Belief in moralistic gods makes people generous—towards coreligionists

I’ve been meaning to write about this paper for some time, but it’s fallen into my backlog of 1000-odd draft posts that I almost never look at. However, I found a printout in my daypack, and so will try to describe it briefly, for the results are somewhat hard to interpret.

The paper, in a January, 2016 issue of Nature, is by Benjamin Grant Purzycki et al. (reference and free download below). The authors’ motivation is this: why, in a world in which we’re far removed from many of the people we interact with, do we still behave nicely and practice reciprocity? After all, we’re not related to those people, so kin selection can’t apply, nor do we live in small bands with them, in which reciprocity is de rigueur because you constantly deal with your groupmates and can’t afford to get a bad reputation. Why don’t we just cheat?

There are many answers, including Peter Singer’s combination of evolution and rationality described in his book The Expanding Circle, and my own theory (which is mine) that one’s reputation also extends distantly these days. (If you cheat in your internet business, you’re not going to do very well.) But Purzycki et al. had another hypothesis involving religion and God. The authors suggest that if you believe in a “moral and punitive god,” the more likely you’d be to distant people who belong to your religion. After all, that god is watching you and will punish you if you’re not generous or fair.

To test whether this hypothesis was true, the authors did a laboratory study with people from eight diverse communities throughout the world, holding a variety of different belief systems. Each society (shown in the table below) had both a general, moralistic god as well as a “local god” (for Christians, the Big God and the Virgin Mary respectively).

Screen Shot 2016-06-08 at 12.26.04 PM

To test the effects of a moralizing god (as well as less omnipotent or omniscient local deities), the authors played two games with each of the 591 subjects, games described in the diagram below. In each case the participant was given 30 coins, which stood for stuff like money or corn, so your allocation reflects what you really got (and the stuff you gave away was also really given away). Subjects were also questioned about the degree to which local gods and Big gods cared about morality (local gods always cared less).

In each test, you allocated your 30 coins between two labeled cups. First, you were asked to mentally choose which cup you wanted to put one of your coins into. Then you were asked to roll a die on which three faces had a single color, and the other three faces another color. If a specified color came up, the subject was asked to put a coin into the cup that was mentally chosen. If the other color came up, you were asked to put a coin into the other cup. Each of the cups were labeled as to the nature of the recipient. The two choices were these:

Self game.  One cup indicated as “self” (i.e., the player); the other as “distant co-religionist” (the paper doesn’t say how distant).

Local co-religionist game. One cup indicated as “local co-religionist”, presumably someone in your community; the other cup indicated as “distant co-religionist”.



Under these conditions, of course, 50% of the coins would, on average, go into each of the two cups. But you could bias the results by cheating, either by not throwing the die, or putting coins into cups contra what the die dictated. Nobody watched the participants, though they thought their gods were watching.

The results:

  • When results were classified, as in the graph below, by answers to the question “does your god punish you,” people put fewer coins in the distant co-religionist cup (i.e., what the subjects themselves didn’t get), than they did when they thought the god was more punitive (bars left to right in figure).  The number of coins allocated toward distant co-religionists compared to the alternative choice increased by a factor of 4.8 in the self game and 5.3 in the local co-religionist game.
  • On average, then, players gave two more coins to the distant-coreligionists when they thought their god was very punitive than when they didn’t know whether their god was punitive. In other words, they were nicer to distant members of their faith when they thought their god was watching and would punish selfishness.
  • Note, though that in all cases the average number of coins given to distant co-religionists was less than half of what you expect under the 50:50 distribution, so even a very punitive god didn’t enforce perfect fairness (the error bars in the latter case do encompass 15 coins, though).
  • Finally, looking at the effect of other variables on the responses, including stuff like economic conditions, education, and number of children, the authors found that only the degree of punitive-ness or omniscience of the moralistic god was correlated with the results, while the effect of other variables, including local gods, wasn’t significant.



