Looking into the human literature on the evolution of cooperation, I feel that studies on humans are often conducted and interpreted poorly compared to studies of cooperation in ants, bacteria, fish, and other nonhuman primates. One point of confusion involves wrong assumptions about what individual humans should maximize and how well they should do it. But another (which I will discuss here) is an overwhelming focus in the social sciences on human groups as homogenous entities in conflict with each other.
According to much of the social psychology literature, humans conceptualize others into an “ingroup” and “outgroup”. This intuitive tribalistic categorization is considered a basic facet of human nature, and many research articles have been written about its causes and consequences. I will give a couple of random examples. This paper in Science states,
Humans regulate intergroup conflict through parochial altruism; they self-sacrifice to contribute to in-group welfare and to aggress against competing out-groups. Parochial altruism has distinct survival functions, and the brain may have evolved to sustain and promote in-group cohesion and effectiveness and to ward off threatening out-groups.
A paper I just saw yesterday on learning and empathy, states,
Deficits in empathy for out-group members are pervasive, with negative societal impact. It is therefore important to ascertain whether empathy toward out-groups can be learned and how learning experiences change empathy-related brain responses.
An implicit notion is that empathy is essentially a dichotomous variable extended to the ingroup but not the outgroup. The prototypical experiment demonstrating the ingroup bias will divide participants into arbitrary groups like “red team” and “blue team” which then predictably leads to all manner of competitive attitudes and biases wherein the ‘reds’ are more judgmental towards the ‘blues’ and more generous and empathetic to fellow ‘reds’. One of the original classic cases is the Robbers Cave study.
This ingroup vs outgroup phenomenon is important, but it’s often over-emphasized to the point of distorting the complexity of the evolutionary design of human social cognition. For example, some influential authors believe that the bias towards cooperating with groupmates and aggressing against outgroupers amounts to a group-level altruistic trait that is only explainable by some form of group selection. A multi-level selection model would assume that, although ancestral humans spent much time competing with fellow groupmates for mates and resources– the real key to their survival and reproduction was successful competition with other tribes. The idea here is that competition between tribes was so severe that it led to heritable tendencies to put the average reproductive success of the group ahead of one’s own relative survival and reproduction within the group.
But humans are not eusocial insects! It is not our evolved human nature to give up one’s life for the group. Humans tend to help their group, because doing so simply helps their own direct fitness. Our tendency for human tribalism is mutually beneficial, not altruistic in the evolutionary sense*. I would hope everyone would agree on these basic points, and yet I’m not so sure when I read the literature.
(*One point of confusion here is that multi-level selection has a different definition of altruism, which allows this human tribalism to be called “altruistic” whereas the orthodox inclusive fitness definition means we can never be as “altruistic” as ants and slime mold. I think this is one additional reason that many social scientists prefer the multi-level selection terminology that most evolutionary biologists don’t prefer.)
Human evolutionary history is not a mere series of tribal wars. Selection continues on between these conflicts. Given that the majority of social behaviors occur ‘within-group’, it is unclear whether the ingroup fitness effects are overshadowed by the fitness effects of intergroup conflict (as assumed in the narrative given above). Within each of our ‘ingroups’, we are still closer to some group members than others (although one might not even detect this fact if one is focused solely on intergroup conflict). A machine well-designed for navigating a social world (i.e. a human’s social brain) would surely possess adaptations for dealing with cooperation and conflict within the group, and not simply sit idle waiting in anticipation for the next tribal war.
Upon closer inspection, a social group is really a social network. And upon even further inspection, the links in that network are weighted differently and change over time. A more nuanced individualistic approach is therefore to imagine that people place others on a spectrum of dynamic social distance, with close friends and family on one end and threatening strangers on the other.
So what happens when experimenters place strangers into red and blue teams? They are sampling two points on that spectrum, thereby creating the discrete group mentality that they claim they are revealing. A complex continuous variable is reduced to a simple discrete one.
A similar mistake is easy to make when we think about discrete personality types. For each personality trait, say extravert vs introvert, there is a continuous spectrum which probably approximates a normal distribution. Most of us are not truly ‘extraverts’ or ‘introverts’ but are rather just average and in the middle. But I’m always surprised by the number of people who think that people actually come in these distinct categories or types and that being moderate or average is therefore rare: “Are you an introvert or extravert? You must be one or the other!”
If you surveyed people who considered themselves definite extraverts or introverts, then placed them in two groups, and tested their behavior–you would see an exaggerated difference between these groups for two reasons. First there is the biased sampling. Second, there is a priming effect. The result would obscure the real complexity and flexibility of human personality often taken for granted: most of us are extraverted or introverted to a near optimal degree for the current social situation.
