The interplay of genetic and cultural evolution

In my last post, I discussed the framework for cultural evolution laid out by Claire El Mouden and colleagues in a new article in the Journal of Evolutionary Biology (ungated pdf and supporting information). By setting out clear definitions for the analysis of cultural evolution, such as cultural relatedness and fitness, a workable framework using evolutionary biology’s Price equation can be developed.

As I noted in that post, it is when the biological and genetic frameworks are laid on top of each other, as is the focus of the dual-inheritance literature, that things get interesting. With their framework, El Mouden and friends tackle a couple of the prominent gene-culture evolution questions.

The first question is to what extent cultural evolution increases or decreases genetic fitness. The authors note that a theme of the gene-culture evolution literature is that cultural evolution made the scale and distribution of today’s human population possible. How does this work?

For there to be a positive correlation between cultural and genetic fitness, those who have the most cultural influence must also leave the most offspring. It is easy to see circumstances where this holds, with those of high social status tending to have both cultural influence and more offspring (look at the number of partners of rock stars).

An example to the contrary is low fertility in developed countries. El Mouden and colleagues reference work by Richerson and Boyd, who suggest that high status is given to professions with high investment in education, which would allow the behaviour to spread despite the negative relationship between education and fertility.

However, El Mouden and friends show that this mismatch between cultural and genetic fitness is evolutionarily unstable as genetic natural selection acts to align genetic and cultural fitness. They suggest two reasons for this, one involving transmission and the other selection.

Their transmission explanation relies on the already evolved propensity for people to avoid behaviours that are harmful to their genetic interests. For example, the nausea produced when consuming toxins would fight against any cultural pressure to eat toxic food. However, it is an interesting question as to how general this transmission effect would be, as many cultural forces are novel and without precedent in our evolutionary history. Transmission may be unlikely to constrain our desire for high status professions that require large investments in education any time soon.

The genetic selection explanation would seem to have broader power. In the supplementary information, the authors ask us to consider a population where cultural and genetic fitness were not aligned. Now, imagine a mutant in that population that causes people to pay attention to a cultural trait that is more highly correlated with genetic fitness. As these mutants have higher genetic fitness, they increase in proportion of the population, and cultural and genetic fitness are now more correlated. Cultural fitness now promotes genetic fitness. In the long-run, the two will be perfectly correlated (the exception being where cultural traits are neutral to genetic fitness).

The catch in that last sentence is the “long-run”. As cultural evolution can be so much faster than genetic evolution, systems can be far from genetic equilibrium until the genetic response evolves. Fertility in developed countries would be an example of this. There may also be some constraints that prevent perfect alignment, such as the presence of appropriate learning mechanisms.

This interaction of genetic and cultural evolution gets most interesting is when we turn to the evolution of altruism. In examining this question we must remember that genetic and cultural fitness are distinct. Cultural altruism reduces the altruist’s cultural fitness; that is, their influence. As a result, the claim that cultural evolution increases genetic altruism (the more common claim in the gene-culture evolution literature) needs to be made carefully.

As an illustration, consider this interesting example from the paper. A stranger is being attacked, so a good Samaritan steps in to defend them and dies as a result. Whether this is culturally altruistic would depend on whether the Samaritan’s deed was copied. If so, then the Samaritan’s act would actually have been increase cultural fitness as it would have increased their influence in respect of that cultural trait.

Conversely, their death is genetically altruistic. As a result, genetic selection would tend to act against it. Those who ignore this cultural trait will have higher genetic fitness, grow in proportion of the population and eventually bring cultural and genetic fitness into alignment.

So what of behaviours that are both culturally and genetically altruistic? Whether the behaviour spreads will depend on the degree of cultural and genetic relatedness.

Evidence suggests that cultural relatedness within ethnic groups is higher than genetic relatedness (although it is still not high in absolute terms, with more within group than between group variation). This means that there are a wider range of circumstances for which cultural altruism can emerge than for genetic altruism. However, that domain in which cultural but not genetic altruism is likely to emerge will be subject to the forces described above to align cultural and genetic fitness.

Another important point is that each cultural trait should be considered separately. Even though a group may have the same language, giving them high relatedness for this cultural trait, this does not mean that they have the same views on giving their lives for strangers, for which they may have low cultural relatedness. Consideration of the conditions for altruism need to consider the specific cultural trait.

There are many other interesting points in the article – I recommend reading the whole thing – but I will close with a point on the practicality of modelling cultural evolution in this way. El Mouden and friends note that there is a host of complications not present in the genetic case. Cultural relatedness can vary wildly across cultural traits, whereas the nature of genetic transmission means that relatedness is similar across most of the genome. Recognising the pattern of inheritance is also a challenge, as ancestor numbers can vary in number and be of vastly different biological ages. In that context, there is no such thing as a standard length of generation.

So although this paper presents a nice approach to cultural evolution, it does not present an approach that is easily applied to empirical observation. However, given the lack of clarity across much of the gene-culture evolution literature, particularly when examined across authors and papers, it is nice to see an attempt to achieve some conceptual coherence.


  1. says

    Jason, nice to see you review that paper. What did you think of the ‘genetic leash’ argument? This is a belief that keeps occurring in evolutionary circles, namely the idea that genetic selection must somehow keep cultural plasticity on a leash, ultimately. I do not see it that way. The “always far out of equilibrium” effect of culture on genetic selection means that genetic selection often gets little traction. Another way of saying it, using Laland’s terminology, is that since both cultural and genetic behavioral endowments modify the environment, but culture does it so much faster, genetic evolution is constrained to respond to selection in a culturally dominated environment. Obviously, dual inheritance theory holds, but the idea that there is some sort of ultimate constraint on culture seems a mistake. Thoughts?

    • says

      I’m relatively sympathetic to the idea of forces of alignment. There are many genetic fingerprints of this alignment, from lactose tolerance in populations exposed to dairying to disease resistance for those exposed to dense living. New cultural traits emerge, but alignment acts on those of the past. I have argued in other work that this alignment is happening today. If there is a large negative effect on genetic fitness, the evolutionary response can also be fast.

      One important point from the paper is that we should think of each cultural trait individually. When we do that, we can see the “leash “or forces of alignment. It is new cultural traits that are out of equilibrium, leading to the high level of disequilibrium we might observe at any point in time.

      I also expect that a lot of cultural traits (most?) are neutral with respect to fitness, so there is no leash but also no negative effect on genetic fitness.

  2. says

    I am also interested to hear your thoughts on Tim’s comment. The standard dual inheritance story is that (1) culture evolves faster than genes and (2) conformity plus culutural norms of punishment/reciprocity (Bowles would likely add reproductive leveling) create and stabilize institutions that influence the direction of genetic selection. It is not clear, at least to me, that genetic selection necessarily holds the leash in cases where cultural institutions create incentives for good behavior (e.g., capital punishment/ostracism/jail time).

    • says

      Henrich’s JEBO article is a nice illustration as to why El Mouden et al’s paper is so useful. Terms such as relatedness, generation or cultural fitness are not defined. Henrich simply gives the cultural Price equation in a between-within group form and that is it. Without those definitions, the subsequent discussion is not as crisp as it could be and it would be hard to actually use the equation as it is given.

      Henrich’s 2004 paper “Demography and Cultural Evolution: How Adaptive Cultural Processes can Produce Maladaptive Losses: The Tasmanian Case” in American Antiquity spends a bit more time on the Price equation for cultural evolution. He does not need to address the relatedness/generational/influence issue as he assumes that everyone in the next generation simply learns from the most skilled person in the last. If you wanted to generalise his work (or examine this argument empirically), you would need to define these terms. (As an aside, I’m not sure he uses the Price equation correctly in this paper, putting mutation in the selection rather than transmission part of the Price equation – and calling selection “selective transmission” makes it more confusing).

  3. Tomás says

    “…although it is still not high in absolute terms, with more within group than between group variation…”

    A blot on an otherwise wonderful article. What’s this, exactly, the Cultural Lewontin’s Fallacy?

