We investigated modulation of gaze behavior of observers viewing complex scenes that included a person. To assess spontaneous orientation-following, and in contrast to earlier studies, we did not make the person salient via instruction or low-level saliency. Still, objects that were referred to by the orientation of the person were visited earlier, more often, and longer than when they were not referred to. Analysis of fixation sequences showed that the number of saccades to the cued and uncued objects differed only for saccades that started from the head region, but not for saccades starting from a control object or from a body region. We therefore argue that viewing a person leads to an increase in spontaneous following of the person's viewing direction even when the person plays no role in scene understanding and is not made prominent.
To have successful social interactions, we must take into account representations of the world of interacting partners, to disambiguate certain meanings. Lacking direct mind-reading abilities, we can only infer these representations from behavioral cues. This can be achieved by gazefollowing (Butterworth, 1991) or head-following (Langton, 2000). The latter makes it possible to infer the locus of attention even at distances at which direct gaze information is not available. Inferring the locus of attention then makes it possible to engage in joint attention (for an overview, see Frischen, Bayliss, & Tipper, 2007). Both gaze direction (Friesen & Kingstone, 1998) and head direction (Langton & Bruce, 1999) have been shown to direct visual attention even when this is uninformative for the task.
In the paradigms of Langton (2000), Friesen and Kingstone (1998), Langton and Bruce (1999), and others, the cuing stimulus was presented in isolation and was therefore quite prominent. It is less clear whether the orienting mechanism can be observed in more complex scenes, when the cuing stimulus does not occupy central areas of the visual field, which might lead to preferential processing per se (see Dukewich, Klein, & Christie, 2008). What is more, these studies did not address whether directed visual spatial attention also modulates eye-movement control, thus leading to a shift in the gaze of the observer. Investigating the effect on overt attention is of particular interest when studying social gaze behavior. During social communication, overt attention can serve as a trigger for further social interaction by establishing a common focus of attention.
The question of whether observed gaze also leads to overt responses was addressed, for example, by Mansfield, Farroni, and Johnson (2003), as well as by Kuhn and Kingstone (2009). Again, however, gaze cuing was made prominent by having only one centrally presented face and one potential target on the screen. Furthermore, in these studies, the gaze cue (and target position) were the only events that changed across trials. Under such conditions, similar effects can be demonstrated with tongue pointing (Downing, Dodds, & Bray, 2004) or symbolic cues (Kuhn & Benson, 2007; Kuhn & Kingstone, 2009). This calls into question whether the demonstrated effect is not a more general one of spatial compatibility. Additionally, instructing participants to produce speeded eye movements to the targets might have increased the tendency for spontaneous gaze-following. To shed light on whether similar overt responses could be elicited when the cuing object is not placed prominently in the screen and participants are not asked to make speeded saccades, the present study investigated gaze behavior of observers when viewing complex natural scenes.
Orientation-Following in the Presence of Persons
In an interesting study, Kuhn, Tatler, and Cole (2009) looked at gaze-following in natural conditions. By manipulating where a magician was looking, the authors showed that observers of magic tricks often directed their gaze toward the same areas as those at which the magician was looking. …