Do People Mentally Represent Automated Tasks? Evidence from Task-Switching Costs Following Takeovers

Abstract

Partially automated vehicles, in which automation performs parts of the driving task, introduce new challenges of automation monitoring and human-automation teaming to the driving experience. We describe a new method for measuring whether operators mentally represent automated task performance using a version of the joint task-switching (JTS) task paradigm. In the JTS task, an operator and a teammate take turns performing two intermixed tasks. Following takeover from the teammate, task-switching costs (slower responses following a task switch vs repetition) indicate that the operator mentally represents the teammate’s performance. We measured performance following task switches and repetitions with and without a takeover from automation. Switch costs disappeared following takeovers, indicating a lack of representation of the prior automated task. We discuss how task switch costs following takeovers can serve as indirect measures of whether operators mentally represent automated task performance in mixed automation situations.