Predictors of belief bias in critical thinking tasks

Description

Belief biases are the subset of biases caused by the inappropriate influence of prior knowledge and/or personal beliefs on reasoning. The current study explored whether instructions to override prior knowledge and/or personal beliefs reduced the magnitude of the belief bias effect. Furthermore, are the individual difference factors cognitive ability and thinking dispositions stable predictors of belief bias? Do task factors influence bias?Participants (195 university students) were randomly assigned to either an 'override' group (explicitly instructed to put aside prior beliefs), or a 'non-directive' group (no reference to prior beliefs). Participants completed a test battery including a formal critical thinking task that featured prior beliefs low in emotional engagement (syllogisms that brought prior knowledge into conflict with logical validity). One informal critical thinking task required generating arguments about two different topics---one topic featured prior beliefs high in emotional engagement while the other topic was low in emotional engagement. Another informal task featured prior beliefs high in emotional engagement and required the evaluation of two experiments with opposing conclusions. Prior beliefs were captured along with measures of cognitive ability and thinking dispositions.The results revealed that instructions can reduce bias on formal and some informal tasks. Individual difference factors contributed to performance on formal tasks with low levels of emotional engagement but not informal tasks (whether high or low in levels of emotional engagement). The results suggest that educators and policy makers should emphasize the importance of 'considering the opposite' in decision-making situations---particularly when the topic under consideration pertains to strongly held personal beliefs.The override instructions significantly reduced the belief bias effect for two tasks. Bias was reduced by 36% for the syllogisms, and by an average of 72% for the argument generation task, while no between-group difference existed for the experiment evaluation task. The override instructions reduced bias for the syllogisms by 63% for participants in the lowest range of cognitive ability compared to participants in the middle (7.4%) and highest (1.2%) ranges. Cognitive ability and thinking dispositions explained performance on the syllogisms but, surprisingly, did not explain performance on the argument generation and experiment evaluation tasks.

Subjects

Links

• URL