Why humans care about sunk costs while animals don't: an evolutionary explanation

While humans often care about sunk investment, animals are not subject to this sort of sunk cost behavior or "Concorde fallacy". This paper investigates a simple two stage decision problem under uncertainty. At the second stage, subjects can commit the Concorde fallacy by sticking to the first stage decision, independent of the state of nature revealed in-between. We investigate whether this can be beneficial in a standard payoff monotonic adaptation process. Committing the Concorde fallacy reduces the payoffs but accelerates the adaptation since it acts like "self-punishment". It will, however, not only reduce the population growth rate in the long run but also the population size at any point in time in a biological evolutionary process. In this sense, animals can never benefit from the Concorde fallacy. Risk aversion gives an extra benefit to a behavior that more rapidly learns to avoid bad outcomes. If the wrong initial decision leads occasionally, albeit very infrequently, to a very low payoff, then risk averse humans will be better off by committing the Concorde fallacy.