Identification of angry faces in the attentional blink

Abstract

According to cognitive and neural theories of emotion, attentional processing of innate threat stimuli, such as angry facial expressions, is prioritised over neutral stimuli. To test this hypothesis, the present study used a modified version of the rapid serial visual presentation (RSVP) paradigm to investigate the effect of emotional face stimuli on the attentional blink (AB). The target stimuli were schematic faces which depicted threatening (angry), positive or neutral facial expressions. Results showed that performance accuracy was enhanced (i.e., the AB was reduced) on trials in which the second target was an angry face, rather than a neutral face. Results extend previous research by demonstrating that angry faces reduce the AB, and that this effect is found for schematic facial expressions. These findings further support the proposal that, when there is competition for attentional resources, threat stimuli are given higher priority in processing compared with non-threatening stimuli.

Acknowledgements

This research was funded by the Wellcome Trust (reference 051076).

We thank Sophia Durrani for her help with pilot work on a related project and helpful comments on an earlier draft of the manuscript.

Francis Maratos is now at the Centre for Psychological Research in Human ehaviour, University of Derby, UK.

Notes

¹The condition in which there was no distractor between T1 and T2 was not included in the analyses because the AB effect is often not found in this position (so-called “Lag-1 sparing”) and there is evidence that different mechanisms are responsible for Lag-1 sparing and the AB effect (Chun & Potter, 1995; Hommel & Akyurek, 2005; Shapiro et al., 1997a). As the aim of the study was to investigate the effect of emotional information on the AB, the main hypothesis concerned identification of T2 faces during the time window of the AB (i.e., 200–400 ms following T1) versus after the AB (>500 ms after T1).

²Targets presented at serial positions 7–10, 11–14 and 15–18 formed the early, mid and late serial position levels, respectively. The data were collapsed from 12 positions into these three categories to reduce floor effects in the analyses.