Is Social Trust a Cultural Trait?



Join 36.9K other subscribers

Yesterday’s blog proposed that the most useful approach to understanding the evolution of large-scale complex societies is to view them through the lens of Cultural Evolution. To make this discussion more concrete, let’s look at a particular cultural trait: social trust (more specifically, ‘generalized trust’).

Trust is highly important for explaining the ability of people, teams, and whole societies to cooperate. Social trust creates mutual bonds between citizens that make them willing to enter into potentially profitable but risky transactions and to participate in collective enterprises that create public goods (more on this in Francis Fukuyama’s Trust). Social scientists, beginning with Alexis de Tocqueville, have known that generalized trust is a critical ingredient for collective action, economic growth, and effective governance (in addition to the already mentioned Fukuyama book, a useful compendium is by Diego Gambetta, Trust: Making and Breaking Cooperative Relations).



Thanks to such organizations as the General Social Survey in the USA and the European Social Survey, we have a lot of quantitative data on how social trust varies within societies, between societies, and how it changes over time.

The standard question that these surveys ask is, “Generally speaking, do you believe that most people can be trusted or can’t you be too careful in dealing with people?” Respondents are then offered four versions of the answer:

  • always trusted
  • usually trusted
  • usually not trusted
  • always not trusted

Note that the spectrum of potential responses is, rather arbitrarily, split into four discrete answers, although it is clear that the degree of belief that most people can be trusted grades smoothly from a complete agreement to a complete rejection of this statement. Relevant to the previous blog, generalized trust doesn’t seem to be a meme, at least not as usually understood by the proponents of Memetics.

But what if it’s not even a cultural trait (as I defined them in the previous blog)? The key question is whether this attitude is socially transmitted, or individually learned.

Clearly, our assessment of whether any specific individual known to us should be trusted, or not, will be affected by our previous interactions with him. Theorists of trust, like Eric Uslaner, refer to this as ‘particularized’ or ‘strategic trust.’

But what about trusting a stranger, someone with whom we have no history of interactions? It turns out that people, even within the same population group, vary quite significantly in their attitudes towards how much strangers should be trusted. And specific encounters (with either a trustworthy or a duplicitous person) have a small effect—these attitudes are characterized by a great degree of stability. At least, this what Uslaner argues in The Moral Foundations of Trust. He makes a distinction between particularized or strategic trust and generalized or ‘moralistic’ trust. In his view, generalized trust is really about whether we should trust people (rather than whether people are in general trustworthy). It’s more a moralistic attitude, rather than an assessment of whether people are generally trustworthy, or not.

Periodic social surveys indicate that generalized trust behaves just as we would expect a cultural trait to behave. National-level studies show that each surveyed population is characterized by a mixture of people holding different beliefs about whether others can be trusted. The relative proportions of different beliefs are quite stable in time—they change, but slowly. In other words, this cultural trait evolves (remember that my definition of cultural evolution is based on the frequencies of cultural traits changing with time).

But the critical question is, how is this attitude transmitted? Uslaner also presents data showing that the most important influence predicting a person’s stance on generalized trust is the attitude of their parents. In other words, if he is correct, we typically learn generalized trust (or distrust) from the previous generation, and that makes it a cultural trait.

Somewhere I read a wonderful story illustrating how trust (or distrust) is transmitted. It went something like this. A father tells his son, who is sitting on a tall fence: “Jump down, and I will catch you!” The son is dubious, but the father says, “Don’t you trust me?” Finally the son jumps, at which point the father steps aside and lets him take a painful fall. “Remember, son, you should not trust anybody.”

I thought I read this story in Edward Banfield’s book The Moral Basis of a Backward Society, but I have been unable to locate it in the copy of the book that I have. Can anybody provide me with the source for it??



Notify of
Most Voted
Newest Oldest
Inline Feedbacks
View all comments

Hypothesis: it’s a genetic trait likely influenced by a degree of outbreeding (that is lack of close kin marriages) within a population.
Would explain a lot quite nicely wouldn’t it?

