Automation bias is one kind of misuse in human interaction with automated decision aids, which indicates the phenomenon that operators take information from automation as a heuristic in decision making, and give up cautious information processing. Two main kinds of error are due to automation bias: 1, Omission error, which means operators miss the signal due to over trust in automation or lack efficient environmental awareness, when automation does not alert the key object. 2, Commission error, which refers to the phenomenon that operators falsely respond to the signal or the key object, when automation makes inappropriate recommendations or false alarms. The percentage difference in frequency between omission errors and commission errors is the main issue of this thesis. Exp 1. The error types of automation (i.e. criterion of a signal detection task) were manipulated to compare whether false alarm-prone systems and miss-prone systems would induce different automation bias numbers. Results indicated that miss-prone systems induced more automation bias, i.e. omission errors were more likely to happen than commission errors. Participants’ trust in, and the perceived utility of, miss-prone systems are higher; however, miss-prone systems harmed participants’ task sensitivity more than false alarm-prone systems. Exp 2. Participants would lose 0.1 RMB in this experiment, if he / she committed an incorrect response. However, in experiment 1, the award rule is to give participants 0.1 RMB, if he / she made a correct response. Results indicated that omission errors were still more likely to happen than commission errors. And also, participants showed less automation biases than in experiment 1. Exp 3. Compared to experiment 1, time urgency was manipulated. Participants had to respond within 1000ms. Results showed that, there was no difference between omission errors and commission errors in this experiment. And also, participants made less automation biases than in experiment 1.