start » program » presentations

Do as eye say: the interaction between gaze cues and language specificity in social interaction

Gaze cues are important in communication and are used in this way even before spoken language has developed. Once language has been mastered, these cues continue to be used in social interactions. Most gaze cues will not be given in isolation; they will be given and followed within the context of a natural dialogue. Despite this, most previous research on gaze cueing has used paradigms without spoken language. As well as focusing on gaze independently of language, previous paradigms have often been highly artificial, using stimuli that are far removed from real-world gaze cues.

The present study uses authentic gaze-cues to investigate the interaction between gaze and language in the real-world by measuring three aspects of gaze utilisation: gaze seeking, gaze following and behavioural benefits. Each participant followed instructions to build a series of abstract structures out of building blocks. Throughout the experiment their eye movements were recorded using a Positive Science LLC mobile eye-tracker, which allowed free head movement. In a 2x2 between-subjects design, the instructor varied the specificity of the instructions given to participants (unambiguous or ambiguous language) as well as the presence of beneficial gaze cues (present or absent). Eye tracking data were coded offline and fixations to the blocks were recorded along with task performance. We measured three dependent variables: (1) percentage of instructions in which the participant looked at the instructor, (2) percentage of first fixations after the first descriptor word on the target block and (3) percentage of correct pick-ups of the target block. When ambiguous instructions were used, participants looked at the instructor more, made more correct first fixations and made more correct pick-ups when gaze cues were given than when they were not. For those given unambiguous instructions, the presence or absence of gaze cues did not significantly affect any of the measures.

Our findings provide new theoretical insights into when and how we use the gaze cues supplied to us by people with whom we interact. These results indicate that rather than being a ubiquitous response to a social interaction, the tendency to engage in gaze seeking and following depends on the informativeness of the gaze cue relative to other available information. In the case of this experiment, when language provides necessary information to locate a block, gaze cues are not sought nor followed. Similarly, if the instructor’s gaze is not informative, these cues will not be sought out, even if the spoken language does not provide sufficient information to locate the block. By innovatively combining language and gaze cues experimentally in an ecologically valid environment, we are able to conclude that people only utilise the gaze cues of others when the cues provide useful information.