Looking beyond the average: Human gaze behaviour towards natural scenes in special populations and across the lifespan

Abstract

Every day, we encounter an overwhelming amount of visual information. However, our visual system is capable of processing only the central area of the visual field with high acuity. This requires us to constantly move our eyes in order to select regions of interest. Which factors influence where we look? For more than a century, empirical research has investigated this question from various perspectives. The key findings indicate that gaze behaviour is mostly directed by (high-level) features of a scene, our goals, and visuo-spatial biases. A common aspect across most of these studies is that they do not explore potential differences among observers. However, recent findings suggest that gaze is highly individual and different among (age) groups. This thesis focuses on the observer, examining how free viewing behaviour towards natural scenes differs between observers, and investigates the effects of face recognition ability and age on these differences. To this end, the first study found that free viewing behaviour towards natural scenes is atypical in Super Recognizers (SRs) – individuals with superior face recognition skills. Compared to controls, SRs showed a stronger tendency to focus on faces but looked less towards touched objects and text. When looking at faces, SRs focused less on the mouth but more on a point just below the eyes, which is considered optimal for face identification. Together, these findings suggest that face recognition ability is related to a visual preference for faces and saccadic landing positions close to the theoretical optimum within them. The second study first replicated large and stable individual differences in gaze behaviour towards objects of multiple semantic categories in an ad-hoc sample of adult observers. It then validated a short test, showing that such fixation tendencies can be estimated with a subset of only 40 as opposed to 700 natural scenes. In study 3, we examined the development of semantic gaze biases and applied this short test to probe them in preschoolers. Compared to adults, children looked substantially less towards text, but more towards touched objects and faces, suggesting that text preference may be guided by levels of literacy and reading experience. Finally, to track the development of gaze behaviour across the lifespan, study 4 applied the short test in a large and age diverse sample. We found that the semantic gaze tendencies in children demonstrated in study 3 undergo a protracted development taking ~ 20 years to become adult-like. Moreover, we showed that the central and horizontal biases unfold until early adulthood. Interestingly, while visual exploration showed a protracted increase, the fixation patterns became less idiosyncratic and more canonical across adolescence. Together, the presented studies show that averages can obscure systematic variation across observers, when studying gaze behaviour, which may inform us about underlying mechanisms. I conclude that gaze towards high-level object attributes is highly individual across the population and atypical in certain cohorts, covering a spectrum of semantic gaze biases. Based on the findings of study 1, I speculate on a mediating role of face recognition ability in shaping fixation biases towards faces. Studies 3 and 4 suggest that gaze behaviour is highly plastic and perhaps affected by changes in visual diet and literacy that mutually reinforce each other. Interestingly, recent findings on cortical recycling in higher visual cortex nicely converge with the protracted development of viewing behaviour presented here. This suggests a link between the development of gaze behaviour, visual diet and cortical recycling that may be addressed by future research.