(From figure): Allocations to distant co-religionists increase as a function of moralistic gods’ punishment. Punishment indices are mean values of a two-item scale (see Supplementary Information section S2.3.2). Error bars represent bootstrapped (1,000 replications) 95% confidence intervals of the mean. Histogram labels are sample sizes per category.


The authors conclude that their results have big implications for understanding cooperation between distant people:

These results build on previous findings and have important implications for understanding the evolution of the wide-ranging cooperation found in large-scale societies. Moreover, when people are more inclined to behave impartially towards others, they are more likely to share beliefs and behaviours that foster the development of larger-scale cooperative institutions, trade, markets and alliances with strangers. This helps to partly explain two phenomena: the evolution of large and complex human societies and the religious features of societies with greater social complexity that are heavily populated by such gods. In addition to some forms of religious rituals and non-religious norms and institutions, such as courts, markets and police, the present results point to the role that commitment to knowledgeable, moralistic and punitive gods plays in solidifying the social bonds that create broader imagined communities

But I don’t buy that for a number of reasons. People are often altruistic towards those of different faiths. Further, the experiment was done in a lab, and it’s not clear how well this short-term experiment translates toward behavior in real life. Finally, the authors use the data to explain why some religions spread at the expense of others (people of those faiths help each other, even when far away), but there is no control study about how people behave towards other people who don’t share their faith.

In other words, this experiment, while showing that the idea that a punitive god is watching may make you give more coins to distant coreligionists than you otherwise would, doesn’t say much to me about the author’s Big Thesis. Yet the paper was published in Nature, one of the world’s most prestigious and hard-to-get-into journals. I can conclude only that Nature‘s editors like the result. Perhaps they have belief in belief.


Purzycki, B. G. et al. 2016. Moralistic gods, supernatural pubishment and the expansion of human sociality. Nature 530:327-330.



  1. Heather Hastie
    Posted June 8, 2016 at 2:19 pm | Permalink

    Even if it’s true, it doesn’t make a punitive god the right approach. Personally, I favour doing the right thing because it is the right thing. Teaching values and ethics doesn’t need a god figure to reinforce it. It’s easy enough to show that doing the right thing is better without the supernatural rubbish, which is, of course, always vulnerable to being shown to be wrong.

    • Mark P
      Posted June 8, 2016 at 2:31 pm | Permalink

      Oh Heather, how silly of you! How can you possibly know right from wrong without referring to the Bible or Koran or some other sacred text?

      • Heather Hastie
        Posted June 8, 2016 at 3:04 pm | Permalink

        Oh well, I used to be Christian so I suppose it’s still part of me from then!🙂

        There are a lot less people I’m supposed to disapprove of now, but my failure to disapprove of them even when I was a Christian was one of the main reasons religion failed for me. How can something that is supposed to be about love include so much hate?

        • Ken Elliott
          Posted June 8, 2016 at 4:33 pm | Permalink

          I’ve been reading Dan Barker’s new book, “God: The Most Unpleasant Character in All Fiction”, which has helped me understand the enormity of immorality in the Bible. It boggles the mind how anyone could think it the source of morality, or the gateway to morality through the vicious, murderous, rancorous God depicted therein.

  2. Posted June 8, 2016 at 3:11 pm | Permalink

    “The authors suggest that if you believe in a “moral and punitive god,” the more likely you’d be to distant people who belong to your religion.”

    I think you may have left out “nice” or “generous” or some such term between the bolded words.

    • Jenny Haniver
      Posted June 8, 2016 at 3:22 pm | Permalink

      You beat me to the punch, I just wrote this:
      Perhaps I’m developing syntactic aphasia, but I don’t understand this sentence: ‘The authors suggest that if you believe in a “moral and punitive god,” the more likely you’d be to distant people who belong to your religion.’ Specifically, I can’t parse the clause, “the more likely you’d be to distant people who belong to your religion.” I can’t find the complement to “be”. Seems like there should be one, but now I’ve confused myself even more than I was already.