T. Greer


4-generational studies of immigrants to America suggest that trust is too plastic to be genetic. See Francesco Giavazzi, Ivan Petkov, Fabio Schiantarelli, “Culture: Persistence and Evolution,” NBER Working Paper, No 20174 (May 2014).


Peter Turchin

Thanks for the ref! Looks very interesting.


Interesting – not convincing but too long to analyze in a blog comment.
In the meanwhile take a look at this

Interesting article and thanks for writing about trust.

You might want to correct the spelling of Eric Uslaner’s name 🙂

Peter Turchin

Thank you for catching it! Now fixed.


I think you are mistaken about what meme enthusiasts think. I for one think all culture is based on memes – and there are plenty of others who agree with me. The main exception I can think of is Susan Blackmore. She has the idea that only things copied by imitation are memes, and things copied by stimulus enhancement or local enhancement don’t really qualify. Boyd and Richerson (1985, p.35) advocated much the same rather odd notion (though using different terminology) in their first book. These days, this sort of thing looks like human exceptionalism.

These days, the idea that memes only underlie some cultural practices mostly looks like a set-up job by critics and opponents. The claim is found on the “Dual Inheritance Theory” Wikipedia page, for instance. It’s provided supporting reference is the first edition of Laland and Brown’s book. I read that edition of that book – and didn’t see how it supported the claim. I would label the idea as mostly a misunderstanding or misinterpretation.

Most evolutionary models of culture apply equally well to all kinds of information that is copied from one agent to another. There isn’t much point in having a theory of culture that only covers some of these types of information – since you then need a very similar theory to cover all the rest. However, I don’t think there’s a good reason to associate this dud idea with memetics, when only a small number of meme proponents ever advocated such a notion – and when prominent memophobes within academia had been promoting the same dubious idea since 1985.

Peter Turchin

Richerson and Boyd continue to criticize ‘meme’ in their latest work (e.g., Not By Genes Alone).


So what? Their criticisms are only of marginally better quality than yours are. They never bothered to find a sympathetic interpretation of memes either. Like most critics they found an unsympathetic interpretation and criticized that. This is a common approach – but rarely a helpful one. Maybe because they invented the rival terminology, they suffer from not-invented-here syndrome. Maybe they prefer terminology that won’t offend their fellow anthropologists by rubbing biology in their faces. It’s hard to know for sure – since they only rarely discuss the topic these days.

Academics in general seem to have a hard time understanding memetics. Probably most of the meme proponents haven’t been folk that academics are interested in affiliating with. Also, memetics is too revolutionary for many of them; many of them are still on the wrong side of its paradigm shift.

Fortunately, academia does seem to be gradually catching up. Its sluggishness and inertia are frustrating, though.

Peter Turchin

My objection to memes is that they are too ‘gene-like’. The big difference between genetics and culture is that in the first there is a clear separation between the genotype and phenotype. It’s genes that are transmitted, but expressed traits that are selected.

In cultural evolution, it is the phenotype that is transmitted. It is then encoded, possibly in very different ways, in different people’s brains. When I learn to dance czardas I most certainly use a different way of encoding it than my wife. I use different muscles in different ways (admittedly, I am a lousy dancer).

So the big point is that cultural evolution is Lamarckian, unlike genetic. So gene-like memes is not the way to go.


There is no “inheritence of the phenotype”. If its inherited, it is genotype, *by definition*. That is the very essence of the distinction between genotype and phenotype. Genotype is heritable information, phenotype is all the things it affects.

This approach to these concepts works well for both organic and cultural evolution. Of course you *could* define these concepts differently, so that they don’t work properly in both domains. However, that would be contrary to the whole enterprise of finding a general theory of evolution. If your concepts apply to some evolutionary phenomena and not others, maybe your concepts aren’t very general, aren’t very good – and you should reconsider them.

