Our choice involves a degree of uncertainty. For example, we may easily decide that a 90-year-old man is old, while it would be unclear whether we can call a 50-year-old man old. This exemplifies that it is harder to make a categorical choice as a stimulus approaches the boundary of categories. In this example, the boundary between old and young would be closer to 50 than 90. This is why it is harder to classify a 50-year-old man. Critically, the boundary is not stable but adjusted trial-by-trial as we experience new stimuli. Because of this boundary updating, it is known that the boundary updating induces a bias in choice called the repulsive bias. However, it has been unknown whether the boundary updating also induces a historical bias on decision uncertainty. Here, we addressed this issue by investigating three measurements that are associated with decision uncertainty. They are response time (RT), pupil size, and the signal of the dorsal anterior cingulate cortex (dACC). We found that all of them reflected the bias induced by boundary updating. Specifically, we simulated the boundary updating model and derived that the uncertainty level increased as the current choice is more congruent with the previous stimulus. We call this effect the pre-congruence effect. The pre-congruence effect was found in all three measurements of decision certainty. This finding suggests that the updating of the category boundary not only biases choice but also decision uncertainty.