|
Abstract
|
Scheduling theory concerns the development of policies determining the optimal allocation of resources to a set of tasks. Scheduling problems have been studied extensively in the context of operations research and computer science, where optimal policies have been established for many cases, but little research has examined how people perform with respect to these optimal policies despite clear applications to selective attention and behavioral allocation. We conducted several experiments in which each subtask is a random dot motion (RDM) judgment, with difficulty determined by the coherence of the motion. Participants were presented with 4 or more subtasks, which are selected one at a time and then completed. We are concerned with the order in which subtasks are selected for completion. When subtasks vary in difficulty but have the same reward value, scheduling is optimal when tasks are completed from easiest to hardest. When reward also varies, then an optimal order index can be constructed from the ratio of reward to completion time. We compared a number of manipulations varying whether tasks appeared in a consistent or varied location, the deadline for completing the four subtasks, how subtask difficulty was indicated, and the nature of the reward (e.g., game-based or points-based). We introduce novel measures of performance and a model of scheduling allowing an analysis of optimality across conditions. In general, performance is more optimal or near optimal responding when rewards are equivalent, when task difficulty is easily determined, when task location is fixed, and when there is a deadline.
|