Abstract
Attentional biases toward threatening faces have repeatedly been studied in the context of social anxiety, with etiological theories suggesting exacerbated biases as a possible cause for the latter. To counteract these postulated effects, research has focused on the concept of attentional bias manipulation (ABM), in which spatial contingencies between succeeding stimuli are traditionally employed in training paradigms designed to deliberately shift automatic attention processes away from threat-related stimuli. The ABM research field has been faced with various methodological challenges, such as inconsistent results, low reliabilities of dependent variables and a high susceptibility to moderating factors. We aimed to combine several recent approaches to address these issues. Drawing upon theories of value-driven attention, we explored reward-based contingencies in a Dot Probe task to improve the training's efficacy, combined with neurophysiological measures for greater reliability compared to reaction times, while evaluating the moderating effect of explicitness in the instruction. In a healthy sample (N = 60) and within a single session, we found a general attentional bias toward angry faces present across all conditions as indicated by the N2pc, which was, however, marked by large intrinsic lateralization effects, with submeasures exhibiting opposing polarities. This prompted us to explore an alternative, intrahemispheric calculation method. The new N2pc variant showed the attentional bias to have disappeared at the end of the training session within the explicit instruction group. Reliabilities of the main dependent variables were varied from excellent to questionable, which, together with the exploratory nature of the analysis, leaves this result as preliminary.