“Lamarckian” is a vague term – rather like “Darwinism”. However, if one takes it to refer (in part) to the inheritance of acquired traits, then both cultural evolution and organic evolution are “Lamarckian”. In cultural evolution, acquired memes are passed on. In organic evolution, many acquired traits are passed on – particularly via symbionts. AIDS is acquired and transmitted, for example. So are fleas and their bites. There have been lots of attempts to paint DNA evolution as non-Lamarckian. However, as far as I can tell, they are all failures on technical grounds. The idea that cultural evolution is Lamarckian and DNA evolution is not is mostly just wrong. The claim is hard to refute – but that is mainly because Lamarck believed a bunch of things, and so the non-Lamarckian claim has become a vague and shifting target.

Carl Coon

Peter, you never did comment on my new ebook, “A Short History of Evolution” but perhaps you’ll now visit the chapter “Deconstructing the Promethean Spark”, esp the section on altruism. The issue of trust you raise can be seen as a facet of altruism; it arises naturally within groups small enough so each individual has direct or indirect experience with the other individuals. This phenomenon can best be considered a basic building block for cultural evolution rather than a product of cultural evolution. We can view cultural evolution as a kind of architectural process that knits together such socially unified entities into larger ones. I don’t go into this in detail but do observe that when you build a multi-storied building you do not start at the top and work down–the philosophy behind many of our do-good interventions abroad.

Peter Turchin

Thanks, Carl. I’ll try to comment on your book in a future post. All: here’s a link to Carl’s book if anyone wants to read it

Gene Anderson

This agrees with what I know about trust. I have heard several variants of the “jump down” story–I too don’t remember the references though. One thing: what people say about trust may have little to do with what they really believe. In many societies, people have to trust kin and at least a few others, but are very distrustful of people-in-general.
One amazing thing is the incredible degree to which trust can prevail. There is a wonderful book by Sebouh Aslanian, FROM THE INDIAN OCEAN TO THE MEDITERRANEAN, about the Persian Armenian business network in the 18th century. People had to trust each other with huge advances of capital, even though letters were scarce and luck shaky, and you might not see or hear from your partners for ten or more years. Trust was almost absolute. A man would trust his partner (usually family–naturally) with a fortune and the partner would duly come back after 10 years out of touch with a substantial profit. I have enough Persian Armenian friends to know this is still sort of the case. There are similar stories of Chinese merchants in southeast Asia at the same time.
Memes and culture traits: There is an absolutely enormous amount of work on this. Nothing remotely like a “meme” exists. Cultural knowledge is not packaged in neat little clumps, does not spread like genes or bugs or viruses from person to person, and does not have a life or identity of its own. People learn things, shaping them as they learn. It’s fine to talk about “culture traits” and “cultural models” because there is a long lit on them and we pretty much know that the “traits” are broad complexes of beliefs and data, integrated in wide and very consciously managed networks of knowledge and constantly refined on the basis of current contexts and situations. The old early-20th-century “culture trait” model of actual, timeless, changeless traits that are isolated and have a specific identity of their own was disproved back in the 1950s when there was a huge debate on it. The “meme” idea is even farther from reality. See any work on cultural reality–Roy D’Andrade is a good writer to start with–

Eric M. Uslaner

I am Eric M. Uslaner (not Uslander). The post is interesting but filled with problems. The author doesn’t mention the difference between generalized and particularized trust (trusting people in general versus trusting only your own in-group, see my work and especially that of Yamagishi and Yamgishi, ) and he doesn’t discuss the connection between inequality and trust, which is even more important than any cultural differences. Also he has the survey question wrong. He is correct that the question is: “Generally speaking, do you believe that most people can be trusted or can’t you be too careful in dealing with people?” But he is wrong when he states:

Respondents are then offered four versions of the answer:
• always trusted
• usually trusted
• usually not trusted
• always not trusted
Note that the spectrum of potential responses is, rather arbitrarily, split into four discrete answers, although it is clear that the degree of belief that most people can be trusted grades smoothly from a complete agreement to a complete rejection of this statement.

This is simply incorrect. There may be a handful of surveys where the answers are posed this way, but in all of the surveys he cites, the question is a simple dichotomy

Peter Turchin

First, again my apologies for misspelling your name – Barbara Kimmel above has already pointed it out, and I fixed it.

Second, I’d like to point out that this was a blog post, rather than an article, and in my blog I aim at engaging the interest of the general audience, rather than at scientific rigor. Hopefully this blog has achieved that goal.

