, the three-way 2 (proposer) × 2 (fairness) repeated measure ANOVA for the P300 amplitudes revealed a significant main effect for proposer [F (1, 23) = 19.156, p < 0.001, η2p = 0.454]. This result indicated that the P300 magnitudes were smaller in proposer-in-need’s condition (M = 1.919 μV, S.E. = 0.927) than in proposer-not-in-need’s condition (M = 3.195 μV, S.E. = 0.874). However, the main effect of fairness was not significant [F (1, 23) = 0.133, p = 0.719, η2p = 0.006]. The interaction effect between proposer and fairness also was not significant [F (1, 23) = 0.174, p = 0.680, η2p = 0.008]. Figure 4A and B show the grand-average waveforms of the P300 at the CPz site among the four conditions and between two conditions. Figure 4C and D contain topographic maps and bar graph for the Grand-average ERP waveforms from channels CPz, in which we can more intuitively observe the difference of the P300 valence effect between the proposer-in-need and the proposer-not-in-need conditions. Table 1 summarized the descriptive results of ERPs.

Table 1 Descriptive results of ERPs
Full size table

GLMM results of MFN and P300

As to the MFN effect, proposer type has a positive effect on MFN (Model 1) but not fairness (Model 2). The interaction term of proposer and fairness is significantly positive (Model 3). As to the P300 effect, proposer (Model 4) and empathy rating (Model 7) has a positive effect on P300, but this is not evident with fairness (Model 5) and the interaction term of proposer and fairness (Model 6). Accordingly, these results are consistent with our ANOVA results. Moreover, we include empathy rating as an additional level-2 predictor based on model 6. This result indicates that empathy rating has a negative effect on P300 (Model 7). Table 2 summarized the GLMM results.

Table 2 Results of GLMM
Full size table

Correlation analysis

We conducted a Pearson correlation analysis between empathy rating and the P300 components for proposer-in-need and proposer-not-in-need. It showed a negative correlation between empathy rating for proposer-in-need and the P300 for proposer-in-need regardless of fairness. However, the correlation between empathy for proposer-not-in-need and the P300 for proposer-not-in-need was insignificant. The results were summarized in Table 3 and showed in Fig. 5.

Table 3 Correlations between P300 amplitude and empathy rating
Full size table
Fig. 5
figure 5

Scatter plots and Pearson correlation coefficient analysis results. The first row shows correlation between empathy rating and P300 under proposer-in-need condition, r- and p-values are highlighted in red. The second row shows correlation between empathy rating and P300 under proposer-not-in-need condition, r- and p-values are shown in black (Unfair offer in A/D, Fair offer in B/E, Average of unfair and fair in C/F)

Full size image

Discussion

This study set out to elucidate how the neediness level of the proposer influences the responder’s fairness perception and examine why and how this phenomenon occurs. The behavioral data indicate that subjects are generally more likely to accept fair offers (4:6 and 5:5) than unfair offers (1:9 and 2:8), consistent with prior studies (Falk, Fehr, & Fischbacher, 2003; Fehr & Schmidt, 1999). Furthermore, the reaction time also supports our hypothesis that reaction is faster for fair offers than for unfair offers. More importantly, our results show that subjects are more inclined to accept the unfair offers from the proposers who are in need compared with those who are not.

Following the behavioral results, we also found the same phenomenon on MFN results in the offer presentation stage. The MFN was more negative-going for unfair than for fair offers in the proposer-not-in-need condition. Probably most strikingly, this MFN difference effect (Fig. 3E) was completely reversed in the proposer-in-need condition. As pointed out in the introduction, MFN is sensitive to expectancy violations (Bellebaum, Polezzi, & Daum, 2010; Boksem & Cremer, 2009; Hewig et al., 2011). Thus, the MFN difference effect in the proposer-not-in-need condition reflected the detection of a violation from the subjects’ expectations (even asset distribution is an expected social norm) (Fehr & Fischbacher, 2004; Fehr & Gächter, 2002; Messick & Sentis, 1983). However, the reversed MFN difference effect in the proposer-in-need condition suggested that the subject’s expectations for the proposers in need were reversed, suggesting that they did not even expect them to offer fair proposals. The inverted MFN difference effect has scarcely been detected in previous UG studies regarding fairness perceptions and social roles. A relatively similar pattern was observed in a study conducted by Ma et al. (2015). They found that participants’ fairness perception would be weakened by facial attractiveness, which reflected in a null d-FRN effect (Ma et al., 2015). In our experiment, however, this MFN difference effect was reversed rather than absent suggested that participants have much less fairness perception for proposers in need, showing a strong “pure altruism” phenomenon. The differential MFN difference result may be because the facial attractiveness of the proposer in Ma et al. (2015)’s study is related to emotion, while we focus on the economic situation of the proposer, which is related to cognitive state. These results also might support the notion of a shared representational network, which suggests that either observing or imagining other people’s situations activates neural networks involved in those situations (Cacioppo & Decety, 2009; Cui, Abdelgabar, Keysers, & Gazzola, 2015; Lamm, Decety, & Singer, 2010; Marsh et al., 2014). Hence, the shared neural network could be evoked in fairness processing when individuals perceived the proposer’s economic neediness. Therefore, in this study’s proposer-in-need condition, responders might feel that it was reasonable and excusable for underprivileged students to offer an unfair proposal and not even want them to offer a fair proposal. This would have activated the prosocial behaviors and influence evaluative processing of (un)fair outcomes.

