Imagine walking into a room with twenty strangers.
Someone flips a coin.
Half the room becomes Group A.
The other half becomes Group B.
There is no history.
No territory.
No ideology.
No family connection.
No reason to believe your side is better.
And yet something strange can happen almost immediately.
People begin to distribute money a little differently.
They expect more cooperation from their own side.
They prefer outcomes that favor people carrying the same meaningless label.
The boundary was invented seconds ago.
But psychologically, it is already doing work.
This is one of the most uncomfortable findings in social psychology:
Humans do not always need deep differences to create an "us" and a "them." Sometimes a category is enough to begin changing behavior.
That does not mean humans are doomed to tribal hatred.
The real story is more interesting.
The tribe can be created in a laboratory
In the late 1960s and early 1970s, Henri Tajfel and colleagues developed what became known as the minimal group paradigm.
The idea was deliberately extreme.
Strip group membership down until almost nothing remains.
Participants could be divided according to trivial preferences, arbitrary classifications or even chance. They might never meet the other members. There was no long history of conflict and often no direct personal gain from favoring the ingroup.
Then participants were asked to allocate rewards or make judgments.
Even under these minimal conditions, researchers repeatedly observed forms of ingroup favoritism.
The implication was not that every arbitrary group produces hostility.
It was that social categorization itself can become behaviorally relevant much faster than common sense suggests.
A label can become a reference point.
Once that happens, the mind can begin asking:
Who are we?
What do people like us do?
What can I expect from them?
What should they expect from me?
Who deserves my cooperation?
The group starts to exist not only on paper, but inside prediction.
How large is the effect?
This is where the story needs numbers rather than mythology.
A major 2014 meta-analysis examined ingroup favoritism in cooperative decision making across 212 studies from 77 publications.
Overall, people were more cooperative with ingroup members than outgroup members, with a small-to-medium average effect of d = 0.32.
But the effect changed with context.
When people were simply categorized and there was no mutual interdependence, ingroup favoritism was smaller, around d = 0.19.
When the situation involved interdependence, such as social dilemmas in which participants' outcomes were linked, favoritism was stronger, around d = 0.42.
That distinction matters.
The popular version says:
Humans see a tribe and automatically become biased.
The evidence suggests something richer.
Categorization matters.
But expectation, reciprocity, interdependence and what people think other group members will do also matter.
The group is not just a badge.
It can become a model of future behavior.
"Us" does not automatically mean "I hate them"
Another common simplification is that ingroup favoritism and outgroup hatred are the same phenomenon.
They are not.
The 2014 meta-analysis found stronger support for the idea that discrimination in cooperation often reflects ingroup favoritism rather than active outgroup derogation.
That is a crucial distinction.
You can prefer to help your family without hating every other family.
You can trust your team more because you share norms, experience and mutual obligations.
You can choose your community first under uncertainty without wanting outsiders harmed.
Of course, favoritism can become discrimination.
And under the wrong conditions, group identity can combine with threat, competition, ideology or dehumanization in dangerous ways.
But the psychological starting point may be closer to:
"I know who my people are."
rather than:
"I hate everyone else."
Understanding that difference makes the system easier to study.
Why the "tribal brain" metaphor is useful and dangerous
You will often hear that humans have a tribal brain.
The phrase captures something real.
Human beings evolved in social groups.
Belonging matters.
Reputation matters.
Cooperation matters.
Group membership can alter perception, emotion and behavior.
But "tribal brain" easily turns into a lazy master explanation.
A political argument becomes tribalism.
A football rivalry becomes tribalism.
A scientific disagreement becomes tribalism.
A religious identity becomes tribalism.
A brand community becomes tribalism.
Eventually the word explains everything and therefore almost nothing.
Modern research shows multiple mechanisms can produce ingroup favoritism.
Social identity is one.
Expectations of reciprocal cooperation are another.
Norms matter.
Competition matters.
Shared goals matter.
History matters.
Threat matters.
Status matters.
The fact that people categorize does not tell you which mechanism is driving the behavior you are watching.
Why opinion can become identity
Groups become especially powerful when the boundary stops describing what you think and starts describing who you are.
A 2025 review and meta-analysis of opinion-based groups synthesized 45 studies, 124 independent effect sizes and more than 30,000 observations.
It found a medium positive relationship between social identity and intergroup phenomena, around r = .46, with substantial heterogeneity.
