Individuals with Autism Spectrum Disorders attend to the natural world in a different manner to typically developing individuals. There is evidence that they displaying diminished interest in social stimuli (Jones & Klin, 2008; Klin et al. 2002) especially in the first few fixations when viewing complex scenes (Freeth et al., in press). Could the impact of visual saliency (low-level properties such as colour, intensity and orientation) account for any of these differences? Actions are often performed in a routinised, repetitive way by individuals with ASD. Is this style evident in their patterns of eye movements when natural scenes are viewed?
- To discover whether individuals with ASD are influenced by visual saliency in a similar or different manner to matched controls when viewing naturalistic scenes.
- To discover whether eye movement scanpaths are predictable over time and whether individuals with ASD are drawn to similar or disparate aspects of natural scenes.
A series of natural scenes were presented in a free viewing task at Time 1 and Time 2 (approximately 6months apart) by 13 high functioning adolescents with ASD and 13 typically developing adolescents, matched on age and ability. The influence of visual saliency was investigated using the Saliency Toolbox (Walther & Koch, 2006). Scanpath similarity was analysed using Mannan similarity scores and string edit similarity.
Preliminary findings suggest that visual saliency is more predictive of fixation location early in viewing than later in viewing in individuals with ASD, a finding that has previously been demonstrated in typically developing individuals, and replicated in this study. Visual saliency appears to influence eye movements to a similar extent in high-functioning adolescents with ASD and their typically developing peers. This was found when comparing fixations on the five most salient regions and the mean saliency at fixation. Scanpath analyses suggest that image properties predict fixation sequences in both typically developing individuals and those with ASD when comparing scanpaths at Time 1 and Time 2. However, within group similarity was found to be lower for individuals with ASD suggesting that individuals in the ASD group were more different from each other than individuals in the typically developing group.
Initial results suggest that visual saliency does not influence the attention of individuals with ASD in a markedly abnormal manner. It is therefore unlikely that attention is being strongly captured by low-level properties of stimuli.
Viewing the same image twice resulted in similarly predictable scanpaths in the typically developing group and the ASD group demonstrating that it may be possible to characterise an individual’s attention profile over time but that eye movements are not abnormally predictable in ASD. Comparison of scanpaths between individuals in the ASD group demonstrated greater within group heterogeneity than between individuals in the typically developing group. This finding highlights the importance of looking into individual differences in attention in ASD and suggests that individuals with ASD, as a group, may be more disparate than typically developing individuals.