Learning from Forced Completion vs the Option to Opt Out: An Experiment on a Hybrid of the Game of 21 and the Centipede Game

This study examines how the option to opt out facilitates learning in the Game of 21. It compares the performance of subjects in the Game of 21 and a Hybrid Centipede-21 Game where players can opt out of the game at any time and receive a decreasing payment for doing so as the game progresses. Additionally, the experiment introduces a novel concept of a ``dumb computer" that always makes suboptimal decisions. Performance against the ``dumb computer" determines the level of foresight of subjects in order to compare the amount of learning between the Hybrid Centipede-21 Game and the traditional Game of 21. Results indicate players drop out strategically and earlier as the game progresses with a few ending immediately as the backward induction solution predicts; however, contrary to our predictions, players learn better when forced to finish the game, particularly women. Some of the difference in learning occurs due to a small set of players giving up when they have the option to opt out while the rest may be attributed to the larger role strategic uncertainly plays or the additional complexity of the game.