We did experiments in a three-round bargaining game where the (perfect) equilibrium offer was $1.25 and an equal split was $2.50. The average offer was $2.11. Patterns of information search (measured with a computerized information display) show limited lookahead rather than backward induction. Equilibrium theories which adjust for social utilities (reflecting inequality-aversion or reciprocity) cannot explain the results because they predict subjects will make equilibrium offers to "robot" players, but offers to robots are only a little lower. When trained subjects (who quickly learned to do backward induction) bargained with untrained subjects, offers ended up halfway between equilibrium and $2.11.
Johnson, Eric, C. Camerer, S. Sen, and T. Rymon. "Detecting Failures of Backward Induction: Monitoring Information Search in Sequential Bargaining." Journal of Economic Theory 104, no. 1 (2002): 16-47.
Each author name for a Columbia Business School faculty member is linked to a faculty research page, which lists additional publications by that faculty member.
Each topic is linked to an index of publications on that topic.