Nevertheless, I do mention the difference between generalized and particularized kinds of trust.That’s the main point of this paragraph:

“Clearly, our assessment of whether any specific individual known to us should be trusted, or not, will be affected by our previous interactions with him. Theorists of trust, like Eric Uslaner, refer to this as ‘particularized’ or ‘strategic trust.’ ”

In the next paragraph I say: “He makes a distinction between particularized or strategic trust and generalized or ‘moralistic’ trust,” referring specifically to your book.

I actually find the connection you make between inequality and trust an extremely interesting idea. However, the good rule of writing blogs is to focus on one idea at a time.

One other thing: the four answers to the question that I list I got from the GSS code book here:

However, I am happy to be corrected that the usual version is just two answers.

Finally, full disclosure: as I mention in my blog on memes, progress, etc, I am now in the middle of writing a book on the evolution of cooperation, especially in large-scale complex societies. The last two blogs had an ulterior motive, to run some of ideas by the readers. So thank you, Dr. Uslaner and all others, for your comments.

Peter Turchin

I just realized that the link I provide above doesn’t lead to the specific variable. Its POEPLE CAN BE TRUSTED OR CANT BE TOO CAREFUL (misspelling is in the original)

T. Greer

Peter, does that mean the ‘cliodynamics of the American Republic” project is no longer moving towards publication as a book?

Peter Turchin

No, it’s on. I am still negotiating with publishers. I tend to purse several projects on parallel tracks. The main ones currently are the book on the struct-dem analysis of the USA, the popular book on the evolution of cooperation, and Seshat (the last one taking the bulk of my time).


Much of trust is a recognition the options for defection of the person being trusted are limited so is a condition generated by the environment. if the Persian Armenian entrusted a fortune in an 18th Century business network were to “take the money and run” where would he run too, and where would he put the money? it would need to be somewhere unknown to his Persian Armenian partner, somewhere he could not be found, and where the wealth would remain safe and accessible to him. that would not have been easy. if there were Swiss bank accounts in the 18th Century they weren’t accessible from Uruguay. ranch in middle of Africa – no chance. trust is related to the sense of belonging to a community and that sense of belonging may rise or fall with the ease one can defect from it – i.e. change communities. this must in turn be related to how similar communities are to each other. members of dissimilar communities – especially when hostile – will not defect to the other. cultural traits which maintain strong barriers between communities – nationalism, racism, fundamentalist religion – exploit lack of options for defection to keep trust high and the community together. the cultural trait might seem to benefit from that trust but this condition of high asibiyah doesn’t last for ever. this does not occur “for the benefit of the meme.”

Gene Anderson

On culture traits…. Consider the bow and arrow, possibly the most classic example of a culture trait that spread all over the world (almost) over 50,000 years as a single package. An anthropologist can easily ABSTRACT it from cultural context, and talk about the “diffusion of the bow and arrow,” but to a Mongol horseman, a San hunter, or an English archer at Agincourt, the bow and arrow is part of a whole lifeway and is associated with a whole world–and those would be three VERY different things, with the bow and arrow meaning VERY different things, and consequently being constructed very differently, used very differently, talked about differently, etc. Thus with all packets of knowledge. They are networked into lifestyles. It is useful to abstract “beliefs,” “traits,” “knowledge systems,” etc., analytically, but one has to remember that to the people that hold and use them, these things are parts of larger wholes. Probably the nearest thing to a mindless, universal, automatically involved cultural rule is a pluralization rule: -s (or -es) in English, etc. Yet even that can get interesting in practice…I could tease my kids by inventing plurals like “sheepen.”

Peter Turchin

Gene, this is a very good point. I try to make a similar one above in my response to Tim.

Eric M. Uslaner

To: Peter Turchin, particularized trust is not the same thing as strategic trust. The former is trust only in people in your own group (however defined) and the latter is trust based upon experience–the sort of trust EdwardT discusses (which is not related to generalized trust at all).. Particularized trust need not be based upon experience–it is more likely to be based upon stereotypes and fear of out-groups. See chapter 2 of The Moral Foundations of Trust.