The same thing is true of human cooperation. Humans are obviously very good at flexibly treating others as either competitors, collaborators, or a mix of both. Of course it makes sense to act tribal when you are divided into warring tribes. But the real question is, how constrained are people in acting this way, when it would go against their own self-interest?
The ingroup-outgroup bias might be reducible to simpler cognitive biases. The first is that people generalize or stereotype people, places, and things into various categories. People prefer to chunk continuous measures into discrete types. It’s easier for us to think in chunks. This is one of the facets of psychological essentialism, where even young children intuitively assume that concepts like bird, female, or red have an underlying discrete reality, even without evidence. This pervades much amateur thinking about biology (e.g. “A virus must be alive or not, so which is it?”).
Another cause of tribalism is the acquisition and enforcement of social norms, which are themselves properties of groups. The difference between adherence to social norms and group loyalty can be subtle (or even identical) when groups are identified by social and cultural norms. Human social groups are often not defined by actually knowing the individuals in the group, but rather by markers such as dress (football jerseys, suits, or tribal garb), language or dialect, or by belief systems such as religions or political parties. In these cases, each individual could switch groups. So I could become a Dallas Cowboys fan, or join Islam, or become a citizen of Spain. But in other cases, group membership is more fixed, as when it is defined by biological traits like race or sex. But here again, the neat-and-tidy discrete lines between ingroup-outgroup may break down. A person might appear to be more “racist” towards a homeless black person than towards a black person wearing a suit and tie. In other words, being a certain race or religion is perhaps not an absolute cue of being an outgroup, it is just one of many cues that the person’s brain uses to decide how socially distant an individual is to them. The subconscious machinery might function to assess questions like: How likely is this person to be an ally to me in a conflict? If a tribal conflict breaks out, what will it be about? And will we be on the same side or opposing sides?
If I’m correct about this, then people’s ingroup-outgroup bias should itself be highly malleable. In a situation where I think I might be caught in a political war, I will suddenly see my country as my ingroup. But if I see a religious war brewing, then I might then see my religion as my ingroup. That is, our loyalties should be context-dependent. If you and I are different races/sexes/religions in one tribe, we will largely forget that when we join forces to fight an army from a completely different tribe. Then when the tribal war is over, those race/sex/religion issues will re-emerge.
But I might be wrong. Perhaps in our evolutionary history, groups were so stable and homogenous that such strategic social positioning was never necessary. Perhaps social group identity is more real and stable than I am imagining. I don’t know. Do people have one set of social instincts for regulating inter-group conflict and another set for regulating intragroup conflict? Or do we have just one set for both situations?
Is the ingroup-outgroup bias a separate evolved cognitive bias serving its own evolutionary function? Or is it merely a byproduct of intuitive essentialism and adherence to social norms? (Or is it somewhere in the middle? 😛 )
One way to answer these questions would be to describe the distribution of perceived social distances in a person’s social network. Are there two discrete humps? One for ingroup or one for outgroup? Or is there a smooth curve? How does this distribution differ in different societies around the world? Social distance itself could be measured either objective or subjectively– and that difference itself might be interesting. Perhaps this has all been done already? Let me know in the comments.
I need to read more of the literature about human social networks. But so far I don’t see people answering the questions that fill my mind.
My own concern is more that the ingroup-outgroup distinction sadly oversimplifies the conversation about human cooperation. If you give people simple questions (ingroup or outgroup?) you will get simple answers. But real social life is full of nuance, as we all recognize simply from being human. Ponder the strategic logic underlying these phrases:
“A friend in need, is a friend indeed.”
“The enemy of my enemy is my friend.”
“A friend to all, is a friend to none.”
There are even less obvious facets to the adaptive design of human friendships. For example, two friends who are better friends with each other than with you, are both less ‘valuable’ to you than two friends that don’t know each other (see the evidence). Even nonhuman animals appear to make strategic social decisions about bonding that are contingent on their shifting place in a social network. In baboons, females are more willing to make new friends when a close relative dies. In vampire bats, those individuals who form more non-kin relationships do better when their kin partners are unavailable to help them.
Many primates not only create and manage their own ‘ingroups’ but they probably manage each social relationship individually according to what is happening in all their other relationships (much like in a market). And moreover, in each population, there might be different social management strategies that are under frequency-dependent selection. Rather than focusing so much on ingroup-outgroup bias in humans, these are the kinds of design principles of social cognition that, in my opinion, would be more illuminating to investigate.