Two-element classification images for the discrimination of emotional expression in upright and inverted face

Abstract

Classification images (CI) can reveal an observer's strategy for discriminating faces (e.g., Sekuler, Gaspar, Gold, & Bennett, 2004, Cur Biol). The present study used a variation of the CI method to examine the strategy used to discriminate the emotional expression of schematic faces. The faces consisted of two eyebrows and a mouth — each represented by a curved or straight line — and two eyes represented as small disks. Expression was manipulated by varying the curvature of the eyebrows and mouth. For example, downward curvature of the eyebrows combined with upward curvature of the mouth produced a happy face. The observer's task was to classify a face as ‘Happy’ or ‘Non-Happy’. Feature curvature was varied across trials with a staircase procedure to estimate a discrimination threshold. Curvature noise was added to the mouth and eyebrows on each trial: noise added to the mouth and eyebrows was independent, but the noise values for the two eyebrows on each trial were equal. We collected responses on 400 trials for upright and inverted faces from 12 Japanese healthy observers. Logistic regression showed that the judgments of all observers were influenced significantly by mouth curvature in both conditions. One observer in the upright condition and five in the inverted condition were not influenced by eyebrow curvature. Additionally, the interaction between eyebrow and mouth curvature was not significant in all but two observers in the upright face condition. Hence, judgments from most observers were well fit by an additive linear model that placed greater emphasis on mouth curvature than eyebrow curvature, especially for inverted faces. Currently we are testing Japanese autistic observers, as well as typical and autistic Canadian observers, to measure the effects of autism on emotional judgments and to determine if the results are similar for observers drawn from different cultures.

Footnotes

 Supported by the CIHR-JSPS Japan-Canada Joint Health Research Program, the Canada Research Chair programme, and Grant-in-Aid for Young Scientists (B)