k10063Despite my disagreements with a lot of its details, I thoroughly enjoyed Ara Norenzayan’s Big Gods.  The book posits an explanation for the current global dominance of the big three Abrahamic religions: Christianity, Islam, and Judaism.

Instead of the “quirks of history & dumb luck” explanation offered in Jared Diamond’s Guns, Germs, and Steel, Norenzayan suggests that the Abrahamic religions have so many adherents today because beneficial economic behaviors were made possible by belief in those religions.

Here’s a rough summary of the argument: Economies function best in a culture of trust.  People are more trustworthy when they’re being watched.  If people think they’re being watched, that’s just as good.  Adherents to the Abrahamic faiths think they are always being watched by God.  And, because anybody could claim to believe in an omnipresent, ever-watchful god, it was worthwhile for believers to practice costly rituals (church attendance, dietary restrictions, sexual moderation, risk of murder by those who hate their faith) in order to signal that they were genuine, trustworthy, God-fearing individuals.

A clever argument.  To me, it calls to mind the trustworthiness passage of Daniel Dennett’s Freedom Evolves:

When evolution gets around to creating agents that can learn, and reflect, and consider rationally what they ought to do next, it confronts these agents with a new version of the commitment problem: how to commit to something and convince others you have done so.  Wearing a cap that says “I’m a cooperator” is not going to take you far in a world of other rational agents on the lookout for ploys.  According to [Robert] Frank, over evolutionary time we “learned” how to harness our emotions to the task of keeping us from being too rational, and–just as important–earning us a reputation for not being too rational.  It is our unwanted excess of myopic or local rationality, Frank claims, that makes us so vulnerable to temptations and threats, vulnerable to “offers we can’t refuse,” as the Godfather says.  Part of becoming a truly responsible agent, a good citizen, is making oneself into a being that can be relied upon to be relatively impervious to such offers.

I think that’s a beautiful passage — the logic goes down so easily that I hardly notice the inaccuracies beneath the surface.  It makes a lot of sense unless you consider that many other species, including relatively non-cooperative species, have emotional lives very similar to our own, and will like us act in irrational ways to stay true to those emotions (I still love this clip of an aggrieved monkey rejecting its cucumber slice).

Maybe that doesn’t seem important to Dennett, who shrugs off decades of research indicating the cognitive similarities between humans and other animals when he asserts that only we humans have meaningful free will, but that kind of detail matters to me.

You know, accuracy or truth or whatever.

Similarly, I think Norenzayan’s argument is elegant, even though I don’t agree.  One problem is that he supports his claims with results from social psychology experiments, many of which are not credible.  But that’s not entirely his fault.  Arguments do sound more convincing when there’s experimental data to back them up, and surely there are a few tolerably accurate social psychology results tucked away in the scientific literature. The problem is that the basic methodology of modern academic science produces a lot of inaccurate garbage (References? Here & here & here & here... I could go on, but I already have a half-written post on the reasons why the scientific method is not a good persuasive tool, so I’ll elaborate on this idea later).

For instance, many of the experiments Norenzayan cites are based on “priming.”  Study subjects are unconsciously inoculated with an idea: will they behave differently?

Naturally, Norenzayan includes a flattering description of the first priming experiment, the Bargh et al. study (“Automaticity of Social Behavior: Direct Effects of Trait Construct and Stereotype Activation on Action”) in which subjects walked more slowly down a hallway after being unconsciously exposed to words about old age.  But this study is terrible!  It’s a classic in the field, sure, and its “success” has resulted in many other laboratories copying the technique, but it almost certainly isn’t meaningful.

Look at the actual data from the Bargh paper: they’ve drawn a bar graph that suggests a big effect, but that’s just because they picked an arbitrary starting point for their axis.  There are no error bars.  The work couldn’t be replicated (unless a research assistant was “primed” to know what the data “should” look like in advance).


The author of the original priming study also published a few apoplectic screeds denouncing the researchers who attempted to replicate his work — here’s a quote from Ed Yong’s analysis:

Bargh also directs personal attacks at the authors of the paper (“incompetent or ill-informed”), at PLoS (“does not receive the usual high scientific journal standards of peer-review scrutiny”), and at me (“superficial online science journalism”).  The entire post is entitled “Nothing in their heads”.

Personally, I am extremely skeptical of any work based on the “priming” methodology.  You might expect the methodology to be sound because it’s been used in so many subsequent studies.  I don’t think so.  Scientific publishing is sufficiently broken that unsound methodologies could be used to prove all sorts of untrue things, including precognition.

If you’re interested in the failings of modern academic science and don’t want to wait for my full post on the topic, you should check out Simmons et al.’s “False-Positive Psychology: Undisclosed Flexibility in Data Collection and AnalysNais Allows Presenting Anything as Significant.”  This paper demonstrates that listening to the Beatles will make you chronologically younger.

Wait.  No.  That can’t be right.


The Simmons et al. paper actually demonstrates why so many contemporary scientific results are false, a nice experimental supplement to the theoretical Ioannidis model (“Why Most Published Research Findings Are False”).  The paper pre-emptively rebuts empty rationalizations such as those given in Lisa Feldman Barrett’s New York Times editorial (“Psychology Is not in Crisis,” in which she incorrectly argues that it’s no big deal that most findings cannot be replicated).

Academia rewards researchers who can successfully hunt for publishable results.  But the optimal strategy for obtaining something publishable (collect lots of data, analyze it repeatedly using different mathematical formula, discard all the data that look “wrong”) is very different from the optimal strategy for uncovering truth.


