The aim of this experiment was to examine the effect of different instructional strategies on student teachers' confirmation bias. Confirmation bias refers to the selectivity in finding and using evidence that fits one's own beliefs or hypotheses while neglecting evidence that is opposite to one's own beliefs or hypotheses (Nickerson, 1998). Dutch student teachers (n = 141) took a confirmation bias pre-test and were then randomly assigned to three conditions; teaching on video (TOV), preparing to teach (PTT) and re-study (CC). All participants received text-based instruction on confirmation bias and how it can be mitigated. They also practised with confirmation bias tasks and they received feedback on their answers. Subsequently, participants in the TOV and PTT conditions prepared a lesson about the instructional content and in the TOV condition they taught this lesson on video. After the learning phase, TOV and PTT participants completed a social presence questionnaire. All participants completed an arousal questionnaire and a confirmation bias post-test and a transfer test. The results showed that confirmation bias was reduced to a similar extent in all conditions. Results also showed that the quality of the prepared lesson was highest for TOV participants suggesting they had gained better understanding of the confirmation bias than PTT participants. Furthermore, in contrast to our expectations, PTT participants reported highest social presence scores. TOV participants experienced higher arousal levels compared to CC participants. Transfer scores did not differ between conditions. We discuss theoretical explanations of the findings from the present study.