Abstract
Where, when and how eye movements are performed to process the visual environment has been a question of great interest for scientists from a long time. Nowadays, eye-tracking technologies have become precise and affordable, generating an increasing amount of research studies. Even if the use of real-world videos in eye-tracking experiments was recommended from a decade (Henderson, 2003), most of the eye-tracking studies in the literature keep using impoverished representations of the visual environment in laboratory experiments. This experimental choice subtends the idea that eye movements obtained in the laboratory are comparable and representative of the ecological visual strategies deployed in the natural environment. Only a few studies have investigated eye-movements differences between laboratory and real-world conditions. However, these studies compared eye movements from the same observers confronted with different settings and tasks. Consequently, it remains unclear whether eye-movements deployed by active walkers freely navigating in the environment are similar to those of observers visualizing the same environment, when both are performing the same task. To tackle this issue, we asked participants to walk and actively look at the environment while wearing eye-tracking glasses (the Walkers). In addition, we tracked the eye movements of another group of observers (the Watchers), while they were actively exploring in the laboratory the videos (with sound) recorded from the Walkers. Walkers and watchers exhibited different fixation and saccade patterns, but similar fixation durations. Moreover, the Watchers exhibited greater sensitivity to visual saliency and motion than the Walkers, except when persons, actionable or readable objects were present in the scenery. Altogether, our data show that results obtained in laboratory do not entirely generalize to real-world vision, at least for ecological spatial navigation. Eye-movements findings obtained in laboratory studies should be interpreted with caution and, when possible, reproduced in ecological settings for validation.
Meeting abstract presented at VSS 2018