Systematic tendencies in scene viewing

Abstract

While many current models of scene perception debate the relative roles of low- and highlevel factors in eye guidance, systematic tendencies in how the eyes move may be informative. We consider how each saccade and fixation is influenced by that which preceded or followed it, during free inspection of images of natural scenes. We find evidence to suggest periods of localized scanning separated by ‘global’ relocations to new regions of the scene. We also find evidence to support the existence of small amplitude ‘corrective’ saccades in natural image viewing. Our data reveal statistical dependencies between successive eye movements, which may be informative in furthering our understanding of eye guidance.

WALD-EM: Wald accumulation for locations and durations of eye movements.

“…With respect to the central and horizontal biases, there is an ongoing debate on what is exactly the cause of these phenomena. Possible explanations range from being caused entirely by the image content (e.g., objects mostly aligned in the center of the images, or objects mostly aligned along horizontal axes or the horizon), to being some sort of interaction between image content and system-P r e p r i n t atic bias towards centers or horizontal saccades, to being completely explained by systematic tendencies, caused by physiological, learned or strategic aspects (Foulsham et al, 2018(Foulsham et al, , 2013Le Meur & Liu, 2015;Tatler & Vincent, 2008;Tseng et al, 2009;van Renswoude et al, 2016van Renswoude et al, , 2019. It is also possible that these three sources of the observed "biases" are not mutually exclusive.…”

“…The top-down factors and mechanisms include characteristics and states of the observer, such as their motivation, purpose, task, (background) knowledge or individual differences (De Haas, P r e p r i n t Iakovidis, Schwarzkopf, & Gegenfurtner, 2019). The third group includes factors that are neither purely bottom-up (i.e., not necessarily tied to features in the environment) nor top-down (i.e., not necessarily unique to states or characteristics of observers), but rather experimentally observed phenomena (Tatler & Vincent, 2008). Systematic tendencies are believed to be relatively stable across stimuli, participants and tasks, such as fixation biases (e.g., central bias; Tatler, 2007;Tseng, Carmi, Cameron, Munoz, & Itti, 2009; van Renswoude, van den Berg, Raijmakers, & Visser, 2019) or saccadic biases (e.g., horizontal and leftward bias; Foulsham, Frost, & Sage, 2018;Foulsham, Gray, Nasiopoulos, & Kingstone, 2013;Le Meur & Liu, 2015;van Renswoude, Johnson, Raijmakers, & Visser, 2016).…”

Gazing at Social Interactions Between Foraging and Decision Theory

“…This is a difficult hurdle to face. Some contextual rules have been proposed in the computer vision field (Torralba et al, 2006) and in the psychological literature (Tatler and Vincent, 2008). However, these might be put into question out of the lab and in dynamic environments.…”

Similarities and Differences Between Eye and Mouse Dynamics During Web Pages Exploration

“…These studies clearly show an influence of bottom-up and topdown factors. However, Tatler and Vincent (2008) and Anderson et al (2015) show that bottom-up influence was higher at the beginning of visual exploration. Thus, both factors alternatively influence visual exploration (Henderson, 2003;Torralba et al, 2006).…”

“…For instance, we know that a fixation last in average 250-350 ms (Mackworth and Morandi, 1967;Yarbus, 1967) and that visual exploration is modulated by bottom-up and top-down factors regardless of the stimulus type (Yarbus, 1967;DeAngelus and Pelz, 2009;Helo et al, 2014;Itti and Borji, 2015). Bottom-up factors are characterized by low-level features of the stimulus, such as luminance, contrast, or edges (Tatler and Vincent, 2008), while top-down factors are characterized by high-level properties representing cognitive processes (Henderson and Hollingworth, 1999). It is generally assumed that the interaction between bottom-up and top-down factors influence how we orient our visual attention (Theeuwes and Failing, 2020).…”