Abstract:

Moran et al. (2021) report a multi-lab registered replication of Olson and Fazio’s (2001) surveillance task. The surveillance task is an incidental learning procedure over the course of which participants observe pairings of conditioned stimuli (CSs) and unconditioned stimuli (USs) while engaging in a distracting secondary task. Unaware evaluative conditioning (EC) effects are inferred if participants who fail to report the CS–US contingencies on a post-hoc measure show preference for the CSpos over the CSneg. Moran et al. claimed to establish such effects relying on the criteria used by Olson and Fazio to exclude contingency aware participants from analyses. Here we reexamine Moran et al.’s data using more fine-grained analytic strategies. We show that the contingency awareness measures used by Olson and Fazio and, by extension, Moran et al. lack adequate reliability and validity. Moreover, even assuming valid awareness measures, Bayesian analyses did not provide unambiguous evidence for unaware EC effects under any exclusion criterion and provided decisive evidence against such effects in most models. Finally, a separate analysis that distinguished between fully aware, partially aware, and fully unaware participants shows that evidence for unaware EC is due to the inclusion of partially aware participants in the purportedly unaware subsample. These reanalyses suggest that unaware EC as indexed by the surveillance task has yet to be convincingly demonstrated. We discuss the conceptual, theoretical, and applied implications of these findings with regard to the potential for unaware attitude formation.

Keywords: