Testing the evaluative nature of masked-prime conditioning with an immediate affective-matching task and improved adaptive presentation

Results from the Greenwald and De Houwer (2017) paradigm with valent targets have been interpreted as evidence for unconscious conditioning. However, it remains unclear whether the observed learning effect reflects an evaluative change in conditioned stimuli (CSs). We test this question by combining the Greenwald and De Houwer (2017) procedure with the affective-matching task (Klauer & Musch, 2002), which serves as a measure of evaluative meaning. In a preceding experiment, matchpoint3, we replicated the standard Greenwald and De Houwer (2017) conditioning effect, but CSs did not produce an affective-matching effect. In contrast, pre-experimentally valent words produced the expected affective-matching effect, indicating that the task implementation was sensitive to evaluative congruency effects. The present experiment addresses the possibility that an initially present evaluative effect was attenuated by intervening tasks in this experiment. A first follow-up, matchpoint4, tested affective matching immediately after acquisition but failed to replicate the valent-word positive-control effect. High accuracy may have limited sensitivity to valence congruency in that study. The present experiment, matchpoint5, repeats the immediate-test procedure with a revised QUEST adaptive staircase that is aimed at maintaining performance below ceiling. Participants will complete the Greenwald and De Houwer (2017) acquisition phase followed immediately by an affective-matching task with CSs as primes. A second affective-matching task with pre-experimentally valent words will serve as a positive control. Evidence for or against both effects will be evaluated using Bayesian optional stopping.