Second, the GSS did experiment with different question wording, but the standard question is a dichotomy. I worked with the American National Election Study on a revised version of the question that has been asked by the National Longitudinal Study of Youth–and convinced them that the standard dichotomy was at least as good.See my essay, which is Chapter 7 in Aldrich and McGraw, eds., Improving Public Opinion Surveys (

Finally, on the connection between trust and inequality, see Chapters 2, 6, and 8 of The Moral Foundations of Trust and the paper I wrote for the Center for American Progress:

I;d be interested to learn about your book in progress. Feel free to contact me directly.

Peter Turchin

Once again, I am happy to be corrected (on the difference between particularized trust and strategic trust). You are quite right. The only thing I would add to the distinction between particularized trust and generalized trust is that a simple dichotomy is probably not sufficient. Humans are typically members of a whole set of nested groups: nuclear family, extended family and friends, village or small town or neighborhood, subethnie (same dialect), ethnie or nation, supranational community, and the humanity as a whole. This means that we need finer gradations for “strangers.” In fact the Greeks had two different words to express ‘otherness’: xenoi (foreign Hellenes) and barbaroi (non-Hellenes).

Thanks for your offer which I will be taking you up on. I am also interested in your other work (The Decline of Comity in Congress), about which I only recently learned and did not have a chance to read yet (I wonder what you say about that _now_). I’ll be in touch by e-mail once I have a text ready for comments…

Peter Turchin

Also, I have collected my posts on inequality and cooperation here:

under the heading Elite Overproduction, Inequality, and the Unraveling of Cooperation (Structural-Demographic Dynamics)


‘Somewhere I read a wonderful story illustrating how trust (or distrust) is transmitted’

It looks to be an internet meme

Peter Turchin

Now here I am willing to admit it’s a ‘meme’!


That seems unscientific, though. In science, part of the point of the enterprise is to find general explanations that apply to lots of things.

If you use terminology with narrow and specific definitions, you wind up with narrow and specific explanations that simply aren’t very useful. That is all very well if your aim is to ridicule the terminology and try to kill it off. However, that isn’t going to go down very well with other scientists who want to use that terminology to do useful work.

Why would anyone want to kill off the meme, though? It has done more to bring public understanding of cultural evolution than anything else has ever done. That benefits all humanity – except maybe those few scientists who want to monopolize cultural evolution and turn it into an esoteric discipline that only they and their students understand.

Peter and Eric (Ric)- glad I was able to connect the two of you, as I sent your blog post to Eric yesterday morning.

Peter- hope you have had a chance to look over the extensive trust resources on our website at

Cheers, Barbara

Peter Turchin

Thanks, Barbara. Interesting project. Perhaps you might want to explore connections with David’s Prosocial initiative:

Eric M. Uslaner

It is always a good idea to read on what you are writing about. The Giavazzi paper is an excellent example of this. Without any grounding on trust. they try to make an original contribution. I don’t understand the difference between third and fourth generation respondents. Their measure of trust is flawed. As I have shown in several papers, fairness and helpfulness are not trust. Their time trends are distinctly different and often are negative. And often the three measures don’t form an acceptable scale. And the generalized trust question is NOT a measure of trustworthiness. They also then take the standard deviation of each generation of European immigrants as a measure of attitudes. Why European immigrants? This mess is precisely what you get when all you have done is read one or two papers by economists on trust. There is a lot of literature on this topic (including a paper of mine, “Where You Stand Depends Upon Where Your Grandparents Sat,” Public Opinion Quarterly, 2008) and lots of work in Europe by political scientists and sociologists. I never write about a topic without reading the relevant literature. Didn’t stop these guys. Nor did their terrible prose.