That last phrase matters.
The relationship was not identical everywhere.
But the pattern helps explain why an opinion can change psychological status.
At first:
I believe X.
Later:
People like me believe X.
Then:
We are the people who understand X.
Finally:
If you reject X, you are rejecting people like us.
Now evidence is no longer entering an empty room.
It is entering a room containing belonging, loyalty and social cost.
Changing your mind can begin to feel less like updating a belief and more like leaving a group.
The hidden power of expected reciprocity
There is another reason group boundaries matter.
Uncertainty is expensive.
When you do not know how a stranger will behave, any available signal becomes useful.
Shared group membership can function as one of those signals.
People may expect ingroup members to be more cooperative.
If I believe you are more likely to help me because we belong to the same group, helping you first becomes more rational.
Now favoritism can reinforce itself.
I cooperate with you because I expect cooperation.
You experience cooperation from me.
You reciprocate.
My belief that "our group takes care of its own" becomes more credible.
This is why modern researchers still debate how much ingroup favoritism comes from identity itself and how much comes from bounded generalized reciprocity, the expectation that help circulates more reliably within the group.
A 2024 Nature Reviews Psychology commentary described this as an active theoretical competition rather than a settled single-mechanism story.
That is good science.
A robust phenomenon can be real while its psychological engine remains debated.
Self-esteem is not a simple prejudice switch
Older versions of social identity theory often became popularized like this:
People feel bad about themselves.
They boost their self-esteem by making their own group look better.
Therefore low self-esteem causes prejudice.
That story is too clean.
A 2024 meta-analysis examined more than 100 effect sizes involving self-esteem and ingroup or outgroup evaluations.
The results did not support a simple rule that low self-esteem automatically produces outgroup hatred.
Self-esteem relationships depended on what was measured, which groups were being judged and the social context.
Again, the useful lesson is not that identity theory failed.
It is that human group behavior refuses to collapse into one motivational switch.
Why groups can make people better
There is a reason humans form groups so easily.
Groups do not only produce bias.
They produce cooperation.
Mutual aid.
Shared knowledge.
Collective action.
Norm enforcement.
Belonging.
Social support.
Sacrifice for people who may never repay you directly.
A group can make a person more generous, brave and reliable.
The same mechanism that says these are my people can create remarkable prosocial behavior.
The problem begins when the boundary changes what counts as morally relevant.
When evidence from outsiders is discounted automatically.
When harm to an outgroup stops feeling like harm.
When disagreement inside the group becomes betrayal.
When status depends on displaying greater hostility than everyone else.
That is no longer merely belonging.
It is an identity system that rewards division.
Why social media can sharpen the boundary
Online groups add a new ingredient.
Identity is visible.
Every like, repost, avatar, flag, badge, phrase and follower network can signal affiliation.
An audience can see whether you behaved like a "real" member.
This changes the incentive structure.
A private opinion can become public performance.
Moderation may look weak.
Uncertainty may look disloyal.
Outrage may demonstrate commitment.
The platform did not invent group identity.
But digital environments can make identity signals faster, more visible and more measurable.
The most extreme member can sometimes become the clearest prototype.
And once people are competing to represent the group, the boundary can harden.
The question that matters more than "Are humans tribal?"
The useful question is not:
Are humans tribal?
The answer is too broad to help.
Ask instead:
What is this group rewarding?
Cooperation?
Accuracy?
Loyalty?
Status?
Humiliation of outsiders?
Expertise?
Compassion?
Certainty?
Dissent?
The same human capacity for group identity can produce radically different outcomes depending on the norm attached to membership.
That means the boundary is not destiny.
The DarkBrain lens
When you feel the pull of "us" and "them," do not immediately shame yourself for being tribal.
Observe the system.
What exactly defines the group?
How meaningful is the boundary?
What do you expect from insiders that you do not expect from outsiders?
What behavior earns status inside the group?
What would it cost you socially to disagree?
Would you judge the same evidence differently if it came from your own side?
And the most uncomfortable question:
Do you dislike the other group, or have you simply stopped examining them one person at a time?
Groups help humans coordinate.
They also compress complexity.
The danger begins when the label becomes more real than the person wearing it.
Because the fastest way to create "them" is not always hatred.
Sometimes it is simply drawing a line and forgetting that you drew it.