Doing cultural evolution right

A sojourn into the literature on cultural evolution can be confusing. Authors use the same terms in different ways. Unique models are used to reach opposite conclusions. And each author seems to find their own way to intertwine genetic evolution into the analysis.

In that light, a new article in the Journal of Evolutionary Biology (ungated pdf and supporting information) by Claire El Mouden and friends seeks to nail down some of the concepts of cultural evolution and to set up a general framework (thank you!). The paper is at a more basic level than that of Geoffrey Hodgson and Thorbjørn Knudsen’s book Darwin’s Conjecture, which also sought to define and generalise concepts in this area.

El Mouden and her colleagues’ paper covers a lot of interesting terrain, so I will cover it in two posts. In the first, I’ll cover the basics of a cultural evolution framework. In the second, I will look at how cultural and genetic evolution interact in this framework.

The authors set up their framework using the Price equation from evolutionary biology. The Price equation divides evolutionary change of a trait into two components. The first is a natural selection component resulting from the covariance between a trait and relative fitness. Where there is large covariance, evolution will be fast. Second is a transmission component, which is the fitness-weighted change in trait value between generations (for example, increasing height with improved nutrition across the population would be considered transmission). The Price equation has the neat property that it can be decomposed into within-group and between-group components, allowing analysis in a multilevel selection framework (although not everyone is happy with this decomposition).

But to use this framework, it is important to clarify some terms (which is one of my bugbears about the cultural evolution literature). First, relatedness. As the units of inheritance are cultural traits, the measure of relatedness is similarity in cultural traits. In a simple model where we have one cultural trait, anyone with that same cultural trait has a relatedness of one. In effect, when passing on a cultural trait to another person, they become kin.

The use of the term relatedness is often confusing in the cultural evolution literature as the relatedness of interest is typically genetic relatedness. That is fine, but we need to distinguish the two types of relatedness. Cultural kin are not necessarily genetic kin.

Second, fitness. Cultural fitness reflects the number of people who learn from an individual, plus the degree of influence that they have on those people. Degree of influence is important because, unlike genetic evolution where you have a known and fixed number of ancestors (one parent in the case of asexually reproducing species, two parents for sexually reproducing species such as humans) who contribute a specific amount of genetic material, the number of cultural ancestors may vary by trait and between people. How many people have influenced your cooking? Who was more influential?

Further, for each cultural trait, people will have different fitness. The authors offer the example of Beethoven, whose influence in cookery did not match his influence in music. This necessitates different measurements of cultural fitness for different traits.

Third, generation. The ancestor-descendent relationship is defined by influence, and can have weak relation to biological age. Plato is still spawning direct cultural descendants today, whereas ideas can also spread through a population in days. However, it is only possible to influence people in the next cultural generation, as that is how generation is defined. If I influence someone, they are the next cultural generation in respect of that cultural trait.

Having defined these concepts, they are relatively easy to slot into a cultural Price equation (the maths is in the supplementary information to the paper). While there is extra complexity from considering the degree of influence rather than just the number of descendants, the form of the Price equation is effectively the same for both the genetic and cultural forms. It is just that each deals distinctly with genetic or cultural fitness.

It is also possible to derive a Cultural Hamilton’s Rule. In biology, Hamilton’s rule states that a gene will spread if the cost of the act to the altruist is less than the benefit accrued by the beneficiaries adjusted by the degree of relatedness. A gene can spread if you help kin who also have that gene, even if it comes to a cost to yourself.

Similarly, the Cultural Hamilton’s Rule states that “a behaviour that reduces the actor’s lifetime cultural influence can only be culturally selected for if the cost to him is less than the product of the cultural benefit to his interaction partners and their cultural relatedness to him”. On this point, the authors give an example of two philosophers with the same cultural views. If one chooses to farm to feed the other, allowing the other to focus on spreading the philosophy, the cultural trait may spread despite one of the philosophers effectively sacrificing his own influence.

Under this definition, cultural kin selection becomes a relatively parsimonious explanation for the spread of many cultural traits, such as altruism (and as noted above, this could also be converted into a multilevel selection framework). If people believe in altruism and help others who also do (who are their kin), then helping each other could assist in the further spread of the cultural trait of altruism.

However, this story of spreading cultural altruism falls somewhat short of covering the examples in much of the gene-culture evolution literature. The issue is that, while culture is a part of the model and analysis, people are typically interested in genetic altruism.

Thus, the question of interest is how cultural evolution affects the evolution of genetic altruism? That will be the subject of my next post.

5 thoughts on “Doing cultural evolution right

  1. Super interesting Jason, thank you

    Hamilton’s [Cultural] Rule neatly explains, but can it predict? Think we cultural evolution bods need to demonstrate predictive power in the real world to thrive.

    For example, Hamilton’s Rule explains why Edward Snowden did what he did (, but why Snowden, why now?

    Likewise, Hamilton’s Rule can explain why academics invest in their students (it’s easier to boost overall inclusive cultural fitness by significantly boosting the low fitness of students rather than marginal increases to personal cultural fitness). But can it predict which academics/which students will give/receive help and under which circumstances?

    I tried to apply this to explain/predict at risk individuals for suicide terrorism – part of my PhD. It’s the predictive bit that we need to crack to avoid becoming another ‘just so’ narrative…

    1. Thanks for your thoughts Paul.

      To be able to predict, you need to understand why a cultural trait is associated with high fitness. Why are some ideas more influential? Hamilton’s Rule provides a basis for determining the degree of influence required, but does not address the question of why that particular trait has influence. As a result, we shouldn’t look to Hamilton’s Rule for predicting anything.

      My favourite discussion of this point is by Christopher Jensen – written in the context of the inclusive fitness debate triggered by Nowak, Tarnita and Wilson’s paper. The challenge, as Jensen and you suggest, is to turn inclusive theory into a predictive science. But that is a large step beyond Hamilton’s Rule (and a purpose beyond which Hamilton initially devised the rule).

      On your suicide terrorism point, there could be a large disconnect between genetic and cultural fitness there (unless somehow brothers/sisters of the terrorist somehow have more children – maybe through status and additional resources). I’ll cover a bit of that territory in my next post.

    2. Explanation and prediction are two sides of the same coin. Explanation requires a model. Prediction involves running the model forwards in time. If an explanation can’t be used to make predictions, it probably isn’t really explaining very much either. So: of course Hamilton’s rule makes predictions. If it didn’t, it wouldn’t be proper science.

  2. Hi. Sorry for posting this up so late after the event, but I’ve only just seen this.

    You might also be interested in a recent paper I co-authored with Nico Claidiere and Dan Sperber, in which we describe a way of doing cultural evolution that, while still very much evolutionary, steps away from some (but not all) of the assumptions of the classic biological models. You can read the El Mouden et al paper and our paper as something of a pair. The El Mouden paper shows the formal limitations of sticking too close to the biological model; our paper discusses which assumptions must be dropped to make things work. Here’s the paper:

    1. Thanks for the pointer Thomas. Interesting paper.

      One side thought – when I looked at the ECMs I kept thinking of them as generalised Lotka-Voltera equations or frequency dependent selection equations, but being used/framed in a different way.

Comments welcome

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

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

Google photo

You are commenting using your Google 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