And Peter–when you write about inequality, you need to look at the works of Piketty and Saez, Katz and Goldin, Frank Levy, Tim Smeeding, Joseph Stiglitz,, Jacob Hacker, and Wilkiknson and Pickett, The Spirit Level (among many others). On the Nordic model, see the paper I wrote with Bo Rothstein in World Politics in 2005 (“All for All”) as well as other works by Rothstein, Larsen ( is a good place to start), and Kumlin (with Rothstein, Making and Breaking Social Capital) and Freeman (


And yet, interestingly enough, generalized trust appears to be – unlike virtually everything else – not very heritable at all:

More Behavioral Genetic Facts | JayMan’s Blog

I want to look at the ESS (as well as the World Values Survey) to get an idea of how generalized trust varies across societies.

Eric M. Uslaner

I would not recommend using the post-1995 WVS. It is filled with errors–notably trust is higher in Iran than in Britain and trust in Canada drops from 54% to 39% in 2000 and in every GSS survey since–despite the fact that I have used three different Canadian studies (independent of each other) that showed the same 54% trust level in 2000 that the WVS showed in 1995. The WVS has other problems (trust in government is 99% in China, Vietnam, and Azerbaijan–and Bangladesh leads the world in the % of its population doling upaid volunteer work for environmental organizations–35%).

The Naef and Schupp paper would benefit from some background in how trust is measured and what it means–and what the basis is of different forms of trust are. And the Van Lange paper has a total of 156 parents, very small sample size. I don’t think that trust is transmitted genetically, but there is a strong socialization effect. See my Moral Foundations of Trust, ch. 6. On the stability of the generalized trust measure, see Chapter 3 and I have three papers on that topic (if you want them, emal me). Parental trust is a very strong predictor of their children’s trust, even decades later.



Well, the low heritability strongly suggests that the measurements used so far are seriously flawed, and don’t actually measure anything.

As far as I know, there is little investigation into the predictive validity of trust measures, this is even taking into account that it is context-specific.

Eric M. Uslaner

Well I told you that I have written a book chapter and three published chapters on this. You don’t have to believe me if you don’t want to. It’s a free country. Believe what you want. The alternative is that the heritability studies are flawed. Let’s call this off. I really don’t think that this conversation is going anywhere, since you establish yourself as knowing more about this than I do. I won’t reply to you any more.


Eric, I think you’re misunderstanding me, because I am (broadly) agreeing with you (save the heritability part). I am not trying to claim superior knowledge on the issue of trust, because I haven’t researched it much and it’s clear you’ve looked at it much more than I have.

That said, I can speak with authority on behavioral genetic studies. I will tell you that virtually every single human behavioral trait observed has shown substantial heritability, and this has been confirmed through multiple very different methods. Those also show broadly that “vertical cultural transmission” (cultural transmission from parent to child) doesn’t exist; parent-child similarity (relative to the culture at large) is entirely genetic in nature.

So, that the few behavioral genetic studies that have looked at trust (which have parted the various forms of trust) show that it has no significant heritability is highly anomalous. Such that it strongly suggests (as was the point of linked post of mine) there is something wrong with the trust measure, at least as used in the study (self-report). Likely some experimental measure of trust is called for, or, at the very least, other-report (peer report).

Eric M. Uslaner

JayMan,you are not agreeing with me on anything. And you are wrong on all of these points. But since I have already made my points I won’t repeat them. So it is best to leave this alone and leave me alone. There is nothing wrong with the trust measures, only with the genetic studies. And here is the problem with your argument: “every single human behavioral trait observed has shown substantial heritability>” This is pure <> since the llnkage of many of these behaviors have no connection to genetics. Young whites are less trusting, more liberal, and more likely to vote Democratic than their age group peers over time. So unless you can say that genes have changed, you have a lot of explaining to do. This is my last reply. I don’t argue with people who don’t know what they are talking about .

SEF Editor

I would like to remind all leaving comments here to stay within the bounds of civilized discourse. I am letting this comment stand, although I have removed the offensive language from it. We all may disagree on various issues and different people commenting on this site have various levels of expertise. Still, this doesn’t mean that insulting each other will be allowed.

Eric M. Uslaner

You may unsubacribe me. I don’t appreciate being dissed by someone who admits he has not read my work.

  1. Home
  2. /
  3. Cliodynamica
  4. /
  5. Regular Posts
  6. /
  7. Is Social Trust a...

© Peter Turchin 2023 All rights reserved

Privacy Policy