Assessing the long-term face memory of highly superior and typical-ability short-term face recognisers

Outstanding long-term unfamiliar face recognition ability is the hallmark of the exceptionally skilled ‘super-recognisers’. Indeed, international police super-recogniser units rely on members identifying unfamiliar suspects from CCTV after substantial delays. Yet, virtually all research has employed brief retention intervals to evaluate super-recognisers’ capabilities. To address this gap in the literature, in Experiment 1, participants (n = 597), including 84 super-recognisers, viewed 10 60s target-actor videos and identified targets from 10 target-present line-ups after random delays of 1, 7, 14, 28, or 56 days. In Experiment 2, participants (n = 1421), including 206 super-recognisers, viewed 20 30s target-actor videos and identified targets from 10 target-present and 10 target-absent line-ups after delays of virtually none, 1, 7, and 28 days. Correct target-present identifications, and with smaller effect sizes, correct target-absent line-up rejections were predicted by short-term face memory and matching test scores, and decision confidence. Delay also predicted correct target-present identifications but not correct rejections. With higher confidence, most super-recognisers (82.5%) individually exceeded overall control mean hit rate accuracy. However, only a minority of comparisons were significant (28.5%). The important applied implication was the demonstration that not all super-recognisers can sustain their skills over longer retention intervals. Therefore, recruitment to super-recogniser research groups, or roles in policing or security require longer-term face memory tests. Important theoretically was the finding that super-recogniser’s estimated forgetting curve was shallower than that of controls, implying that their enhanced mnemonic system for faces allows representations in memory to be more reliably sustained for far longer.