It sounds counterintuitive at first, but if you think of real world examples, it makes a lot of sense. It’s the entire principle that casinos and blind bag toys operate on: you do the requested action (like placing a bet or buying the blind bag), you get something you didn’t want or expect, and you get a little mad that it didn’t go the way you wanted, so you do the requested action again.

Yup, I suspect the lack of consistency, especially for dependant animals and gamblers, drives anxiety, which is disproportionately relieved by a successful outcome, which is a recurring survival driver in the wild (again I blame evolution, where persistence can be rewarded by survival)

“Diffusion of responsibility” is a good one, guessing it's testable.

In that vein, we have the Stanford prison experiment, though it's repeatability seems to be questioned.