Different types of facial description alter the confidence-accuracy relationship

Eyewitnesses commonly provide facial descriptions after the commission of a crime. Although research suggests that this description may improve memory (i.e., verbal facilitation), less is known about what types of descriptions (e.g., parts-based featural, trait-based holistic) are most suitable to maximize accuracy. To address this gap, participants studied two blocks of 12 faces (i.e., 24 total study faces) and performed either a Description (i.e., featural, trait) or Control (i.e., counting, addition) task after each (condition names representing block order). Following a short distractor, participants completed two types of recognition memory tests. The inclusion test instructed participants to respond "yes" to all studied faces in both blocks and "no" to new faces. Conversely, the exclusion test instructed participants to respond "yes" to studied faces in Block 1 and "no" to studied faces in Block 2 and new faces. Participants subsequently provided confidence and source-memory judgments. With respect to accuracy, either type of Description (as opposed to Control) facilitated memory. Participants who completed a Description-Control or Control-Description block order combination outperformed Control-Control participants. Participants' exclusion test performance showed a similar description advantage; Description-Control outperformed Control-Description and Control-Control. Although a general description advantage emerged for overall accuracy, the confidence-accuracy relationship manifested in different ways. The number of trait descriptors increased accuracy without affecting confidence. The number of featural descriptors increased confidence without affecting accuracy. We discuss these findings and their implications for guiding person descriptions for different types of eyewitnesses with and without advanced preparation of witnessing a criminal event.