Accuracy of Head Orientation Perception in Triadic Situations: Experiment in a Virtual Environment

Perceiving where another person is looking: the integration of head and body information in estimating another person's gaze

The process through which an observer allocates his/her attention based on the attention of another person is known as joint attention. To be able to do this, the observer effectively has to compute where the other person is looking. It has been shown that observers integrate information from the head and the eyes to determine the gaze of another person. Most studies have documented that observers show a bias called the overshoot effect when eyes and head are misaligned. That is, when the head is not oriented straight to the observer, perceived gaze direction is sometimes shifted in the direction opposite to the head turn. The present study addresses whether body information is also used as a cue to compute perceived gaze direction. In Experiment 1, we observed a similar overshoot effect in both behavioral and saccadic responses when manipulating body orientation. In Experiment 2, we explored whether the overshoot effect was due to observers assuming that the eyes are oriented further than the head when head and body orientation are misaligned. We removed horizontal eye information by presenting the stimulus from a side view. Head orientation was now manipulated in a vertical direction and the overshoot effect was replicated. In summary, this study shows that body orientation is indeed used as a cue to determine where another person is looking.

The effect of face eccentricity on the perception of gaze direction

The perception of a looker's gaze direction depends not only on iris eccentricity (the position of the looker's irises within the sclera) but also on the orientation of the lookers' head. One among several potential cues of head orientation is face eccentricity, the position of the inner features of the face (eyes, nose, mouth) within the head contour, as viewed by the observer. For natural faces this cue is confounded with many other head-orientation cues, but in schematic faces it can be studied in isolation. Salient novel illustrations of the effectiveness of face eccentricity are 'Necker faces', which involve equal iris eccentricities but multiple perceived gaze directions. In four experiments, iris and face eccentricity in schematic faces were manipulated, revealing strong and consistent effects of face eccentricity on perceived gaze direction, with different types of tasks. An additional experiment confirmed the 'Mona Lisa' effect with this type of stimuli. Face eccentricity most likely acted as a simple but robust cue of head turn. A simple computational account of combined effects of cues of eye and head turn on perceived gaze direction is presented, including a formal condition for the perception of direct gaze. An account of the 'Mona Lisa' effect is presented.

Psychophysics of perceiving eye-gaze and head direction with peripheral vision: implications for the dynamics of eye-gaze behavior

Two psychophysical experiments are reported, one dealing with the visual perception of the head orientation of another person (the 'looker') and the other dealing with the perception of the looker's direction of eye gaze. The participant viewed the looker with different retinal eccentricities, ranging from foveal to far-peripheral viewing. On average, judgments of head orientation were reliable even out to the extremes of peripheral vision (90 degrees eccentricity), with better performance at the extremes when the participant was able to view the looker changing head orientation from one trial to the next. In sharp contrast, judgments of eye-gaze direction were reliable only out to 4 degrees eccentricity, signifying that the eye-gaze social signal is available to people only when they fixate near the looker's eyes. While not unexpected, this vast difference in availability of information about head direction and eye direction, both of which can serve as indicators of the looker's focus of attention, is important for understanding the dynamics of eye-gaze behavior.