Abstract
It is widely accepted that emotion processing is impaired in schizophrenic patients (SPs). It is also believed —although evidence to support this claim remain scarce— that SPs confuse (i.e. mis-categorize) some emotions more than healthy individuals. While previous work using the Bubbles technique (e.g. Lee et al., 2011) has revealed aberrant use of facial information that agrees with emotion processing deficit in SPs, their use of only two facial expressions made it difficult to study the sources of the confusions. Here, we examined this question using Bubbles in a four-facial-expression identification task (happy, fearful, angry, or neutral). We first mapped which parts of the face at different spatial scales were used by SPs, and confirmed and extended previous findings. Second, we computed a confusion matrix over the ~13,000 trials completed by all SPs (N=13). Four mis-categorizations had a proportion of responses significantly higher than what is expected by chance (all chi2>60, Bonferroni-corrected ps< .001), including angry faces confused for neutral faces and fearful faces confused for angry faces. Finally, we revealed the specific facial information that drove SPs to commit these two confusions. We discovered that low-spatial frequency (LSF, i.e. coarse information < 10 cycles per faces) from the mouth area and nose area (i.e. philtrum, nose and nasolabial folds), respectively, led to angry faces being mistaken for neutral faces and to fearful faces being mistaken for angry faces. Previous studies revealed magnocellular system abnormalities in the form of below average LSF sensitivity in SPs (Butler et al., 2009). Our results indicate that the inability to extract useful coarse spatial information from expressive faces underlies the mis-labeling of perceived emotions in SPs.
Meeting abstract presented at VSS 2018