Abstract
How do people decide to abandon a problem? Participants were presented with unsolvable water jar problems, having been accurately informed of the prior probability of solvability. Across three experiments, we discovered effects of prior probability of solvability and of problem size (number of distinct problem states) on measures of effort and confidence. If a problem is more likely to be solvable and allows more problem states, a problem solver spends longer trying to solve the problem. Giving-up decisions are informed by the same judgments of probability of success and costs of solution that inform move-choice in a rational model of problem solving.