Human Error as an Emergent Property of Action Selection and Task Place-Holding
该研究通过计算过程模型模拟人类认知机制，解释了中断如何导致多种错误类型，并发现提供更多任务状态线索可减轻中断引发的错误。
OBJECTIVE: A computational process model could explain how the dynamic interaction of human cognitive mechanisms produces each of multiple error types. BACKGROUND: With increasing capability and complexity of technological systems, the potential severity of consequences of human error is magnified. Interruption greatly increases people's error rates, as does the presence of other information to maintain in an active state. METHOD: The model executed as a software-instantiated Monte Carlo simulation. It drew on theoretical constructs such as associative spreading activation for prospective memory, explicit rehearsal strategies as a deliberate cognitive operation to aid retrospective memory, and decay. RESULTS: The model replicated the 30% effect of interruptions on postcompletion error in Ratwani and Trafton's Stock Trader task, the 45% interaction effect on postcompletion error of working memory capacity and working memory load from Byrne and Bovair's Phaser Task, as well as the 5% perseveration and 3% omission effects of interruption from the UNRAVEL Task. CONCLUSION: Error classes including perseveration, omission, and postcompletion error fall naturally out of the theory. APPLICATION: The model explains post-interruption error in terms of task state representation and priming for recall of subsequent steps. Its performance suggests that task environments providing more cues to current task state will mitigate error caused by interruption. For example, interfaces could provide labeled progress indicators or facilities for operators to quickly write notes about their task states when interrupted.