You'd think it'd be difficult to argue against the following straightforward recommendation: Increase empathy towards outgroups, and you should be able to reduce prejudice and improve intergroup relations. Alas, as with most things having to do with human behavior, the reality is not straightforward, and empathy is not a silver bullet against intergroup negativity.
Jacquie Vorauer at the University of Manitoba in Canada is one of the most original researchers in the field of intergroup relations, and it is thus no surprise that research stemming from her lab would show how empathy, ironically, can actually be damaging for intergroup interaction. A lot of her work has focused on intergroup relations among Native and White Canadians (a refreshing change from the emphasis on Black-White relations that pervades U.S. psychology). In a 2009 paper published in the Journal Psychological Science, Vorauer and Stacey Sasaki report findings from a study in which they looked at the effects of an empathy induction on intergroup vs. within-group interaction. The researchers had White participants view a video documentary describing the housing hardships and injustices experienced by Native Canadians titled "Wrapped in Plastic: Housing Manitoba First Nations," and instructed the viewers to either view it from an objective stance ("try not to get caught up in how __ feels; just remain objective and detached") or to actively empathize with the film's protagonists ("try to imagine how __ feels about what has happened and how it has affected her life.").
Thus, participants were either in an empathy condition or a neutral condition. Additionally, though, they were led to believe that they would then talk about the film with either another White Canadian or with an Aboriginal Canadian. No interaction actually took place, but right before participants expected to interact with their partner, they were asked how much they were looking forward to the interaction, as well as their levels of prejudice.
The straightforward prediction would be that among participants who were asked to empathize with the injustices of the Native Canadian protagonists in the film, the more likely they would be to show reduced prejudice, and the more likely they would be to want to interact with their partner in talking about the film.
The results, however, were surprising: the predictions were correct for the White Canadian viewers who expected to talk about the film with other White Canadian viewers. But among participants who expected to talk about the film with an outgroup member, there was no evidence of prejudice reduction, and participants did NOT look forward to intergroup interaction. This was especially true among highly prejudiced people, whom empathy interventions might be naturally targeted towards.
How do we explain these findings? Vorauer and Sasaki found that in the intergroup condition, White participants became worried about metasteretoypes in other words, the stereotypes that their Aboriginal Canadian partner would view them under. In other words, a perspective-taking manipulation had the effect of making participants anticipate the negativity that they themselves might be viewed with, and in anticipation, recoil and react against the upcoming interaction.
Although the findings can be disheartening, they are also hugely important in underlining the boundaries and road-blocks that our well-intentioned ideas (e.g., let's build empathy and the world will be a better place!) may face. This is not to say that empathy doesn't have a place in intergroup relations--in the presence of intergroup contact and friendship, for example, research suggests that a positive side-effect is empathy towards out-groups. However, in the context of friendship, rejection concerns are likely to be less pronounced, and this is important to know.
The question is not whether empathy fosters or hinders positive intergroup interaction, but rather, when does empathy foster vs. threaten intergroup interaction?
Cross-posted from Rodolfo Mendoza-Dentons blog on the Psychology Today website.