This is a response to Steve Randy Waldman’s post about Rational Regrets, which I regret not commenting in. SRW imagines an option between a guaranteed average payout and a risky high earning payout. In the world of expected utility, we should choose the risky payout, but because most people who choose this payout will lose, will will end up, ex-post in a world where most people have regrets that they did not choose the safe payout.
And this makes me think of evolution — i.e. where does our sense for valuing risk and reward come from?
Suppose, for example, that our utility functions are the result of a toy natural selection process over the 200,000 years or so that we survived as hunter-gatherers. OK, I don’t know anything about evolutionary biology, utility theory, or hunter gatherers, but I know how to Google! Let’s say that early man lived in tribes of about 40 people, with a gini coefficient of about .25, with the richest individuals enjoying at most about twice the reproductive success of the average individual.
Now in a bet in which you have a 90% chance of being barely alive (say Reproductive Success of 0.5xAverage) and a 10% chance of being filthy rich (which is bounded above by a reproductive success of 2xAverage) means that if you are likely to take this bet, then your expected reproductive success will be less than if you take the guaranteed safe bet (and get Average Reproductive Success).
.9 * .5 Average Reproductive success + .1 * 2 *Average Reproductive success = 0.65 Average Reproductive success < 1*Average Reproductive success
And in that case, our utility function would have been selected by evolution to make choices so that we do not live in a world of rational regret. What would such a utility function look like? Well, it would depend on the definition of “average”, i.e. if the median personal income in your group is $30,000, then your personal reference-dependent utility function might, in a very degenerate sense, be defined as u(y) = (2 – 30,000/y), where y is real income. We are not using the full notion of reference dependent utility here, just something very simple: I.e. the number might be the average income of your peers, or perhaps be replaced by what you expect the lifetime NPV income of your peers to be, etc.
This is just for fun, and we can speculate that evolution would condition us to value other things that might affect reproductive fitness as well — e.g. social rank. But the main point is that if for the majority of our species’ lives, there were small and bounded (for each individual) relative reproductive gains to obtaining large sums of wealth then this might affect how we instinctively value these types of tradeoffs and why most people now would avoid rational regret.