Differences in eye-movement patterns between anorexic and control observers when judging body size and attractiveness

Abstract

Attentional biases may influence the eye-movements made when judging bodies and so alter the visual information sampled when making a judgment. This may lead to an overestimation of body size. We measured the eye-movements made by 16 anorexic observers and 16 age-matched controls when judging body size and attractiveness. We combined behavioural data with a novel eye-movement analysis technique that allowed us to apply spatial statistical techniques to make fine spatial discriminations in the pattern of eye-movements between our observer groups. Our behavioural results show that anorexic observers overestimate body size relative to controls and find bodies with lower body mass indexes more attractive. For both judgments, the controls' fixations centre on the stomach, but the anorexic observers show a much wider fixation pattern extending to encompass additional features such as the prominence of the hip and collar bones. This additional visual information may serve to alter their behavioural judgments towards an overestimation of body size and shift their ideal body size towards a significantly lower value.

Fixation patterns, not clinical diagnosis, predict body size over‐estimation in eating disordered women and healthy controls

“…However, the comparison between rAN and CN(OVER) was not ( t  = 0.94, p  > 0.05). These results suggest that accurate body size estimation took approximately 1 to 2 fixations less than inaccurate estimation, and critically, that participants appeared to make their decisions after ∼2 to 2.5 s. Consistent with George et al,13 this therefore justified restricting the eye movement analysis to the first 2,000 ms following stimulus onset.…”

“…In earlier work13 using this analysis method, we showed that the distribution of fixation durations in each cell of the sampling grid is rarely normal, but that the correlation between fixation duration and fixation count is very high (typically r  > 0.9). Therefore, as before, we chose fixation counts per cell (also known as fixation density) as our outcome variable.…”

“…This pattern of fixations differs from the results reported here. This could be because one of these studies focused on the perception of attractiveness rather than adiposity,16 and there are differences in the pattern of fixations for these two judgements 13, 20. However, in a previous study we found that when recording the eye‐movements of women with AN making attractiveness judgements, their more distributed pattern of fixations along the torso meant that they looked more at the upper torso than controls 13.…”

“…In order to examine the spatial distributions of fixations on the reference body, and to compare fixation patterns across observer groups, we constructed a sampling grid of square cells, 20 × 20 pixels each, and applied it across the entire reference image (1,024 × 574 pixels) (for further details see Ref. 13). This cell size (20 × 20 pixels) represents a compromise between capturing as many fixation samples per cell as possible to optimize statistical power (which ideally requires large cells) versus retaining good anatomical resolution (which ideally requires small cells).…”

“…However, although they report differences in the fixation patterns between the women with AN and controls, there is only limited consensus on which areas are favoured or avoided 13, 14, 15, 16, 17, 18, 19. This may be due to methodological differences.…”