Too bad: Bias for angry faces in social anxiety interferes with identity processing

摘要

The recognition of faces across incidences is a complex function of the human brain and a crucial ability for communication and daily interactions. This first study on ERP correlates of emotional face learning in social anxiety disorder (SAD) investigates whether the known attentional bias for threatening faces leads to a corresponding memory bias. Therefore, 21 patients with SAD and 21 healthy controls (HCs) learned faces with emotional facial expressions (neutral, happy, and angry) and were later asked to recognize these out of novel identities all presented with a neutral facial expression. EEG was recorded throughout. Behaviorally, the faces' emotional expression modulated later recognition in terms of accuracy, response times, signal detection parameters and ratings of valence, but with better performance for happy than angry faces in HC as well as in SAD. In the learning phase, attention- and memory-associated event related potentials (ERPs) P100, N170, P200, N250/EPN, and LPP indicated enhanced processing of angry faces, which was restricted to patients with SAD in N250/EPN and LPP. In the test phase, familiarity effects emerged in N250, FN400 and LPP. While N250 was affected by learned-angry faces, FN400 and LPP reflected image learning of neutral faces, which was restricted to SAD in LPP. We replicated the attentional bias to threatening faces, which was not restricted to early ERP components, but was prolonged to later stages of conscious processing, especially in SAD. In contrast to what had been expected, sustained hypervigilance to the emotional content seems to have impaired the processing of the facial identity, resulting in a happy face advantage at the behavioral level. This could be explained by prominent models assuming separate processing of facial emotion and identity. Hypervigilance in SAD might be a disadvantage in those studies focusing on other aspects of face processing than emotion. (C) 2016 Elsevier Ltd. All rights reserved.