Using performance art to promote intergroup prosociality by cultivating the belief that empathy is unlimited

Abstract

Empathy is important for resolving intergroup conflicts. However, people often tend to feel less empathy toward people who do not belong to their social group (i.e., outgroup members). We propose that this tendency is due, in part, to the belief that empathy is a limited resource. To overcome this issue, we develop an intervention synthesizing psychology and art to increase the belief that empathy is unlimited. In six studies (n = 2118), we find that the more people believe empathy is limited, the less outgroup empathy they experience. Moreover, leading people to believe that empathy is unlimited increase outgroup empathy, leads to greater support for prosocial actions toward outgroup members, and encourages more empathic behaviors toward outgroup members in face-to-face intergroup interactions. These intervention effects are observed across various intergroup contexts involving different ethnic, national, religious, and political groups. Thus, changing beliefs about empathy may improve intergroup relations, and conveying this belief through art may promote social change.

Fig. 1. The interactive effect of condition (ingroup targets, n = 92 vs. outgroup targets, n = 90) and belief about empathy as an unlimited resource on participants’ empathic reactions toward ingroup and outgroup members.

A significant Condition × Belief interaction (b = 0.41, SE = 0.17, t(178) = 2.40, p = 0.017; 95% CI = [0.074, 0.76]). A significant difference between empathy towards ingroup and outgroup members among participants who believed empathy is more limited (t(178) = −3.52, p < 0.001; 95% CI = [−1.6, −0.45]). No significant difference in empathy towards ingroup and outgroup members among participants who believed empathy was less limited (t(178) = 0.03, p = 0.98; 95% CI = [−0.64, 0.65]). Error bands represent a 95% confidence interval. All tests are two-tailed. Source data are provided as a Source Data file.

Fig. 2. Empathic reactions toward outgroup targets, as a function of Condition (i.e., limited empathy, n = 99; and unlimited empathy, n = 101) and order of empathy-inducing testimonies (i.e., 1–4).

A significant main effect of condition on empathic reactions, F(1, 198) = 8.93, p = 0.003, d = 0.423 (a repeated-measures ANOVA). The plot presents the maximum and minimum values as whiskers, the interquartile range as the vertical length of each box, the mean as the x marker, and the median as the horizontal line within each box. All tests are two-tailed. Source data are provided as a Source Data file.

Fig. 3. Empathic reactions and support for prosocial actions.

a Empathic reactions toward ingroup and outgroup targets, as a function of condition (i.e., limited empathy, n = 51; unlimited empathy, n = 52; and control, n = 47). A significant target identity × condition interaction, F(2, 147) = 4.50, p = 0.013, d = 0.5. A significant difference in empathy toward ingroup and outgroup members in the limited condition (p < 0.001) and in the control condition (p < 0.001). No significant difference in empathy towards ingroup and outgroup members among participants in the unlimited condition (p = 0.654). b Support for prosocial actions toward ingroup and outgroup targets as a function of condition (i.e., limited empathy, n = 51; unlimited empathy, n = 52; and control, n = 47). A significant target identity × condition interaction, F(2, 147) = 4.76, p = 0.01, d = 0.51. A significant difference in support for prosocial actions toward ingroup and outgroup members in the limited condition (p < 0.001) and in the control condition (p = 0.002). No significant difference in support for prosocial actions toward ingroup and outgroup members in the unlimited condition (p = 0.72). The plots present the maximum and minimum values as whiskers, the interquartile range as the vertical length of each box, the mean as the x marker, and the median as the horizontal line within each box. All tests are two-tailed. Source data are provided as a Source Data file.

Fig. 4. A mediation model of the effect of condition on intergroup bias in support for pro-social actions through intergroup empathy bias.

A mediation analysis shows that intergroup empathy bias mediates the effect of manipulated beliefs about empathy as a limited resource on intergroup bias in support of prosocial actions. * indicates p < 0.05, **p < 0.01, ***p < 0.001. All tests are two-tailed. Source data are provided as a Source Data file.

Fig. 5. The performance-experiment setting in Study 4.

Participants met actresses who were seated in a theatrical set inspired by the architecture of “airport immigration booths” welcoming one participant at a time for an “entry” interview.

Fig. 6. Empathic reactions toward outgroup targets, as a function of condition (i.e., limited empathy, n = 53; and unlimited empathy, n = 55) and order of empathy-inducing testimonies (i.e., 1–4).

A significant main effect of the condition on empathic reactions, F(1, 106) = 6.34, p = 0.013, d = 0.49. The plot presents the maximum and minimum values as whiskers, the interquartile range as the vertical length of each box, the mean as the x marker, and the median as the horizontal line within each box. All tests are two-tailed. Source data are provided as a Source Data file.

Fig. 7. The performance-experiment phases in Study 5.

The event included four main phases: (1) Introduction in which participants were randomly assigned to an experimental or a control condition; (2) Intergroup encounter in which participants met separately with an Arab actor and with a Jewish actor, each of whom shared a sad personal story with them; (3) Debriefing and science outreach in which all attendees watched a video art and a live stage performance sharing a review of the literature on outgroup empathy and the main findings of this research project thus far; and (4) Open discussion in which all attendees could share their experience with each others and talk directly with the research team.

Fig. 8. Empathic reactions toward ingroup and outgroup targets, as a function of condition (i.e., unlimited empathy, n = 88; and control, n = 84).

A significant target identity × condition interaction, F(1, 170) = 5.09, p = 0.025, d = 0.35. A significant difference in empathy toward ingroup and outgroup members among participants in the control condition (p < 0.001). No significant difference in empathy towards ingroup and outgroup members among participants in the unlimited condition (p = 0.344). The plot presents the maximum and minimum values as whiskers, the interquartile range as the vertical length of each box, the mean as the x marker, and the median as the horizontal line within each box. All tests are two-tailed. Source data are provided as a Source Data file.

Fig. 9. Empathic behavior (no interpersonal touch; handshake; hug) toward ingroup and outgroup targets, as a function of condition (i.e., unlimited empathy, n = 88; and control, n = 86).

A significant difference in selecting the most empathic behavior (i.e., a hug) toward ingroup and outgroup members among participants in the control condition (p < 0.001). No significant difference in selecting the most empathic behavior toward ingroup and outgroup members among participants in the unlimited condition (p = 1; McNemar’s chi-square test). All tests are two-tailed. Source data are provided as a Source Data file.