Abstract
Perceiving gaze in others is important for everyday life. In order to define the area in which observers perceive a model’s gaze as direct, the gaze cone concept was introduced. The most frequently used methods to measure the gaze cone can be summarized into two broad groups of tasks: First, the Gibson Task, which is a method of constant stimuli. Second, the Gamer Task, which is a method of adjustment. Gaze, however, may also be perceived to be directed at a “third” object, which is measured using a third task, the Anstis Task. Previous studies found evidence for a considerable range of gaze cone widths in the Gibson and the Gamer Task, with some variation left unexplained and they found an overestimation of gaze in the Anstis Task. Our study compares all three methods and aims to answer two questions: First, do the two dyadic tasks of Gamer and Gibson lead to comparable gaze cone width results? Second, does the overestimation bias that can be measured in the Anstis Task, influence judgements in the two dyadic tasks in the same manner? Results indicate a 4° wider cone in the Gamer Task (10.5°) compared to the Gibson Task (6.6°). This difference even increases when the overestimation is taken into account. Uncorrected cone width measures for the two tasks do not correlate with each other. Interestingly, after correcting both cone widths by the estimation biases, a substantial correlation between Gibson and Gamer Task was obtained.