Understanding Strategic Information Use During Emotional Expression Judgments in Williams Syndrome

ABSTRACT

Detailed analysis of expression judgments in Williams syndrome reveals that successful emotion categorization need not reflect “classic” information processing strategies. These individuals draw upon a distinct set of featural details to identify happy and fearful faces that differ from those used by typically developing comparison groups: children and adults. The diagnostic visual information is also notably less interlinked in Williams syndrome, consistent with reports of diminished processing of configural information during face identity judgments. These results prompt reconsideration of typical models of face expertise by revealing that an age-appropriate profile of expression performance can be achieved via alternative routes.

Acknowledgments

We thank Michael Papasavva, Inês Mares, Hayley White, Cathy Wong, Charles Wigley and Marguerite Cullity for assistance with data collection as well as the participants, their families and the schools for the time and effort expended in these studies.

Funding

This research was supported by a grant from the Leverhulme Trust: RPG-2013-019 awarded to MLS, EF and AKS. Additional support was generously provided by Gillian Rhodes and the Person Perception Node of the Australian Research Centre of Excellence in Cognition and its Disorders CE110001021.

Notes

^1. Note that Smith et al. (2005) found no significant difference in the diagnostic information for expression judgments across male and female faces.

^2. For ease of explanation, we told participants that the bubbles concealed things from view; however, in truth the bubbles more accurately revealed information (like windows through an opaque mask).

^3. Note that one participant from the WS group had a severe and selective impairment for fearful facial expressions only, registering only 12% correct across the course of the experiment in this one condition whilst performance for the remaining three expressions was well above chance (M = 58% correct). Similarly a second WS individual was selectively poor with anger (11% correct). Group average data has been used in place of the individual values for these two cells of the ANOVA.

^4. Pilot studies which used equivalent levels of information sampling in a separate adult control group (N = 14 completed an identical task to the WS group) resulted in a group level significant difference in performance accuracy, F(1,30) = 5.8, p = 0.023). That is, when the number of bubbles was matched between groups, performance accuracy was not. These new typical adults performed extremely well (79% correct on average): outperforming relative to the WS group, as well as the 75% accuracy criterion target because the task was simply too easy for them.

^5. Note that very few participants made use of the “don’t know” option, with no difference between the groups: median use in WS: 2.9% vs. typical adults: 3.7%, U = 148.5 (Z = .43), p = 0.67.

Additional information

Funding

This research was supported by a grant from the Leverhulme Trust: RPG-2013-019 awarded to MLS, EF and AKS. Additional support was generously provided by Gillian Rhodes and the Person Perception Node of the Australian Research Centre of Excellence in Cognition and its Disorders CE110001021.