Dunsmoor and Murphy (2014) found that conditioned fear spreads from typical to atypical category exemplars but not vice versa, paralleling the principles of category-based induction. Their work was foundational in establishing that fear can generalize within conceptual structures along a dimension of category typicality and offers insight into how higher-order reasoning might be integrated in classic conditioning. This registered report aims to replicate and extend the finding of Dunsmoor and Murphy by investigating how idiosyncratic category structures and trait anxiety affect the asymmetry effect in conceptual fear generalization. It includes several methodological improvements and a comparison of different methods for processing skin conductance responses to allow for an identification of the source of contradicting findings on the effect in the literature.
Dunsmoor and Murphy (2014) found that conditioned fear spreads from typical to atypical category exemplars but not vice versa, paralleling the principles of category-based induction. Their work was foundational in establishing that fear can generalize within conceptual structures along a dimension of category typicality and offers insight into how higher-order reasoning might be integrated in classical conditioning. This registered report aims to replicate and extend the findings of Dunsmoor and Murphy by investigating how idiosyncratic category structures and trait anxiety affect the asymmetry effect in conceptual fear generalization. It includes several methodological improvements and a comparison of different methods for processing skin conductance responses to identify the source of contradictory findings on the effect in the literature. Results from this multiverse approach revealed no asymmetry effect in explicit expectancy ratings or in skin conductance responses, contrary to the original study’s findings. These findings suggest that while conceptual fear generalization relates to typicality gradients, the asymmetry effect is neither as robust nor as universal as originally proposed. The original premise of a direct parallel between associative learning and category-based induction warrants reconsideration, with additional research needed to clarify the conditions under which these cognitive domains interact.