An N2pc was elicited in response to task-irrelevant fearful face singletons among neutral faces on no-change trials. This finding demonstrates that fearful faces can bias the spatial distribution of attention even when attention is allocated to a continuous visual monitoring task at fixation, and peripheral faces can be entirely ignored. The fact that an N2pc was still present shows that the attentional capture by emotionally salient events is not restricted to situations when attention is initially unfocussed (as in previous behavioural visual search studies where observers were preparing to find a target in a visual search array; e.g., Öhman et al., 2001a, 2001b
), but can also be triggered by task-irrelevant fear-related stimuli that are presented outside a narrow central focus of attention.
The enhanced negativity triggered ipsilateral to a neutral face singleton among fearful faces on no-change trials is most likely to represent an N2pc triggered by the fearful face located next to fixation, and opposite to the neutral face singleton. Given that attention was narrowly focused at the screen centre, and all but the two innermost faces were centred at eccentricities of 8.9° of visual angle and beyond (see ), only the face pair closest to fixation may have been processed to the level where emotional expression could be discriminated. If this was the case, arrays containing fearful and neutral singletons would be equivalent in terms of attentional capture by fearful faces, as both contain one fearful and one neutral face close to fixation, and peripheral faces that remain subjectively indeterminate in terms of emotional content. In line with this assumption, the N2pc observed for fearful face singletons and the ‘reversed’ N2pc to neutral face singletons were equivalent in terms of their time course and amplitudes (, bottom).
Finally, N2pc amplitudes were reduced on trials where a task-relevant luminance change occurred at fixation simultaneously with the onset of the face array relative to no-change trials. The attenuation of the N2pc on these trials suggests that the presence of a target, and associated target identification, response selection, and response execution processes, reduce the ability of emotionally salient peripheral stimuli to capture attention. Analogous reductions of N2pc amplitudes as a function of concurrent target processing have recently been reported in ERP studies investigating the attentional blink (Jolicoeur et al., 2006, in press
). In these experiments, the N2pc to peripheral targets was attenuated when these were presented in close temporal proximity to a previous task-relevant stimulus, suggesting that the consolidation of target events in visual short-term memory interferes with subsequent shifts of attention. This may also explain why previous studies in our lab (Eimer et al., 2003; Holmes et al., 2003
) have found that ERP effects of emotional facial expression processing are eliminated for emotional faces at unattended locations, suggesting that the processing of emotional faces is strongly gated by spatial attention. In these studies, unattended faces were always presented simultaneously with other target stimuli, and the processing of non-face targets may have been sufficient to prevent attentional capture.
In summary, the present results provide new electrophysiological evidence for the hypothesis that task-irrelevant fearful faces can trigger attentional capture even when attention is narrowly focused. Capture is reduced by the simultaneous presentation of a target event. Future experiments need to investigate whether analogous results can also be obtained for other facial expressions. For example, attentional capture may be even more pronounced in response to an immediate threat signaled by angry faces.