Gernot Horstmann

The perception of direct gaze provides multiple benefits for the observer. Previous studies have investigated how the information from both eyes is used to estimate gaze direction, showing that the perception of gaze direction differs when only the externally rotated eye versus only the internally rotated eye is visible. We examined the width and center of the area of direct gaze by presenting the observers with either the externally or internally rotated eye, or both eyes with the task to judge whether a computer avatar is looking at them. Two experiments yield evidence for a wider area of direct gaze for the externally rotated eye (around 6°) than for the internally rotated eye (around 4°). The area of direct gaze for both eyes was found to be the same as for the internally rotated eye, but smaller than for the externally rotated eye. When both eyes were present, our results indicate that the perception of direct gaze is more likely to follow the internally rotated eye. The discussion substantiates a new generalization that observers base their judgments on the more rotated eye, which can differ due to angle kappa and vergence, in our study it is the internally rotated eye.

The cone of gaze is a looker's range of gaze directions that is accepted as direct by an observer. The present research asks how the condition of mild strabismus, that is, when the two eyes point in slightly different directions, influences the cone of gaze. Normally, both eyes are rotated in a coordinated manner such that both eyes are directed to the same fixation point. With strabismus, there are two fixation points, and, therefore, two directions into which the two eyes point. This raises the question of the direction and the shape (i.e., width) of the gaze cone. Two experiments are conducted with simulated mild strabismus. Three conditions are tested, the two strabismic conditions of esotropia, and exotropia and one orthotropic (nonstrabismic) condition. Results show that the direction of the gaze cone is roughly the average of the directions of the two eyes. Furthermore, the width of the gaze cone is not affected by simulated strabismus and is thus the same for the strabismic and the orthotropic conditions. The results imply a model where at first the direction of gaze based on both eyes is perceived, and where the gaze cone is implied on the basis of the combined gaze direction.