A later cognitive processing stage of fairness evaluation, reflected in the P300, also supported these results. A larger P300 was observed in the proposer-not-in-need condition than in its opposite condition, irrespective of the fairness of the offer. As outlined in the introduction, high motivational stimuli would induce more positive P300 than low motivational stimuli (Jin et al., 2020; Leng & Zhou, 2009; Ma et al., 2011). Accordingly, the P300 results support our hypothesis that the motivational salience and cognitive load toward the proposals were higher in the proposer-not-in-need condition than in the proposer-in-need condition.

Surprisingly, we also found a negative correlation between empathy rating and the P300 amplitude in the proposer-in-need condition regardless of the fairness of the offers, while this pattern did not exist in the proposer-not-in-need condition. According to the cognitive significance of P300 (X. Hu et al., 2017; Nieuwenhuis et al., 2005; Yang et al., 2014), this negative correlation indicated that the higher the empathic concern for the proposer in need, the less motivational/affective salience and less cognitive resources were allocated for his/her offers. However, the empathic concern was not significant in its correlation with proposer-not-in-need meant that the empathic concern remained unchanged regardless of the economic conditions of proposers. According to our hypothesis, these results demonstrate that the difference in empathic concern of the two conditions was due to the increase in the proposer-in-need condition but its opposite condition remained unchanged. It is consistent with the previous findings that the empathic concern would be increased by the perceived neediness of a person in need (Batson & Ahmad, 2001; Batson & Moran, 1999; Stocks et al., 2011; Woltin et al., 2011). Yet, these studies are from the perspective of self-report, while largely ignoring the intrinsic neural mechanism. Sanfey (2007) reported that revealing neural mechanisms can help us better understand the underlying reasons behind social phenomenon, such as cooperation and fairness perception (Sanfey, 2007). Our result provided neural evidence that perceived neediness not only would increase the empathic concern, but also reduce the fairness perception. Besides, previous studies have pointed out that the empathic concern would be decreased due to the social roles, such as the intergroup empathy bias (Bruneau, Cikara, & Saxe, 2015), attractiveness of children in need (Fisher, 2014), or upper-class of decision-maker (Cote et al., 2012). In the proposer-not-in-need condition, empathic concern remained unchanged could be explained that these proposers are general strangers without the above characteristics. Hence, the empathic concern was not increased or decreased in such a condition.

Both behavioral and ERP results indicate that participants’ fairness perception was deeply affected by the degree of neediness. Crucially, a reversed MFN difference effect was found to reinforce the importance of economic neediness and demonstrated that fairness perception was highly context-dependent. Additionally, this study points out a central role for empathic concern in social decision-making related to fairness consideration. Our findings showed that perceived neediness stimulates individuals’ empathic concern and further weakens the fairness perception only in the condition of economic neediness, while people focus on the pursuit of fairness in the proposer-not-in-need condition and their empathic concern remained unchanged.

The current study may have several critical implications. First, our findings suggest that the proposers’ neediness level may lead to different behaviors in terms of fairness. Specifically, an individual’s empathic concern may be promoted by the perceived neediness and further influence our fairness perception of monetary division, whereas this effect did not exist with proposers who are not in need. This subtle process may enhance disparities that emerge between different social roles. Second, several studies have confirmed that situational induced empathy affects an individual’s altruistic behavior (Liu et al., 2020), while the relationship with fairness perception remains unclear. These results have noteworthy theoretical significance for recent advancements in state empathy. Third, a reversed MFN difference effect between the proposer-in-need condition and proposer-not-in-need condition was observed in the current study. The MFN difference results suggest that participants are prone to be unfair rather than fair in the proposer-in-need condition. This idea indicates that subjects are willing to suffer a loss themselves in order to benefit the proposers who are in need, which showed a strong “pure altruism” phenomenon.