Unintended effects of memory on decision making: A breakdown in access control

A hybrid evaluative-conditioning/source-monitoring paradigm is put forward as an alternative to the standard evaluative-conditioning paradigm. The first experiment paired brand names with a small number of attractive or unattractive female faces and used a likeability rating task as well as a source monitoring task. The second experiment paired words which differed along a masculine–feminine dimension with male and female faces, and used a speeded judgment about whether words were stereotypically masculine or feminine. The third experiment paired words that differed along an active–inactive dimension with male and female faces and used a variation of the Implicit Association Test where judgments about whether words were active or inactive were mixed with judgments about whether faces were male or female. In all three experiments, we observed transfer between the recently acquired information and the judgment task. In addition, the three experiments progressively reduce the probability of demand characteristics. We explain the results in this paradigm, and in many other paradigms, as a breakdown in access control. We also point to several similarities between existing theories of evaluative conditioning and memory phenomena/theories that have gone unnoticed in the evaluative conditioning literature.