The Righteous Mind: Religion, Cooperation, and Evolution

I’ve read a book.

In perfect candor, this is a feat I’ve accomplished once or twice in the past, but it never fails to stoke a certain sense of accomplishment and smug self-adulation. After all, I’ve forsaken untold hours of watching TV and playing video games in favor of an identical amount of time spent turning pages and reading words. Basically, the sort of opportunity cost only saints are meant to bear.

In this case, the book came with the additional reward of containing a surfeit of the sort information the late French pedant Claude Levi-Strauss might have called “good to think”.

Without further delay, the book: The Righteous Mind, by social psychologist Jonathan Haidt. I won’t go so far as to give an exhaustive review – suffice it to say that the book was good and you ought to read it, providing as it does a succinct and provocative run-down of research into the psychological underpinnings of our moral and political inclinations.

Of more interest here is an argument Haidt makes about the relationship between religious belief and cooperation. It’s an interesting proposal, not entirely original to Haidt, that gets to the ruddy flesh of a longstanding cultural and evolutionary puzzle.

Today, depending on the developmental status of the country in question, somewhere between 74% and 44% of people live in large urban centers – surrounded on all sides by strangers. Most people buy their groceries from strangers, have their appliances serviced by strangers, and are policed by strangers. Almost universally, these urban centers are part of larger polities, again, comprised primarily of people largely unfamiliar to one another. Nevertheless, these people usually manage to do a better than middling job of getting along. Twelve thousand years ago, human social interaction unfolded at the level of the tribe or band – groups structured around bounds of genetic kinship and personal familiarity.

For decades, the question of how people managed to make the transition from tiny bands of closely related individuals to larger political entities, full of relative strangers, has remained unresolved. Kin selection and reciprocal altruism are likely foundational, but these elements alone are widely considered to be insufficient to explain humanity’s unique capacity to get along across wide genetic chasms. Haidt makes a compelling case that religion has played a role in helping humans to cooperate with unrelated neighbors, a critical step in the process of building large-scale societies.

Now, it’s important to be precise about this hypothesis. Haidt and those who share his view are not arguing that religious beliefs have much of a role to play in facilitating large-scale cooperation in modern societies. That has much more to do with the rule of law and the adoption of liberal, democratic values (in the classical sense) than it does with religious belief. Instead, the argument here is that certain varieties of religious belief may have helped bridge the gap between communities in which kinship, familiarity, and reciprocity could explain cooperation in communities too large for those rudimentary elements to sustain amicable, productive relationships.

Punitive Gods and Stable Cooperation

Recently, work in experimental economics, anthropology, and psychology has revealed that punishment plays an instrumental role in building and enforcing stable patterns of cooperation. People are more likely to cooperate and behave honestly when they perceive violations of established norms of interaction come with a reliable threat of reprisal. That is, people will be most likely to do the right thing if they believe cheating someone out of their hard-earned cash or trespassing a sacred cultural taboo will earn them some sort of stiff rebuke. Punishment is particularly effective when it involves a cost to the punisher, meaning that the amount of time, effort, or material resources forsaken in favor of punitive actions intensify their efficacy.

A reliable threat of punishment requires two things. First, it requires that people be convinced that their misdeeds will not go unnoticed. Someone – or something – needs to be watching. Second, it demands that people assume getting caught will spell serious trouble. If your fellow citizens or regulatory agency or tribe members catch you doing something nasty, they are going to respond by going out of their way to make you very uncomfortable.

Lately, researchers like Ara Norenzayan and Joseph Henrich have begun to argue that certain forms of religious belief can fill both roles. People with strong beliefs in an omnipotent god – one capable of and interested in punishing them for their transgressions – seem to be more likely to cooperate with one another. The basic thinking is that, even in situations where they can feasibly disguise their crimes from their neighbors, people are less likely to behave selfishly or maliciously if they think a powerful deity is watching them. They might not be punished by their peers, but they won’t escape the inevitable wrath of god – here on earth or in the afterlife.

If true, this could go a long way toward explaining the historical success of the monotheistic faiths that spread out of the Levant and the Arabian Peninsula. These groups were able to outcompete their neighbors and, purely as a product of cultural selection, the underlying belief systems spread.

Haidt wants to take this argument a little further. He argues that religion and prosociality coevolved in response to multilevel (i.e. group) selection in the late Pleistocene and throughout the Holocene. The central claim – that certain forms of religious belief can benefit the groups that entertain them by giving them a competitive advantage in patterns of ingroup/outgroup competition – is perfectly in line with the aforementioned research. Groups that cultivate the kinds of religious beliefs that encourage social cohesion and discourage free-riding and defection should be able to outcompete groups with religious beliefs that encourage independence or fail to incentivise cooperation and/or punish defection.

Under these conditions, religious belief may have carried certain selective advantages. Those groups with the highest levels of prosocial religiosity might have regularly outcompeted their less pious neighbors. Any existing psychological traits that encouraged supernatural belief – such as false agency attribution, elevated neurochemical feedbacks from ritual experience, credulous acceptance of traditional knowledge, or strong tendencies to monitor and enforce social norms, for instance – would grease the skids for participation in religious ceremonies that nourish social cohesion and fear in the sort of punitive gods that discourage defection. If these traits have a genetic basis, they may have been favored by selection.

This is all very reasonable. It makes sense that certain cultural traits could confer advantages on the groups that possess them. In a way, it’s almost obvious. Groups with bow and arrow technology would have exercised a strategic advantage over those still using atlatls and darts – and the archaeological record bears this out. If there is an underlying psychological architecture – significantly correlated in its expression with the inheritance of genetic information – that promotes or enhances people’s ability to engage in these cultural practices, then there could very well be forces of biological selection at work in shaping relevant patterns of gene-culture coevolution.

In attributing the invention and spread of prosocial religions to group selection, Haidt’s error is twofold. First, he is wrong to attribute the evolution of any underlying psychological traits to group selection. Second, he ignores the ways in which cultural group selection – entirely distinct from biological group selection – can shape the spread of specific prosocial traditions. I’ll deal with these problems sequentially.

Prosocial Group Selection

First, the matter of group selection. In attributing the evolution of psychological traits that foster religious prosociality to biological group selection, Haidt’s foremost mistake is in thinking that because religiosity would have benefited the group, any selection for the underlying traits must have been “for” that purpose. This is an argument that exposes the potential risks of adaptationist reasoning. It claims that because a trait confers an advantage for a certain task – in this case, intragroup cooperation and intergroup competition – it must have arisen specifically for that reason. This is precisely what evolutionists mean when they say a certain trait is an adaptation for a specific challenge.

Let’s consider four plausible candidates for psychological traits that might foster religious belief:

  1. False agency attribution
  2. Strong tendencies to accept traditional knowledge
  3. Strong tendencies to establish and police social norms
  4. Elevated neurochemical response to ritual participation

All four could easily have evolved absent of and prior to the development of supernatural belief systems. As Daniel Dennett and Michael Shermer have been keen to point out, something like an evolved predilection toward false agency attribution – automatic type-2 error, as it were – would be useful independent of any beliefs in an underlying supernatural order: better to think the rustling in the grass is a lion when it’s just the wind than to dismiss it as just the wind when it’s really a lion.

A strong tendency to credulously accept traditional knowledge could likewise be favored in an environment devoid of religiosity. The causal mechanics of most cultural innovations – be they social institutions or hunting technologies – are frequently opaque to their users because they emerge from long-term processes of collective engineering – cycles of trial-and-error and social learning that unfold over generations. In most environments, it pays to accept traditional knowledge about prey behavior and hunting techniques without question, because these practices entail generations of accumulated wisdom. Under the conditions that prevailed throughout the Pleistocene, blind acceptance of extant information – foraging techniques, ritual practices, supernatural beliefs – would have often been a cheap route to success.

Again, any evolved inclination to monitor and enforce social norms could be selected on a thoroughly irreligious landscape. This is because a tendency to keep an eye on your neighbor and respond punitively when they step out of line is good way to overcome the free-rider problem. Living in larger groups and engaging in behaviors that require collective effort is a precarious business, because individual strategies that exploit the fruits of cooperation will rapidly come to dominate the system and cooperation will deteriorate. Groups can get around this problem by imposing heavy costs on the decision to become a free rider, encouraging the evolution of traits that help their bearer both reliably detect and consistently punish cheaters. Here, the only necessary condition is that the benefits of detection and enforcement exceed their costs.

Finally, evolving elevated neurochemical responses to ritual participation could simply be a matter of bootstrapping extant mechanisms for reinforcing social bonds. Any system that used neurochemical feedbacks to reward social engagement – pair bonding, grooming, food sharing – could be coopted (exapted, in evolutionary parlance) to encourage ritual engagement. That said, there is a case to be made that these feedbacks could be intensified and elaborated by the novel selective pressures associated with group living and increasing levels of supernatural belief.

Only in the latter two cases – tendencies to establish and police social norms and elevated neurochemical responses to ritual participation – do we see compelling candidates for group level selection. Both have the potential to confer distinct advantages on the groups that exhibit them, simultaneously stabilizing and intensifying social bonds. But to extend from that recognition – that those traits benefit groups – to the stronger assertion that they evolved for that purpose is to fall into a rather dastardly teleological trap.

