Proceedings of the Annual Meeting of the Cognitive Science Society

People routinely collaborate to accomplish goals more quickly, but deciding who should do which tasks, and when, poses a formidable coordination problem. How do people allocate work across collaborators? Related problems have been studied in distributed computer systems, where load balancing algorithms divide heterogeneous tasks among processors with heterogeneous capabilities. Here, we investigate whether human scheduling reflects sophisticated principles from load balancing or simpler heuristics such as taking turns. In a restaurant management task, participants (N=99) assigned customer orders to chefs with heterogeneous processing speeds, either planning schedules in advance or allocating tasks dynamically. People substantially outperformed simple heuristics, discovering efficient assignments that jointly accounted for collaborator and task heterogeneity. We also uncover systematic biases, including differences between upfront and sequential planning and a preference for completing shorter jobs early. More broadly, linking computational load-balancing theories with cognitive theories of coordination offers a valuable framework for understanding human collaboration.