Distinct contributions of foveal and extrafoveal visual information to emotion judgments and gaze behavior for faces.

The precise contributions of foveal and extrafoveal visual processing to facial emotion recognition and to how individuals gaze at faces remain poorly understood. We used gaze-contingent masking and windowing to control foveal and extrafoveal inputs while observers (N = 35) classified the emotion (anger, disgust, fear, surprise, sadness) on face images. Emotion classification performance was substantially reduced by the absence of extrafoveal information but was unaffected by the absence of foveal information. Gaze decoding showed that fixation patterns discriminated viewed emotion categories regardless of whether either foveal or extrafoveal information was absent or both were present, more so when observers provided correct responses. Although fixations clustered around the eyes, nose, and upper mouth, emotion-specific biases in fixation densities aligned with regions previously identified as emotion diagnostic, and, for trials with incorrect responses, with locations informative of the most confused emotion. Even without extrafoveal information, necessitating top-down guidance of gaze, fixations were biased to these same emotion-informative regions. Yet, the spatiotemporal sequencing of fixations differed in the absence versus presence of extrafoveal information. Fixation patterns also predicted stimulus presentation conditions, most evident in differences due to the absence versus presence of extrafoveal rather than foveal inputs. Thus, where one looks on a face impacts the ability to determine its emotional expression, not only via the higher resolving power of foveal vision but also by the extrafoveal extraction of task-relevant information and guidance of gaze, and possibly also via the interplay between foveal and extrafoveal vision that underpins presaccadic attention.