Abstract
Individuals show strong and reliable differences in their eye movements towards static (e.g. Bargary et al., 2017, de Haas et al., 2019) and dynamic (Constantino et al., 2017) scenes. Here, we find that this also applies to observers in an eyetracking experiment (n = 16) watching a feature film (‘Shaun the Sheep’), who varied up to factor 2 regarding saccadic rates and amplitudes. What are the consequences of such differences? Does individual gaze lead to individually different ventral representations of the same external events? Or does individual gaze aid inter-individually more similar representations by compensating for differences at earlier stages of processing (such as individual visual field geometry; Moutsiana et al., 2016)? In order to juxtapose these hypotheses, we used functional MRI in a subset of participants (current n = 4; data collection ongoing). Participants watched segments of the movie as in the eyetracking experiment in the scanner, either freely moving their eyes or fixating centrally. We then used a customized version of hyperalignment (Haxby et al., 2011), in order to test how predictive inferior temporal representations of one observer were for those of another. This was done separately for each pair of observers and condition. Pilot MRI results (n = 4 participants) show a trend for better cross-person decoding in the central fixation condition (t = 2.49, p = .054). This would be in line with individual gaze leading to distinct representations of the visual world. We plan to test this hypothesis in a full sample and build a computational model to probe the role of several facets of gaze for representational similarity.