Abstract
Do humans follow gaze to locations that the gazer cannot see? Previous work using the gaze-cueing paradigm has found that obstacles blocking the gazer’s line of sight did not decrease the gaze-cueing effect at 400ms SOA (Cole et al., 2015). This suggests that the early gaze-following responses are insensitive to context. Here we hypothesize however that a simple lack of salience in the crucial manipulation of line of sight may be responsible for the lack of an effect. Experiment 1 (N = 38) conceptually replicated the previous study by Cole and colleagues (ibid.). Here, barriers were positioned such that in each trial, the agent could see either both the possible target locations, or none. In line with previous findings, the cue validity * sightline interaction at 400ms SOA was not significant (F = .07, p = .79), and the difference in the gaze cueing effect was small: 60ms vs. 56ms in the seeing vs. non-seeing conditions. In Experiment 2 (N = 58) the barriers were positioned such that they blocked the view of either one target location or the other, thus making salient in each trial the difference between where the agent could see vs. not. In contrast to the first experiment, the size of the gaze cueing effect differed significantly at 400ms SOA depending on whether the agent could see the target (cue validity * sightline interaction, F = 6.42, p = .014). The seeing condition yielded a gaze cueing effect of 77ms [95% CI 57ms, 97ms], whereas the non-seeing condition yielded a much smaller gaze cueing effect of 46ms [95% CI 26ms, 65ms]. We conclude then that the agent’s line of sight affects the magnitude of gaze cueing effects, but this only holds when the manipulation of this factor is made sufficiently salient within individual trials.