From gaze cueing to perspective taking: Revisiting the claim that we automatically compute where or what other people are looking at

Abstract

Two paradigms have shown that people automatically compute what or where another person is looking at. In the visual perspective-taking paradigm, participants judge how many objects they see; whereas, in the gaze cueing paradigm, participants identify a target. Unlike in the former task, in the latter task, the influence of what or where the other person is looking at is only observed when the other person is presented alone before the task-relevant objects. We show that this discrepancy across the two paradigms is not due to differences in visual settings (Experiment 1) or available time to extract the directional information (Experiment 2), but that it is caused by how attention is deployed in response to task instructions (Experiment 3). Thus, the mere presence of another person in the field of view is not sufficient to compute where/what that person is looking at, which qualifies the claimed automaticity of such computations.

Keywords: Gaze; attentional cueing; joint attention; perspective-taking; theory of mind.

Figure 1.

Upper panel: Illustration of the gaze cueing paradigm. A gazing face is first presented alone (SOA ≥ 50 ms) followed by the presentation of a target (often a letter to identify) at a location that is either congruent or incongruent with the gaze cue. Lower panel: Illustration of the level-1 visual perspective-taking paradigm. Two prompts indicate the perspective to take (here “YOU” instructs to take the self-perspective) and the perspective content to verify (here “2” refers to the number of discs) followed by the presentation of a scene in which another person is presented simultaneously (SOA = 0 ms) with the targets; the other person can either see the same or a different number of discs than participants.

Figure 2.

Illustration of the timing of the events on each trial of the modified gaze cueing task of Experiment 1. On this trial, participants had to press on the key “2” because two discs were visible in the room. The last screen displayed the feedback about participant's accuracy of response.

Figure 3.

Gaze congruency indexes across Experiment 1, 2, and 3 in the modified gaze cueing paradigms. SOA = stimulus onset asynchrony, IES = inverse efficiency score. Error bars indicate the 95% confidence intervals. ** p < .01, *** p < .001.

Figure 4.

Illustration of the timing of the events on each trial of the modified gaze cueing task of Experiment 2. On this trial, participants had to judge whether there was one disc with a red border (see magnified view of the discs used) visible in the room. The last screen displayed the feedback about participant's accuracy of response.

Figure 5.

Illustration of the timing of the events on each trial of the modified gaze cueing task of Experiment 3. On this trial, participants had to judge whether there were two red discs visible in the room (see magnified view of the colour and number prompts superimposed on the avatar). The last screen displayed the feedback about participant's accuracy of response.