Normal susceptibility to visual illusions in abnormal development: evidence from Williams syndrome

Abstract

The perception of visual illusions is a powerful diagnostic of implicit integration of global information. Many illusions occur when length, size, orientation, or luminance are misjudged because neighboring visuospatial information cannot be ignored. We asked if people with Williams syndrome (WS), a rare genetic disorder that results in severely impaired global visuospatial construction abilities, are also susceptible to the context of visual illusions. Remarkably, we found that illusions influenced WS individuals to the same degree as normal adults, although size discrimination was somewhat impaired in WS. Our results are evidence that illusions are a consequence of the brain's bias to implicitly integrate visual information, even in a population known to have difficulty in explicitly representing spatial relationships among objects. Moreover, these results suggest that implicit and non-implicit integration of spatial information have different vulnerabilities in abnormal development.

Figure 1

Model and copies by WS participants (age in years; months).

Figure 2

Pair of illusion context trials (a–b) and baseline trial (c) in the Ponzo, Müller-Lyer, Kanisza-occlusion and Ebbinghaus illusion. Illusion trials: Though the physical target size difference is identical in both expansive (a) and compressive (b) trials, the size difference between the targets is perceptually increased in the expansive trials (a), while perceptually reduced in the compressive trials (b). In both expansive and compressive trials, the position of the illusion context is constant. For the Ponzo illusion the longer target line is on top for the expansive trials, and on the bottom for the compressive trial. For the Müller-Lyer and Kanisza-occlusion illusions the longer target line is on right for the expansive trials, and on the left for the compressive trials. For the Ebbinghaus illusion the larger target circle is on left for the expansive trial, and on the right for the compressive trial. In our experiment, we noted the difference in size at which the participant made the first error on a compressive trial in the illusion condition, which reflects the physical size difference perceptually nullified by the illusion context. Baseline trials: The longer/larger target is on the bottom, right, left and right for the Ponzo, Müller-Lyer, Kanisza-occlusion and Ebbinghaus illusions, respectively. The physical target size difference is identical across the baseline and illusion trials.

Figure 3

(a) Ponzo, Müller-Lyer, Kanisza-occlusion and Ebbinghaus stimuli. The central lines or circles are physically the same size, but are perceptually different in size. (b) Mean proportion of responses consistent with the illusion when the central lines or circles were physically the same size. Proportions were not reliably different between normal adults (≥18 year olds) and WS. In the Ebbinghaus illusion, mean proportion of 3–4 year olds was not significantly different from chance (0.50), and was significantly different from the other participant groups. (c) The difference in size at which the participant made the first error, the geometric mean of size differences, on baseline trials without the illusion context (open circles) or on critical trials with the illusion context (closed circles). Size differences averaged over context and no context conditions were reliably lower in normal adults than in WS or in 3–4 year olds. (d) Crucially, geometric means of difference ratios (size difference _illusion/size difference _baseline), which represent the illusion magnitude, were not reliably different between normal adults and WS participants. This suggests that WS individuals perceive visual illusions to the same degree as normal adults (*significant post-hoc Tukey HSD between participant groups at p<0.05).

Figure 4

Frequency histograms of illusion magnitude (i.e., log difference ratio) in normal adults and WS individuals for the Ponzo, Muller-Lyer, Kanisza-occlusion and Ebbinghaus illusions. Scores greater than zero represents an effect of the illusion. Across all four illusions, the distributions between normal adults and WS individuals highly overlap, and are not significantly different (see text for details).