Fundamentally, this is where the real trick of adaptationist thinking lies. It’s not a tool for actually discovering adaptations. Rather, it’s a tool for discovering plausible hypotheses about adaptation. Recognizing that a trait might be good tool for accomplishing a certain task doesn’t mean it actually exists for that purpose. Used correctly, adaptationism provides a powerful heuristic for thinking about evolution and posing useful questions about the underlying processes. Used incautiously, it can lead us into traps like erroneous claims about group-level adaptation.

This seems to be what’s going on with Haidt’s claims about the traits that foster belief in prosocial religions. Certainly it’s possible for the benefits of monitoring and enforcing social contracts to accrue at the level of the group – but it’s also true that these things could be rewarding at the level of the individual. Participation in cooperative hunting and foraging, for instance, can have the effect of both increasing per capita caloric returns and buffering against day-to-day uncertainty. Consequently, individuals have strong incentives to make sure cooperative efforts run smoothly – keeping an eye on each other to make sure patterns of reciprocity are enforced. If individual experience benefits from cooperation that exceed the costs associated with seeking out and punishing defectors, there is no need to invoke group-level processes to explain the evolution of those traits.

Much the same can be said for ritual response. If traits evolve that make people feel good when they are participating in community rituals, they could realistically be expected to stabilize and reinforce intergroup relationships and foster a sense of shared identity useful in intergroup competition. This could benefit the group enormously. But again, all the explanatory work can be done at the level of the individual. If, as I’ve speculated, the relevant feedbacks rely on existing neural architecture associated with pair-bonding and food-sharing and the like, we’ve already gone a long way toward establishing mechanisms for reinforcing ritual observance. Then it may simply be the case that a tendency to feel an increased sense of shared identity may encourage cooperative behaviors that ramify into an array of downstream benefits: for instance, your neighbors may come to recognize you as a reliable collaborator, preferential sharing with you resources and knowledge that will aid in survival and reproduction on the belief that you are likely to reciprocate.

Of course, all of this is speculative – particularly the existence of traits related to elevated neurochemical responses to ritual participation. The point here hasn’t been to disprove Haidt’s claim about group selection, but rather to illustrate plausible routes by which individual-level selection could sculpt traits that look like group-level adaptations. All of the decisive work should be done empirically: observe, experiment, repeat – ad nauseum. But the simple fact that individual/gene-level selection are established mechanisms suggests strongly in their favor – one should convincingly demonstrate that they are not at work before seeking to anchor explanations in other, more spurious, processes.

The Case for Stabilizing Selection


When it comes to explaining the existence of traits that encourage religious prosocialitiy, there’s really no need to invoke group-level processes. To begin with, there is no reason to think that belief in a punitive god and participation in socially enriching rituals depends on an entirely distinct psychological architecture from other forms of supernatural belief. The underlying feedbacks could work just as well for promoting beliefs that involve the workings of capricious forces arising from rocks and trees.

More fundamentally, there is reason to think the psychological preferences around which the pantheon of religious traditions – both historic and prehistoric – have been scaffolded rely on traits that were selected for prior to the advent of punitive gods and deeply prosocial belief systems. That doesn’t mean selection stopped there. Rather, my inclination is that punitive gods and prosocial belief systems would likely have resulted in a form of stabilizing selection, wherein deviant traits like skepticism or a disinclination to participate in ceremony were actively penalized. In the language of economics, failure to conform to religious tradition could have entailed a massive opportunity cost. Being an agnostic in the neolithic Levant might have dropped your reproductive success to zero, or at least severely limited it through decreased access to alloparenting, often considered essential to the business of raising costly human offspring.

Daring momentarily to tread the fraught rhetorical ground of the anecdotal, allow me to use a point of personal history. I was raised in a religious family, taught from an early age that I had been fortunate enough to be born into the one true religion – i.e., the Church of Jesus Christ of Latter Day Saints. In the LDS faith, outgroup interactions are both implicitly and explicitly discouraged. Not prohibited, mind you, but there is strong consensus that it’s best to spend your time with coreligionists and, more importantly, marry within the church.

Fortunately, I was also born into a more or less pluralistic society, where the only tax on my growing disbelief was that my former coreligionists didn’t want to hang out with me and Mormon women were much less likely to consider me a suitable mate. I found new friends outside of the Church and eventually married a fellow heathen.