    Let’s repeat the arguments stated in the thirties, the sixties, the seventies, and for the whole past decade until today (this time translated to cultural terms): cultural traits also (statistically and strongly) correlate between them, just like genes. In fact, even the laughable objection put together to defend Lewontin’s Fallacy doesn’t hold wáter in this case, because even using just a bunch of cultural traits you can predict, with a high likelihood of success, other cultural traits. If someone supports marriage privileges for homosexuals and is vegan, you can safely assume he defines himself as a “progressive” in politics.

    In order to defend the classic Fallacy, they went so far as to say that using “hundreds” of loci it was still “unsafe” to classify humans into big continental-scale populations (y’know, ehem…races… cough, cough). Hundreds of loci! Will we have enough computational power to manage such huge, gargantuan numbers?

    But this Cultural Lewontin’s Fallacy is even worse, given that with a dozen of highly predictive cultural traits you can predict several hundreds more. Gay and student of “afrocentric studies”? I bet he doesn’t go hunting very often.

    Anyway, without that blot, the post is really good.

    • says

      You are confusing the factual basis for Lewontin’s fallacy with the Fallacy itself. The Fallacy is not that there is more within group than between group variation – that statement is true (and important for the analysis of the level of selection) – the Fallacy is to say that this relative difference in between/within group variation means that there is no justification for between group classifications.

      • Tomás says

        Not really. The Fallacy purposely forgets about the correlation of the data with the only known goal of reducing between group variation. If you’re studying the variation between humans in order to “demonstrate” that between group classifications* are invalid, that is, if you’re comparing sets of data, you cannot forget, no way, about the inner correlation of the data within the sets. Specially when some people has previously told you about the point.

        ( (*): well, some of them, because there is a host of other between-group classifications that were fine and OK for Lewontin, but I’m digressing… sorry.)

        I mean: at the end of the day, we can turn it into a byzantine discussion as much as we want to. It remembers me of that people -I guess pretty much everyone has spotted this game before- that believes that by re-defining terms, or playing with definitions, “A” and “not A” can be true at the same time.

        For example: “Yes, technically you’re getting rid of that group of cells you have labeled as a ‘human developing organism’, but I absolutely refuse to define those cells as a ‘human person’ “.

        Or something like that.

        In fact, by the way, that was what some people said at that time about Lewontin and Edwards: “both of them are right”. Just figure.

        Moreover, “variation” could be defined in such a way as to make Lewontin’s statement false from the start. The problem here is that, for some reason that is unknown to me, we have to implicitly assume a hidden definition of “variation” that seems engineered to save Lewontin’s statement. I think the point is clear: “what’s variation for you?”

        For me, to take the statistical correlations out of the very concept of variation is disingenous, to say the least. But given that every definition is posible, I guess everybody is right (and wrong).

        And in the cultural realm, I really believe you cannot take out the correlations from the definition of variation, either. Even less than in the genetic área, because it’s my intuition (I acknowledge it’s a intuition, I could be wrong, of course) that there is even more correlation. A correlation that is highly predictive.

        For example: “your way of dividing society between ‘rightists’ and ‘leftists’ has no value, no predictive power, it’s not justified, because there’s more intragroup variation that intergroup variation. In fact, it is ‘politicist’ …”

        I still think I’m correct on this. “Cultural Lewontin’s Fallacy” is a good name.

      • says

        The correlation within groups is not forgotten in El Mouden et al’s paper – it is explicitly captured in the measure of relatedness if using the inclusive fitness approach and the measure of within group variation using the multi-level approach.

        Lewontin used a standard definition of variation from population genetics. The issue with Lewontin’s approach was not his measure of variation, but that he then used this measure to answer a question it wasn’t suited to.

  4. Heinrich Magen says

    “Cultural altruism reduces the altruist’s cultural fitness; that is, their influence.”

    I am not sure I follow this point. Can this be elaborated on?

    • says

      It’s worth going back to my first post on this topic as it lays out the core concepts. In short, the measure of cultural fitness is influence. To be altruistic, you sacrifice your influence in order to increase someone elses’. That, by definition, reduces your cultural fitness.

Comments, thoughts, suggestions?