In human-machine interaction, automation brings both advantages and potentially unpredictable disadvantages to human cognitive performance. In this study, we hypothesized that active behavior improves cognitive performance in human-machine interaction, and verified this hypothesis through three experiments. Experiment 1 examined the relationship between activeness and reaction time in a target-search task. Experiments 2 and 3 analyzed the factors that improved cognitive performance. Experimental results demonstrated that activeness positively affects cognitive performance and suggested that predictability associated with activeness plays a key role in improving cognitive performance.