Abstract
Visual search studies have shown that detection of a highly familiar (one's own) face compared to a recently learned face is particularly efficient, perhaps because highly familiar faces require less attention (Tong & Nakayama, 1998). To test this idea, we employed the attentional blink (AB) paradigm, a procedure that temporally divides attention between two successive targets imbedded in a rapid serial presentation (RSVP) of stimuli. Typically, the second target (T2) is poorly detected if it succeeds the first target (T1) by less than 500 ms (Raymond, Shapiro & Arnell, 1992). However, when T2 is a highly salient and familiar word (one's own name), no AB is found (Shapiro, Caldwell & Sorensen, 1997), suggesting that familiarity aids recognition when attention is divided. Using grey-scale face images presented in RSVP (85 ms/item), we asked observers to detect as T2 the presence versus absence of a famous or recently learned novel face. The T1 task required categorisation of an elliptical, abstract, patterned image. When T2 was a recently learned face, detection varied significantly with the duration of the T1 – T2 interval (an AB effect), being poorest for values less than 500 ms. However, when T2 was a famous face, no effect of lag was observed and performance at all lags was better than that for recently learned faces (although not at ceiling). These findings indicate that an AB can be reliably obtained with face stimuli and that highly familiar faces require less attentional resource for recognition than recently learned ones.