Abstract
Eye movements during free exploration of real-world scene imagery often show a pattern of longer fixation durations over time, but decreasing saccade amplitudes which has been explained as a shift from ambient (global) to focal (local) processing. In the present study, we asked whether these established spatio-temporal eye movement patterns would extend to scenes deviating from our everyday experience, namely aerial scene views. Previous research showed that gist processing takes longer for aerial than terrestrial views, and aerial views were unaffected by scene rotation, unlike terrestrial views (Loschky, Ellis, Sears, Ringer, & Davis, 2010). The latter finding suggested that terrestrial views are processed holistically, with aerial views being processed featurally. Because the focal mode is associated with processing object features, we asked whether aerial views engender more focal processing than terrestrial views. Alternatively, since aerial view gist processing takes longer, perhaps the ambient mode would last longer when viewing aerial than terrestrial scenes. To test these competing hypotheses, we showed 30 viewers both aerial and terrestrial scenes, either upright or inverted, for 6.5 seconds, and analyzed their eye movements. For aerial views, we found prolonged first fixation durations, consistent with aerial gist processing taking longer. Thereafter, we found that the ambient-to-focal strategy was preserved across both perspectives, but for aerial views both, fixation durations and saccade amplitudes were longer. Furthermore, for aerial views, scan paths were less similar between observers, and fixation locations were better predicted by visual saliency, consistent with the idea that processing of aerial scenes requires more "bottom-up" processing than terrestrial scenes. Thus, the results confirm the stability of the ambient-to-focal shift for scenes deviating from our normal experience. However, the greater processing effort might be reflected by a hybrid processing mode that was part ambient (long saccades) and part focal (long fixations).
Meeting abstract presented at VSS 2013