Abstract
Individuals can extract the summary emotional information from groups of multiple faces, an ability called ensemble emotion perception. However, not every member in the group is weighted equally. For example, it was shown that members presented at (or close to) fixation weighed more in ensemble emotion perception, revealing a ‘foveal input bias’. To test directly whether the foveal input bias is ubiquitous, we measured participants’ capacity to perceive the average emotion of a group of faces while asking them to voluntarily ignore the foveal input. The stimulus consisted of nine faces - a central face at fixation surrounded by 8 flankers. Presentation time was 100ms. The emotion of the central face was either varied (happy and angry randomly interleaved) or kept constant (a single emotion: happy, angry, neutral, scrambled, or absent) throughout a block of trials. In two experiments, participants either judged the average emotion of the face set (Experiment 1) or were asked to ignore the foveal input and judge the average emotion of only the surrounding faces (Experiment 2). The results of Experiment 1 revealed strongly impaired performance with varied foveal input: participants tended to report the foveal input’s emotion rather than the average emotion of the face set, revealing a foveal input bias. With constant foveal input, performance was similar with absent, neutral, and scrambled foveal input, and slightly improved compared to varied foveal input, showing that constant foveal input diminished the foveal input bias. In Experiment 2, where participants were asked to ignore the foveal face, performance was similar in all conditions: there was no foveal input bias. These results revealed that observers can ignore the foveal input, suggesting that foveal input bias is not ubiquitous but can be overcome by voluntary control.