Abstract
A recent eye tracking study by Leek et al. (in press, Journal of Vision) has provided evidence that the regions of an object that participants fixate during an object recognition task is driven by underlying shape properties and is remarkably consistent across viewpoint. While eye tracking is a powerful tool to determine the areas participants tend to view, it does not provide direct evidence of the relative importance of the areas that are fixated during the viewing of an object compared with those areas that are not fixated. Here we present a new gaze contingent object recognition paradigm that is designed to investigate the relative importance of those areas that are fixated in object recognition tasks. In this task, participant’s eye movements were recorded as they viewed a series of novel 3D objects that they were informed they would later be asked to recognise. The eye movement data from this viewing phase was analysed and the regions that the participants fixated for each object was determined. The regions the participant viewed during this phase were considered ‘critical’ regions. The participants then performed a gaze contingent object recognition task where the participants received a brief exposure of either a previously seen ‘target’, or a previously unseen ‘non-target’ object. Critically, during the recognition task, the target stimuli were presented at fixation in such a way that the participant’s gaze fell upon either (a) a previously fixated ‘critical’ region, or (b) a ‘non-critical’ region which could be anywhere in the object, but not within the area of a critical region. The results show that participant’s performance is superior when they are shown a critical region compared with a non-critical region providing converging evidence of the selectivity of fixated regions in object recognition.
Meeting abstract presented at VSS 2012