Abstract
We investigated whether task-irrelevant stimulus attributes of faces are encoded alongside relevant attributes. If so, repetition of an attribute present at study should facilitate performance when that attribute is repeated at test. In the study task, subjects classified faces along one dimension (emotion or gender) that was perfectly correlated with an irrelevant dimension (e.g., happy faces were always female for one group). At test, half of the faces were repeated from study (old) and half were new. Old faces were shown with the same emotion as seen during study (gender and emotion uncrossed) or with the opposite emotion (gender and emotion crossed) along with uncrossed and crossed foils. Subjects indicated whether the face was old or new (explicit task) or made a speeded judgment about the emotion or gender of the face (implicit task). Regardless of study condition, explicit recognition of crossed faces was at chance whereas recognition of uncrossed faces was demonstrated on 80% of trials. Crossing increased false recognition of new faces by about 10%, and this effect was somewhat larger when emotion was studied than when gender was studied. In the implicit memory task, emotion classification times were faster for old uncrossed items than for crossed items (a priming effect), and more priming was observed when emotion was studied, rather than gender. No priming was shown in the gender classification task. We found that: 1) explicit recognition is strongly affected by irrelevant attributes encoded at study; 2) encoding irrelevant attributes facilitates emotion classification (but not gender classification) and 3) implicit memory performance can be predicted from explicit memory performance suggesting that explicit recollection mediates the implicit memory contribution to the classification task.
This project was supported by a grant from the Natural Sciences and Engineering Research Council.