Female digger wasps prey on katydids. But they don’t kill them. They paralyze them and then store them in little holes they dig in the ground. They are preparing nests where they will lay eggs and when the eggs hatch, the larvae will feast on the katydids.
Richard Dawkins and John Brockman observed that it sometimes happens that two digger wasps are unknowingly tending the same nest. Naturally, once they figure this out, there’s going to be a fight. Dawkins and Brockman noticed two things about these fights. First, the wasp that wins is usually the one that has contributed more katydids to the common nest. Second, the duration of the fight is predicted by the number of katydids contributed by the eventual loser.
For Dawkins and Brockman the wasps are revealing a sunk-cost fallacy. Evidently, their willingness to fight is not determined by the total reward, but instead by the individual wasp’s past investment. The more they invested, the more they are willing to fight.
A more nuanced interpretation is that the wasps’ behavior is not a fallacy at all, but a clever hack. The wasps really do care about the total value of the nest, but their best estimate of that value is (proportional to) their own contribution to it. For example, a wasps may be able to “remember” the number of katydids she paralyzed (and she must if she is able to condition her fighting intensity on that number) but not be able to count the number of katydids in the nest. The former is going to be correlated with the latter.
Sunk cost bias: a handy trick.