But consider the circumstances faced by a similarly skeptical human in the Bronze Age Levant. Deigning to spend their time designing new farming implements or inventing more effective forms of animal husbandry – or simply drinking in the woods – they regularly fail to show up for religious ceremonies. They don’t express much confidence in the existence of a punitive god and, in any case, seem to place little stock in such a being’s will. Over time, his neighbors begin to suspect his commitment to the group. They wonder if a man who doesn’t believe in the retributive power of god can be trusted to act honorably. Fathers and mothers discourage their daughters from considering him a suitable mate. Eventually, networks of social pressure have the effect of dropping his reproductive prospects to zero within the group.

Unfortunately, he doesn’t have much chance of finding sympathy in a neighboring group. Not only are they equally religious – and therefore likely to be equally suspicious of his irreligious tendencies – they also maintain a mutually hostile posture toward people of his ethnicity. They might greet him with death or enslavement. Sadly, that means his reproductive prospects have dropped to zero, period.

If I’d been born in a religious tradition like the Mormon church, but on a social landscape like the ancient Levant – where religious belief was ubiquitous but heterogeneous, draw along the lines of city-states with specific patron gods – my prospects for ever siring offspring would have been similar. And this is all you need for the proliferation of traits that favor prosocial religiosity. These traits confer an advantage on the bearer by making them more amenable to engagement in a social network where, critically among other benefits, they will be more likely to secure a mate and raise children with the additional help afforded by relatives and neighbors. But more importantly, the absence of these traits imposes an incredibly steep tax. Consequently, the range of phenotypes narrows to a more ardently pro-religious mean.

Now, this does ultimately have the effect of benefitting the group. More group members are more willing to entertain beliefs in the supernatural and participate in the associated rituals and ceremonies. If this god is properly punitive and the associated religious activities happen to reliably encourage internal cohesion, group members may have an easier time cooperating with each other. This would make the group itself more likely to outcompete neighboring groups with less prosocial beliefs. The group will grow and their beliefs will spread.

At no point does it make sense to attribute this to group level adaptation. All the important selective work has been done at the level of the individual.

Which brings us to the second flaw in Haidt’s reasoning. Punitive gods, properly done, seem like strong mechanisms for policing the unseen. However, there is no reason to image those belief system emerged as a direct product of biological evolution – the invention and proliferation of those beliefs doesn’t demand any specific gene-level selection. All the important work can be done at the level of social learning and cultural transmission, as beliefs that encourage intragroup cooperation flourish at the expense of more individualistic strategies in grander patterns of intergroup competition. This opens the door to sensible discussion of cultural group selection, because we’re talking about the differential success of groups based on the proliferation of cultural traits like  traditions, institutions, and beliefs.

This is quite distinct from biological group selection, where the traits of individuals are thought to be a product of the benefits they confer on the group. Instead, cultural group selection concerns the success or failure of beliefs and ideas that exist and spread by virtue of human interaction. Culture doesn’t exist without groups – there is no such thing as a culture of one – so it makes sense to think of cultural traits, by their very nature, as features of groups. Humans acquire cultural traits individually, but once acquired, they aren’t necessarily subject to the same ruthless calculus as biological traits. They can be beneficial or costly. Plenty of cultural traits are adaptive, aiding people in their quest for food and mates, but others are clearly maladaptive – hurting their prospects for survival and inhibiting their ability to reproduce in whatever environment they find themselves. The only real requirement is that they appeal to the suite of preferences – some learned, others evolved – that bias people toward certain flavors of thinking.

Inasmuch as there is a compelling case to be made that there are adaptations that grease the skids for religious belief, there is no similar case to be that religious beliefs are themselves adaptations – at least not in the strict biological sense. However, as cultural software running on biological hardware, they can be conceived of as cultural adaptations – it just depends on how beneficial they prove to be for the people who adopt them.

Scaffolded on a pre-existing architecture of evolved psychological preferences, belief in omniscient, punitive gods may very well have had a role to play in fostering large-scale cooperation in early Mesopotamian and Levantine city-states. Thereafter, they may have modified the selective landscape, eliminating some variation in the underlying traits (false agency attribution, credulous acceptance of traditional knowledge, monitoring and enforcing social norms, and neurological feedbacks from ritual observance). But nowhere does the plausibility of these arguments offer succor to the hypothesis that religious beliefs are themselves adaptations, much less the notion that they arose as a product of selective forces acting at the level of the group.

Additional reading:

Big Gods, Ara Norenzayan

The Secret of Our Success, Joseph Henrich

Genes, Memes, and Human History, Stephen Shennan

Foundations of Human Sociality, Joseph Henrich, Robert Boyd, Samuel Bowels, Colin Camerer, Ernst Fehr, and Herbert Gintis

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s