      Glad someone else found an anomaly and so perhaps I’m not developing syntactic aphasia after all.

  3. rickflick
    Posted June 8, 2016 at 3:17 pm | Permalink

    “Why don’t we just cheat?”

    It seems to me we DO just cheat. But, I think it’s calculated. In real life, people litter the streets and avoid paying taxes even when they know it’s not a very socially conscious thing to do. Whether religious or not, people will often do the calculation (mental whispers) to see whether the likelihood of being caught red handed and the size of the reward tilt in their favor. Some of the devout will figure in how much of a sin it really is (does the Catholic God really, really hate abortion?), and the rest of us will be watching the rear view mirror before deciding to push the pedal to the metal. For big stuff, like armed robbery or murder, the down-side is just too deep to give a second thought.

    • boggy
      Posted June 9, 2016 at 8:26 am | Permalink

      Frank Sinatra’s mother was a Catholic a midwife and an abortionist so I think we can say with confidence that God does not hate abortion.

  4. Vaal
    Posted June 8, 2016 at 3:22 pm | Permalink

    If it were the case that believing we are monitored by a God helps us act morally, it’s intriguing to think how this is (perhaps) in the process of being replaced
    by a reality-based monitoring system: the Internet. We continually learn more about how thoroughly our actions are recorded and monitored by Google and many other internet entities. (And not just internet activity – all sorts of other activity as more of our actions, shopping etc, interact with databases). Some don’t see this as a bad thing, sort of like being under the eye of God.

    I think Sam Harris (if I remember right) is one of those who suggest that the increasing loss of privacy may be balanced by the positive behavioral effects of knowing we are being “watched.”

    • Posted June 16, 2016 at 2:20 pm | Permalink

      Yes, we can control shoplifting with CCTV and no longer need this function to be performed by a punitive G*d with his severe side effects and ever-decreasing efficiency.

  5. Posted June 8, 2016 at 4:07 pm | Permalink

    This study looks like it took an immense amount of work, including a lot of work by researchers at my own school (UBC). So I’m kind of sad to have to say that I don’t think their study shows much of anything. They seem to do a fair amount of modelling in the paper, but then don’t really do anything with it in the Nature article. And the results that they end up focusing on don’t really support what they are trying to claim.

    The main issue here is that their data do *not* show that “allocations to distant co-religionists increase as a function of moralistic gods’ punishment” (Figure 2). The point estimates may increase, but the authors don’t factor in the uncertainty in these estimates. They do give helpful error bars in the figure, and while it’s technically incorrect to compare these bars pairwise to detect an effect, the reported uncertainty is so great here that there is no way the claimed effect is supported by the data.

    Also, I take serious issue with those “increase by a factor of 4.8 (or 5.3)” figures. To me, the way they derive these figures is completely unnatural. And yet again these derivations ignore the uncertainty in the point estimates, which when factored in don’t support the claim of an effect.

    The most interesting part of the paper I think is at the end when they actually talk a bit about the models they built. Here is the only place I find evidence that the threat of a moralizing, punisher god increases the likelihood (odds, actually) that a coin is given to a distant co-religionist.

    The more articles I see coming out of Nature, the less impressed I am. It seems that the journal tries to just present snappy, scientific sound-bytes, with no regard for the statistical basis of the results. And moreover, the journal seems to actively encourage oversimplifying results to the point that they no longer actually hold up. Here, we see the focus on the results that are actually supportive of no effects precisely because they are easy to understand (i.e. don’t require any knowledge of modelling, etc.). But just because something is easy to understand, doesn’t make it right!

  6. Kevin
    Posted June 8, 2016 at 4:34 pm | Permalink

    This sounds like a case of:

    Us and them
    And after all we’re only ordinary [people who prefer to like people who believe in something like what we believe]

  7. Posted June 8, 2016 at 6:15 pm | Permalink

    This study makes me shudder. We’ve known forever that coreligionists are more charitable to coreligionists, whether near or far. (Why are we looking at charity as a form of morality?) So, charity of religionists is not directed to everyone and is, therefore, not nearly as commendable as some would have us believe. We also know that religionists are charitable to non-religionists when proselytizing, thereby trying to increase the fold. I may be wrong, but I think there are relatively few religious charities that are made available to people outside the faith.

    I dislike the assumption that belief in a deity is required for there to be morality
    and that, especially, the deity believed in must be a punitive one. The many ancient versions of the Golden Rule would make it clear that morality can be extended to others based on knowledge of yourself, not by the fear of punishment by a deity. I don’t even think this is a replicable, valid or useful study.

  8. Scott Draper
    Posted June 8, 2016 at 6:45 pm | Permalink

    “so kin selection can’t apply”

    Why not? We don’t have any build-in means of identifying kin, so kin selection must work off of some sort of surrogate. Surely kindness towards strangers could be a result of this kin-detector being fooled in some way.

  9. DiscoveredJoys
    Posted June 9, 2016 at 2:37 am | Permalink

    I’m reading ‘Surfing Uncertainty: Prediction, Action and the Embodied Mind’ by Andy Clark which proposes that how we get to know the world relies on

    a simple but remarkably powerful trick or stratagem. That trick is trying to guess the incoming sensory stimulations as they arrive, using what you know about the world. Failed guesses generate ‘prediction errors’ that are then used to recruit new and better guesses, or to inform slower processes of learning and plasticity. The book suggests that ‘prediction, errors and actions’ overturns the concept of ‘picking good data out of noise algorithm’ which underpins a lot of cognitive current concepts. The book also explains how ‘prediction/error/action’ could generate the feeling of agency and consciousness (but I still have to thoroughly read those bits!).

    But the connection to the post is that I propose if we learn to expect kindness from our kin (genetic altruism), as babies and children, then as we move out into the wider world our ‘predictions’ against which we manage our perceptions may by default expect kindness from others and predispose us to be kind to them.

    Of course that also means that if we have learned to ‘fear god’ then that will be built into our predictions too, making it difficult to unpick because ‘the prediction of god’ rarely encounters error signals (at the subconscious level) leading to revised models and predictions.

    • Torbjörn Larsson
      Posted June 9, 2016 at 3:22 am | Permalink

      That is along the mechanism I believe could be at play. But the details depends on the age when people start to display the trait. Here is what I composed, before I saw your analysis:

      why, in a world in which we’re far removed from many of the people we interact with, do we still behave nicely and practice reciprocity? After all, we’re not related to those people, so kin selection can’t apply, nor do we live in small bands with them, in which reciprocity isde rigueur because you constantly deal with your groupmates and can’t afford to get a bad reputation.

      As long as people have a place to stay, they will interact with familiar people (neighborhood, internet locales). The default in such interactions is set by a Nash equilibrium, of reciprocity (modified by slight, i.e. a little bit of random, punishment).

      Our biology should be wise to that equilibrium. Indeed I think could be, since such behavior can be seen in toddlers, if I remember correctly.

      • Torbjörn Larsson
        Posted June 9, 2016 at 3:27 am | Permalink

        I should also add that it isn’t only a mechanism of showing communal behavior in the neighborhood, since distant interactions are considered in the experiment (and in reality).

        As long as we return to familiar places (routes, stays, et cetera), we risk recurring meetings that may or may not be remembered. Then the Nash equilibrium applies.

  10. DiscoveredJoys
    Posted June 9, 2016 at 2:39 am | Permalink

    Rats! The bit ‘a simple… …plasticity’ should be indented but everything shouldn’t be indented.

  11. Posted June 9, 2016 at 9:42 am | Permalink

    Thanks for posting this, very interesting. Especially the last paragraph. 🙂

%d bloggers like this: