Abstract
How is long-term memory for faces impacted by the presence of uninformative face masks? In an initial across-subject experiment, subjects were tasked with remembering 40 faces without masks (N=20) and 40 faces with surgical masks (N=20). After a delay, they were presented with a 2-AFC memory test. Our results were consistent with existing research showing that face masks disrupt holistic processing (Freud et. al 2020), in that we found worse memory performance for masked faces than those without masks. The second experiment tested whether memory for faces is impacted by face masks that are known to be uninformative for the memory test, but are always present. Subjects (N=40) studied 80 faces each with uniquely patterned face masks. Critically, during the memory test, subjects were shown 40 “old” faces with the same unique masks originally studied, and 40 “old” faces with new unique masks that were not seen before. The 2-AFC always paired old faces with foil faces with matching masks to ensure the masks were not informative for discrimination. We asked (1) whether people effectively ignore masks during encoding and (2) how much a novel mask affects memory for a previously seen face. We found an encoding specificity effect: better performance for faces with unique masks that were the same at study and at test, compared to faces with unique masks which were different between study and test, even though memory for the mask itself was completely uninformative at test (t(36)=3.30, p=0.002, dz=0.54). This was the case even though the instructions stated that masks will not be helpful for memory. Overall, these results suggest that face masks harm long-term memory for faces, and that face masks different from those which were encoded with the face further harm memory for faces, even when people are explicitly told to ignore the masks.