Abstract
Purpose :
Central Vision Loss (CVL) is prevalent in several eye diseases. This impacts many aspects of everyday living including facial expression perception, however, this has not been thoroughly quantified in this population. In this study, participants created subjective representations of facial expressions using a generative face model and genetic algorithms.
Methods :
8 patients with CVL (≥1.3logMAR in better eye) and 8 normally-sighted controls selected computer-generated faces that manifest each of 13 primary affects (Amusement, Anger, Awe, Contempt, Disgust, Embarrassment, Fear, Happiness, Interest, Pride, Sadness, Shame, and Surprise). For each affect, 6 generations of 12 faces were generated from the Basel Face database, in which 199 parameters modulate the structure of each face. After the 1st generation, 6 faces were created by combinations of the parameters of faces selected in the previous generation. The remainder were random to introduce genetic variability.
Results :
A pilot study showed that subjects were most sensitive to changes in the first 20 face parameters, so the analysis was restricted to these. A 2-way ANOVA for the 20 mean coefficients values were compared between the patients and controls. There was a significant (P<.05) main effect for Amusement, Anger, Awe, Disgust, Fear, Interest, and Surprise and a significant (P<.05) interaction for all affects. Overall, coefficient values were closer to 0 for the CVL group, indicating that less extreme face expressions were selected. However, there was no significant difference between groups in the total number of faces selected for each affect besides Fear. This suggests that the tendency towards more neutral faces in CVL patients was not simply a consequence of low sensitivity to affect expression, additionally demonstrating that neither group has a positivity or negativity bias for affective stimuli.
Conclusions :
These data suggest that CVL patients have significantly different perceptions of facial expressions. CVL patients have a tendency to choose more neutral-looking faces to represent a given affect, whereas the control group chose more extreme, caricatured faces. Elucidating individual differences in emotion representation and recognition has implications for populations including neurotypical, visually impaired, and neurodivergent people. This methodology may be a novel tool for quantifying visual function outcomes, and the efficacy of future patient therapies.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.