Nothing “Evaluative“ about Evaluative Priming? Semantic Relatedness Beyond Evaluation Is Not Necessary for a Facilitation of Evaluative Judgements

Ever since Zajonc proposed that preferences need no inferences, psychologists have debated whether evaluations can be conceptualized and measured independently of co-activated semantic content. One prominent arena for this debate has been the Evaluative Priming Task, which tracks the effects of conflicting or congruent evaluative relationships between stimuli. Surprisingly, two prior lines of research found that effects of evaluative congruency in the task do not emerge in the absence of additional semantic relationships. This not only conflicts with other findings, but also challenges dominant theoretical accounts of the task, whose assumptions entail that congruency effects should arise irrespectively. In Studies 1 and 2 (Ns=81, 85), we re-examine these two lines of research and show that when vulnerabilities in the original experimental and analytical designs are rectified, evaluative congruency effects do reliably occur (even in the absence of strong semantic relationships). A third study (N=80), that implements strong semantic relationships under especially controlled conditions, also yields reliable congruency effects. In Study 4 (N=81), the trials with low semantic relatedness from the first three studies are combined, replicating consistent and reliable, but small congruency effects. Our results thereby strengthen models that accommodate an independent role of evaluations, and we illustrate how these models might be modified to account for an augmenting influence of semantic relatedness. Currently, in domains for which evaluations and their measurement are integral (like the measurement of race-based evaluations, consumer or health behavior), the pronounced and potentially biasing effects of semantic relatedness remain largely unrecognized as a serious confound.