Here’s one way to understand why much of modern academic publishing isn’t really science: in general, results are publishable only if they are positive (i.e. a treatment causes a change, as opposed to a treatment having no effect) and significant (i.e. you would see the result only 1 out of 20 times if the claim were not actually true).  But that means that if twenty labs decide to test the same false idea, 19 of them will get negative results and be unable to publish their findings, whereas 1 of them will see a false positive and publish.  Newspapers will announce that the finding is real, and there will be a published record of only the incorrect lab’s result.

Because academic training is set up like a pyramid scheme, we have a huge glut of researchers.  For any scientific question, there are probably enough laboratories studying it to nearly guarantee that significance testing will provide one of them an untrue publishable result.

And that’s even if everyone involved were 100% ethical.  Even then, a huge quantity of published research would be incorrect.  In our world, where many researchers are not ethical, the situation is even worse.

Norenzayan even documents this sort of unscientific over-analysis of data in his book.  One example appears in his chapter on anti-atheist prejudice:

In addition to assessing demographic information and individual religious beliefs, we asked [American] participants to rate the degree to which they viewed both atheists and gays with either distrust or with disgust.

. . .

It is possible that, for whatever reason, people may have felt similarly toward both atheists and gays, but felt more comfortable openly voicing distrust of atheists than of gays.  In addition, our sample consisted of American adults, overall a quite religious group.  To address these concerns, we performed additional studies in a population with considerable variability in religious involvement, but overall far less religious on the whole than most Americans.  We studied the attitudes of university students in Vancouver, Canada.  To circumvent any possible artifacts that result from overtly asking people about their prejudices, we designed studies that included more covert ways of measuring distrust.

When I see an explanation like that, it suggests that the researchers first conducted their study using the same methodology for both populations, obtained data that did not agree with their hypothesis, then collected more data for only one group in order to build a consistent, publishable story (if you’re interested, you can see their final paper here).

Because researchers can (and do!) collect data until they see what they want — until they have results that agree with a pet hypothesis, perhaps one they’ve built their career around — it’s not hard to obtain publishable data that appear to support any claim.  Doesn’t matter whether the claim is true or not.  And that, in essence, is why the practices that masquerade as the scientific method in the hands of modern researchers are not convincing persuasive tools.

I think it’s unfair to denounce people for not believing scientific results about climate change, for instance.  Because modern scientific results simply are not believable.

scientists_montageWhich is a shame.  The scientific method, used correctly, is the best way to understand the world.  And many scientists are very bright, ethical people.  And we should act upon certain research findings.

For instance, even if the reality underlying most climate change studies is a little less dire than some papers would lead you to believe, our world will be better off — more ecological diversity, less asthma, less terrorism, and, yes, less climate destabilization — if we pretend the results are real.

So it’s tragic, in my opinion, that a toxic publishing culture has undermined the authority of academic scientists.

And that’s one downside to Norenzayan’s book.  He supports his argument with a lot of data that I’m disinclined to believe.

The other problem is that he barely addresses historical information that doesn’t agree with his hypothesis.  For instance, several cultures developed long-range trust-based commerce without believing in omnipresent, watchful, morality-enforcing gods, including ancient Kanesh, China, the pre-Christian Greco-Roman empires, some regions of Polynesia.

CaptureThere’s also historical data demonstrating that trust is separable from religion (and not just in contemporary secular societies, where Norenzayan would argue that a god-like role is played by the police… didn’t sound so scary the way he wrote it).  The most heart-wrenching example of this, in my opinion, is presented in Nunn & Wantchekon’s paper, “The Slave Trade and the Origins of Mistrust in Africa.” They suggest a casual relationship between kidnapping & treachery during the transatlantic slave trade and contemporary mistrust in the plundered regions.  Which would mean that slavery in the United States created a drag on many African nations’ economies that persists to this day.

That legacy of mistrust persists despite the once-plundered nations (untrusting, with high economic transaction costs to show for it) & their neighbors (trusting, with greater prosperity) having similar proportions of believers in the Abrahamic faiths.

Is it so wrong to wish Norenzayan had addressed some of these issues?  I’ll admit that complexity might’ve sullied his clever logic.  But, all apologies to Keats, sometimes it’s necessary to introduce some inelegance in the pursuit of truth.

Still, the book was pleasurable to read.  Definitely gave me a lot to think about, and the writing is far more lucid and accessible than I’d expected.  Check out this passage on the evolutionary flux — replete with dead ends — that the world’s religions have gone through:

CaptureThis cultural winnowing of religions over time is evident throughout history and is occurring every day.  It is easy to miss this dynamic process, because the enduring religious movements are all that we often see in the present.  However, this would be an error.  It is called survivor bias.  When groups, entities, or persons undergo a process of competition and selective retention, we see abundant cases of those that “survived” the competition process; the cases that did not survive and flourish are buried in the dark recesses of the past, and are overlooked.  To understand how religions propagate, we of course want to put the successful religions under the microscope, but we do not want to forget the unsuccessful ones that did not make it — the reasons for their failures can be equally instructive.

This idea, that the histories we know preserve only a lucky few voices & occurrences, is also beautifully alluded to in Jurgen Osterhammel’s The Transformation of the World (trans. Patrick Camiller).  The first clause here just slays me:

The teeth of time gnaw selectively: the industrial architecture of the nineteenth century has worn away more quickly than many monuments from the Middle Ages.  Scarcely anywhere is it still possible to gain a sensory impression of what the Industrial “Revolution” meant–of the sudden appearance of a huge factory in a narrow valley, or of tall smokestacks in a world where nothing had risen higher than the church tower.

Indeed, Norenzayan is currently looking for a way to numerically analyze oft-overlooked facets of history.  So, who knows?  Perhaps, given more data, and a more thorough consideration of data that don’t slot nicely into his favored hypothesis, he could convince me yet.