Verified by 
Artificial Intelligence
The New Science of Psychology and Evolution
Psychology and evolution have a difficult relationship, but AI may help mend it.
Posted December 13, 2023 Reviewed by Monica Vilhauer
Key points
- Our modern perceptions of social groups can lead us astray from logically valid scientific inference.
- The complex relationship between human behavior, evolution, and genetics is not yet well-understood.
- AI approaches like computational evolution will allow us to simulate evolution and its effect on behavior.
- Computational evolution allows us to test hypotheses about how and why we evolved different capacities.
From popular culture to the dark corners of the internet, there are many claims about “fundamental” differences between people on the basis of gender, racial or ethnic groups, sexuality, or other social distinctions we make. Sexism and racism are often based on the tacit assumption that one group is better or more well-suited to certain roles than another based on their genes. One of psychology’s most important roles is to inform and (where appropriate) correct these claims — yet fields like evolutionary psychology have a reputation for perpetuating them.
One of the reasons for this is that the relationship between behaviors we observe in humans — including real or stereotyped differences between groups of people — and the evolutionary process is a complex one. It’s extremely difficult to study the evolution of any species, but it is especially difficult with humans because our evolution cannot be directly observed. Our intuitions and observations about the world as it is, and the social perceptions we have formed, can contaminate our understanding of why people are the way they are and even contaminate the supposedly scientific study of things like gender and race.
A key point to remember when delving into the study of genes is that evolution does not have intents or destinations — and just because we observe something (e.g., a difference between men and women) does not mean that it is the result of evolution. Some of the most dangerous applications of evolution are ones that confound cultural and social factors with biological ones. For example, the observation that race or skin color is related to income in the United States has nothing to do with biological factors and everything to do with historical and social factors. Likewise, the division of labor between women and men inside the home and outside the home can be almost entirely explained by cultural factors (e.g., the advent of agriculture).
The study of selection, inheritance, and variation in real organisms is messy at the best of times, and typically involves extremely simple model organisms whose capacities are quite different from those of humans. As a result, much of evolutionary psychology has worked backward — identifying patterns of behavior and coming up with evolutionary explanations for why they are observed. The problem is that this approach is a minefield in terms of logical inference.
Logical Flaws
The process of inferring evolutionary events from contemporary patterns of behavior runs into a classic inference problem. If we think behavior X was evolutionarily advantageous, then we should expect humans to exhibit that behavior. However, observing behavior X now does not mean it is or has been evolutionarily advantageous; this is not a valid logical inference. There are many, many reasons why we might observe behavior X: it might be something we learn to do; it might be a cultural practice that has been handed down; it might be the side effect of some other evolved capacity; it might be random genetic drift, or epigenetic effects, or the result of a bottleneck in human history that allowed only a few random families to survive extinction and pass on their genes. There is no strong reason to think that any particular behavior we observe that people do day-to-day is one that was directly selected by evolution.
The validity of evolutionary claims about why we see certain patterns in human behavior boils down to either this invalid inference, or to axiomatic assumptions that behavior X would be evolutionarily advantageous. These types of assumptions are typically based on a caricature of prehistoric environments. For example, people often make claims based on the assumption that work was divided by gender such that men were hunters and women gatherers. Nearly all of the available physical and DNA evidence refutes this view and indicates instead that women were involved in hunting, and nearly all of the other physical activities that men were, up until the advent of agriculture (around 10,000 years ago).
Solutions and Other Problems
If our typical approaches to understanding the human mind as a product of evolution have led us so far astray, then how can we fix it? Understanding evolution is critical to understanding ourselves; after all, nearly all the capacities that make us human are somehow influenced by evolution.
One approach that has emerged in recent years is that of computational evolution: simulating artificial organisms and examining how they change from generation to generation under selection pressures. This approach allows us to answer questions like “Why might behavior X have evolved?” or “What decision strategies would evolve in environment Y?” An experimenter has complete control over the evolutionary environment and selection pressure, the agents and how they reproduce, and the conditions the agents encounter during their lives. It’s about as close as we can get to lab-controlled evolution, and it works pretty well even for agents with complex, four-base-pair genomes. With some simple tools and activities, you can even watch evolution play out in real-time right before your eyes.
Naturally, computational evolution uses relatively simple organisms and yields artificial intelligence that falls well short of humans. However, as a tool for studying concepts of evolution and questions about how certain behaviors and capacities evolved, it’s hard to match. With enough development, it might just be the breakthrough that allows us to link evolution and psychology. Until then, take most evolutionary claims about human behavior with a grain of salt.
When someone tries to explain differences in behavior as men and women being “wired different” in their brains or convince you that one race or group is genetically superior to another, it’s almost never a scientifically valid claim (or it’s based on bunk science). We simply can’t study the evolution of human behavior that well, and human behavior is so strongly impacted by learning, social, and cultural factors that there is no need to point to genetics. In the near future, we might have the tools to better understand evolution and its effect on our capacities — but such a science might be more productive if it focuses on human universals like memory, perception, decision-making, or language as opposed to our artificial social distinctions.
References
Hansen, C. W., Jensen, P. S., & Skovsgaard, C. V. (2015). Modern gender roles and agricultural history: the Neolithic inheritance. Journal of Economic Growth, 20, 365-404.
Hu, W., Hao, Z., Du, P., Di Vincenzo, F., Manzi, G., Cui, J., ... & Li, H. (2023). Genomic inference of a severe human bottleneck during the Early to Middle Pleistocene transition. Science, 381(6661), 979-984.
Kvam, P. D., Hintze, A., Pleskac, T. J., & Pietraszewski, D. (2019). Computational evolution and ecologically rational decision making. In Taming uncertainty (pp. 285-304). Cambridge, MA: MIT Press.
Hintze, A., Edlund, J. A., Olson, R. S., Knoester, D. B., Schossau, J., Albantakis, L., ... & Adami, C. (2017). Markov brains: A technical introduction. arXiv preprint arXiv:1709.05601.
