Abstract
We used the Remember–Know procedure (Tulving, Citation1985) to test the behavioural expression of memory following indirect and direct forms of emotional processing at encoding. Participants (N=32) viewed a series of facial expressions (happy, fearful, angry, and neutral) while performing tasks involving either indirect (gender discrimination) or direct (emotion discrimination) emotion processing. After a delay, participants completed a surprise recognition memory test. Our results revealed that indirect encoding of emotion produced enhanced memory for fearful faces whereas direct encoding of emotion produced enhanced memory for angry faces. In contrast, happy faces were better remembered than neutral faces after both indirect and direct encoding tasks. These findings suggest that fearful and angry faces benefit from a recollective advantage when they are encoded in a way that is consistent with the predictive nature of their threat. We propose that the broad memory advantage for happy faces may reflect a form of cognitive flexibility that is specific to positive emotions.
Acknowledgements
This research was supported in part by grants to the third author from the Natural Sciences and Engineering Research Council of Canada and a Faculty of Arts Small Projects Grant (Ryerson University) to the second author. We are also grateful to the Ontario Graduate Scholarship and the Ontario Mental Health Foundation who provided generous support to the first author during the completion of this work.
We thank Lixia Yang, Julia Spaniol, Paul Whalen, and Andrew Yonelinas for helpful comments on earlier versions of this manuscript. We also thank Wojtek Grabski, April Khademi, and Danoush Hosseinzadeh for their technical assistance and Courtney Gray for her assistance with data collection.
Notes
1Long ISI intervals were used to collect measures of physiological arousal; these data are not presented here.
2False alarms (FA) were not distinguishable by encoding task. New faces were matched to old faces by emotion type only and presented in the context of a single recognition task. Thus, when calculating any measure that required an FA correction, we equally applied the false alarm rate to both the indirect and direct conditions.