The effect of social roles on gaze cue utilisation in a real-world collaboration

The effect of social roles on gaze cue utilisation in a real-world collaboration Ross G. Macdonald (rgmacdonald@dundee.ac.uk) School of Psychology, University of Dundee, Nethergate, Dundee, DD1 4HN, UK Benjamin W. Tatler (b.w.tatler@activevisionlab.org) School of Psychology, University of Dundee, Nethergate, Dundee, DD1 4HN, UK could slow down responses in a Posner (1980) task, suggesting that the artificial gaze cue stimuli automatically shifted attention away from the target. Variants of this study looking at eye movements have found that participants will also look in the direction of the distracting gaze cue, even though they know there is no reason to do so (Ricciardelli, Bricolo, Aglioti, & Chelazzi, 2002; Galfano et al, 2012). These findings have been used to suggest that humans are “hard-wired” to automatically follow the gaze cues of others (Emery, 2000). The above research shows that people look at eyes and follow gaze cues when viewing isolated static images of others. However, in the real world, gaze cues usually occur alongside spoken language. There appears to be an intimate link between gaze allocation and spoken language, with people making anticipatory eye movements to objects that relate to what they hear (Altmann & Kamide, 1999). Gaze cue utilisation in particular has been shown to be affected by spoken language; changing the syntactic structure of a sentence, whilst maintaining meaning changes the timing of gaze following (Knoeferle & Kreysa, 2012). Reciprocally, Stuadte and Crocker (2011) showed the gaze cues can affect the understanding of spoken language; participants were shown videos of a robot describing the spatial and featural relations between a series of visible objects, whilst providing gaze cues. The robot made mistakes in his descriptions that could have been corrected in two different ways. The experimenters found that participants would correct in the way that was congruent with the gaze cue, suggesting that they were inferring meaning from the robot’s gaze and assuming that the robot meant to refer to the object that it was gazing at. Given the effect gaze cues and language have on each other, it is important to use language in a paradigm investigating how gaze cues are used naturally in collaboration. As well as mostly occurring alongside language, all gaze cues in the real world are provided in a social context. When interacting with another, where we look can be affected by our proximity to this other person (Argyle & Dean, 1965). Social effects specifically on gaze seeking were investigated by Laidlaw, Foulsham, Kuhn and Kingstone (2011), who found that participants sitting in a waiting room were significantly more likely to look at a person on a monitor than the same person present in the room. Gallup et al (2012) found similar results for gaze following rather than seeking. They observed people walking past an attractive Abstract During collaboration, people communicate using verbal and non-verbal cues, including gaze cues. Social factors can affect gaze allocation, however most research on gaze cueing has not considered these factors. The presence of social roles was manipulated in a collaborative task whilst eye movements were measured. In pairs, participants worked together to make a cake. Half of the pairs were given roles (“Chef” or “Gatherer”) and the other half were not. Across all participants we found, contrary to the results of static image experiments, that participants spent very little time looking at each other, challenging the generalisability of the conclusions from lab-based paradigms. When given spoken instructions, listeners in the roles condition looked at the speaker significantly more than listeners in the no roles condition. We conclude that our tendency to seek the gaze cues of collaborators is affected either by our social perceptions of the collaborator or their perceived reliability. Keywords: eye movements; joint attention; real world; gaze cues; social interaction. Introduction When collaborating with another on a task, we need to communicate. As well as using spoken language, there are a number of non-verbal cues we can use, with the directional gaze cues given by the eyes being the most well-researched of these. Gaze cues are first used very early in life and continue to be given and followed throughout adulthood. People have a tendency to orient to and follow the gaze cues of others and can to do this with ease. However, there is evidence that the language accompanying a gaze cue and the social context of the cue can affect how people orient to and follow gaze cues. In the real world, gaze cues will always occur within a social context, yet this context is removed in most studies. The aim of the present study is to measure eye movements in a real-world setting to observe how the utilisation of gaze cues can be affected by social context in a natural collaboration. When viewing images of faces, people have a tendency to look at the eyes (Yarbus, 1967) and when viewing images of social scenes people will seek out faces and eyes (Birmingham, Bischoff & Kingstone, 2007; 2009) even when the person being fixated is not visually prominent and has no role for understanding the scene (Zwickel & Vo, 2010). As well as orienting to these cues, people show a tendency to follow them. Friesen and Kingstone (1998) showed that incongruent gaze cues presented at fixation