Confidence ratings (CR) are one of the most frequently used measures in psychological research. However, recent evidence has suggested that eliciting CR from participants may result in changes to cognitive performance, so called reactivity. Here, we examine whether reactivity to CR can be better explained by added task-relevant introspection, or, alternatively, the unintentional priming of confidence-related beliefs. First, we compare participants' performance in a group making CR with a group making a task-irrelevant control rating, and a second group who made the same task-irrelevant rating, but with the word 'confident' included in the rating's wording. The results suggest that reactivity is driven by the presentation of the word 'confident', and reactivity does not require task-relevant introspection. Additionally, we show that rephrasing CR to remove the word 'confident' neutralises reactivity. This suggests that reactivity may represent a significant problem for researchers using CR, but rephrasing CR may remedy these concerns in relatively simple fashion.
Keywords: Confidence ratings; Metacognition; Priming; Reactivity.