Humans are natural emotion detectors and particularly predisposed to attend to emotions conveyed by facial expressions. The compulsory attraction to emotions, however, gets challenging when multiple emotional stimuli compete for attention, as in the emotion comparison task. In this task, participants are asked to choose which of two simultaneously presented faces displays a specific emotion (happiness/anger). When the two faces both express a different emotion intensity, the selection is driven by a purely perceptual attentional component: participants respond faster to the face displaying the strongest emotion (i.e., the emotional semantic-congruency effect, ESC), and this effect is stronger for face pairs that contain globally positive rather than negative emotional faces (i.e., the size-effect). In this experiment, we exploited the temporal dynamics of the emotion comparison task by tracking participants' eye movements using gaze-contingent displays. We expected that a perceptually driven attentional component like the ESC will rise over time following a purely automatic attentional component based on a left-to-right visual asymmetry. Analysis of fixation accuracy and dwell time fully corroborated this expectation. On the first fixation, participants were more accurate and dwell longer on the left face when it was the target, according to an automatic attentional component. The pattern of accuracy and dwell time on the second fixation was instead consistent with the ESC. Overall, our pattern of results indicates that attentional capture in emotion comparison task arises from the linear combination of purely automatic and perceptual (i.e., the ESC) attentional components over time [This research was supported by a founding for the research (RESRIC-FANTONI2018, Department of Life Sciences, University of Trieste) to CF and an international Fellowship within the European Social Fund 2014-2020 programme of Regione Autonoma Friuli Venezia Giulia to GB.]
