In 1971, Henri Tajfel grouped strangers based on whether they preferred the paintings of Klee or Kandinsky. Then he let them distribute small amounts of money among the others. Despite having no history, no contact, no reason to favor one over another, they systematically gave more to members of their own “group” (Tajfel et al., 1971).
The finding has been replicated with coin flips, the letter of the alphabet your name starts with, and whether you underestimated or overestimated the number of dots on a screen. The group doesn’t have to mean anything. The bias shows up anyway.
The minimal group paradigm
This is among the most robust findings in social psychology. In the simplest version:
- Strangers enter a lab
- They’re randomly split into two groups based on something trivial
- They’re asked to distribute small rewards to others
- They give about 65-70% to their own group, 30-35% to the other
The groups share no history, no values, no interests. The participants often don’t even know who else is in their group. The bias appears in minutes.
Tribal identity doesn’t need a tribe. The label alone is enough.
It’s not just preference. It’s perception.
Xu and colleagues (2009) put people in an fMRI scanner and showed them videos of hands being pricked by a needle. The brain’s anterior cingulate cortex (a core empathy region) lit up strongly for in-group members and significantly less for out-group members (Xu et al., 2009).
Same pain. Same stimulus. Different neural response.
Related effects:
- Cross-race recognition deficit. We’re about 1.5 standard deviations worse at distinguishing individual faces of an out-group race (Meissner & Brigham, 2001 meta-analysis). “They all look alike” is a real cognitive phenomenon, not a slur.
- Memory asymmetry. In-group members are remembered as individuals; out-group members get remembered as category members.
- Dehumanization at the extreme. For sufficiently “distant” out-groups, fMRI shows reduced activity in the brain regions we normally use to process persons (Harris & Fiske, 2006).
The same action means different things by group
Identical behavior gets interpreted through the group lens:
| Action | In-group actor | Out-group actor |
|---|---|---|
| Violence | “They were pushed too far” | “That’s what they do” |
| Success | “They earned it” | “They got lucky / cheated” |
| Failure | “Bad circumstances” | “Confirms the pattern” |
| Cooperation | “Of course, we’re good” | “Suspicious, what do they want?” |
This isn’t conscious rationalization. The interpretation happens pre-reflectively:
- Out-group action gets coded as dispositional. “That’s their true nature.”
- In-group action gets coded as situational. “Just the circumstances.”
This is the ultimate attribution error (Pettigrew, 1979). It’s one of the engines that keeps intergroup conflict stable even when both groups behave identically.
Common enemies unify
Muzafer Sherif’s Robbers Cave experiment (1954) split 22 boys at a summer camp into two random groups. Within days:
- They invented rivalries
- They raided each other’s cabins
- They burned each other’s flags
Then Sherif tried to reduce the hostility. The results were asymmetric:
| Intervention | Worked? |
|---|---|
| Watching movies together | No |
| Sharing meals | No |
| Fixing a broken water supply together (shared threat) | Yes |
Superordinate goals dissolve group boundaries. Shared enemies temporarily unite against an even bigger out-group.
Politicians, coaches, and demagogues have been exploiting this for millennia.
Why this matters
- You’re always in multiple groups at once. National, religious, political, professional, sports. Which one is salient right now depends on framing.
- The out-group is lumped, the in-group is individuated. Watch your own language. “They” is a tell.
- Out-group signals are amplified in modern media. You hear about out-group crimes, never about their ordinary mornings.
- Your cheater-detection runs hotter on out-group members. The same behavior will register as suspicious when they do it and reasonable when you do.
- Superordinate goals work. Contact alone doesn’t. Forcing a shared task with a shared outcome beats “just be exposed to each other.”
The smallest groups are enough to trigger tribalism. The largest goals are needed to dissolve it.