Rational Culture

For millennia, thinkers have spent a lot of effort figuring out how an ideal “rational” agent would reason and act. While we know that we humans were not built to completely fulfill this idea, we do in many ways approximate it. It looks like natural selection has been trying to make us more rational.

Most people are “rationalist” in the sense of trying to some degree to more closely approximate this ideal, at least in many circumstances. But most also put limited effort into this, and are wary of becoming too rational in key contexts. Those who explicitly embrace the label “rationalist”, in contrast, claim to try harder and go further in such efforts. 

I’ve been talking a lot lately about humanity’s serious problem with “cultural drift”. Culture has long been our superpower, but in the last few centuries we have greatly cut cultural variety and selection pressures, and greatly increased internal drift rates. As a result, behavioral features that are hard to vary within cultures have likely been drifting at an accelerating rate into maladaption for a few centuries. (Features than can vary within culture, in contrast, are doing fine.) 

How would a rationalist fix this? They would of course try to see the usual efforts in the area of culture as a crude approximation to the actions an ideal rational agent, and then try to move their actions to become a closer approximation to that ideal.

Typically, humans accept the norms and values of their culture, and then change those as their culture changes. Cultures change due to natural selection, and due to fights between cultural activists. And I think the most natural way to interpret human cultural evolution as the actions of a rational agent is this: think in terms of an agent whose constant-in-time preference is to create more adaptive culture-enabled descendants, and who tries out different cultural variations to gain info on which cultural features are actually likely to be more adaptive in near future contexts. After all, if you squint your eyes, natural selection can be see roughly as an agent trying to create adaptive descendants.

To more closely approximate this ideal, a cultural rationalist would not see themselves primarily as an agent with core values equal to those that their culture taught them in the usual way. They would instead see such given values as tentative trial priorities being tested as candidate versions that might lead to more adaptive behavior. To the extent that they can see a pattern of such trial priorities becoming less adaptive over time due to maladaptive cultural drift, they would lower their estimates of more recent versions actually being adaptive. 

A rational cultural agent would instead look at their entire historical dataset of known cultures, their values, and their adaptive successes, combine that with their best theories of natural selection and the world, to estimate the culture priorities most likely to be adaptive in relevant contexts. (Such contexts might include their DNA, if that couldn’t be changed.) If such an agent had to act alone, it would do this analysis as best it could on its own, and then use a weighted average of such best guess action priorities as its guide to action. 

If such rational culture agents could coordinate with others, however, they might do better to conduct a much more detailed analysis together, and then simultaneously explore many possible cultural values, in order to more quickly discover which were more adaptive. And they might try to coordinate to change their shared contexts, to enable them to be more adaptive. 

The key point, however, is that such a rational culture agent would not especially identify with or embrace the values of the culture in which it grew up, or now lives. Yes, there would be many practical advantages to behaving in ways that inspired confidence in, and did not offend, associates. But that is quite different from fundamentally embracing and promoting such values. 

Cultural change driven by rational culture agents might plausibly improve at far faster rates that do cultures driven instead by dumb natural selection. All else equal, cultures driven this way would likely outcompete others. Natural selection should select for rational culture agents!

Alas, in my experience so far, few who call themselves “rationalist” seem willing to go this far. They mostly want to accept the values of their culture, and then try to be more rational in how to achieve such values, regardless of how adaptive they are. Unless this changes soon, and among enough elites in some society to influence the direction of that society, cultural drift seems unlikely to be fixed soon by rationality. 

Added 5Sep: Our book The Elephant in the Brain, shows just how hard it is to be rational; we are quite reluctant to admit to our real motives. Would be even harder to change such motives.