Abstract

Rationality has taken a hit recently. A veritable torrent of work – both technical and popular – in psychology and economics has challenged the notion that the human brain is designed to make rational decisions. However, this observation raises a paradox. By almost any measure, Homo sapiens is a spectacularly successful species. From humble origins approximately two million years ago, humans have grown to a population that exceeds seven billion and have colonized – and come to dominate – nearly every terrestrial biome. This phenomenal growth suggests that, on average, our ancestors made very good decisions. Yet this work from psychology and economics suggests that the decision-making software that our brains run is profoundly flawed — that we are, in a word, irrational. How is it possible that a species apparently so defective in its ability to generate sound decisions can be so incredibly successful?

Humans are biological entities and, as such, we are subject to the laws of evolution. When the rules for rational decision-making were discovered and formalized, however, this fact was ignored. It turns out that the rules for a living organism, anchored in the present and subject to a force of selection which is extremely averse to extinction, are quite different from the rules of abstract, formal rationality. I will show how the all-important need to avoid extinction in a world that is at best incompletely known has profound implications for preferences, utility, and rationality. By ignoring the condition of existential uncertainty, the theory of rational decision-making has developed distorted expectations of how an organism working in its own best interest should behave.

I work through how an evolutionary perspective on decision-making changes our interpretations of two examples from the canon of human fallibility, namely, time-inconsistent preferences and non-probabilistic weighting of decisions. What we see is that uncertainty, in fact, favors what has come to be generally known (albeit incorrectly) as ‘hyperbolic discounting.’ Even if the decision-maker has a foundation of consistent preferences, the existence of uncertainty means that the best decision-making policy will favor earlier payoffs and be more indifferent to time delays later on. In other words, optimal time preferences in the presence of uncertainty look an awful lot like the dreaded irrational time-inconsistency. I then discuss work in which we have shown that (evolutionary) fitness maximizing agents will show the type of putatively irrational weighting of risky decisions that Kahneman and Tversky famously discuss in their work on Prospect Theory.