Browsing by Author "Plysjuk, Nadja"

Abstract Objectives of the study. Humans are adapted to recognize faces. Turning faces upside down causes an inversion effect that makes it much harder to recognize them. According to previous research, conflicting results have been obtained regarding which facial features are used the most in face recognition and short-term memory, and how turning the face upside down affects the use of features. It has also been unclear how turning the face upside down affects the use of these features. The aim of this thesis was to investigate, using a perceptual psychology experiment and a visual working memory task, whether different facial features are used to different extents in recognition and remembering and whether the facial features used differ in terms of whether the face is presented upright or upside down. Methods. The thesis used a new type of change detection test based on morphing. The material of the thesis consisted of ten students at the University of Helsinki. 40 neutral-looking faces from the Chicago Face Database served as face stimuli. Female and male face pairs were formed from the faces, whose facial features were morphed independently and randomly from 0–20% or 0–80% from the first identity to the second identity. The subject's task was to answer whether faces separated during the memory period, presented upright or upside down, were of the same or different identity. By comparing the number of morphs and the test subjects' answers with each other using a generalized linear model, weighting coefficients were obtained for each facial feature, which told how much the test subjects had used each feature in the test task. The weighting coefficients were normalized, after which they were subjected to a two-way analysis of variance of repeated measures, which was used to examine whether the use of features differed between the upright and upside down situations. In addition, possible differences in face recognition accuracy between upright and upside down situations were investigated using the discriminability index d'. Results and conclusions. According to the results, facial features differed in terms of how much they were used in the working memory task. However, the relative importance of different facial features was similar regardless of whether the face was presented upright or upside down. In line with previous findings, this study also found that the eyes are particularly significant in terms of face recognition and short-term memory. However, in the case of faces presented upside down, the identity recognition accuracy was weaker, which was reflected in a higher number of errors in the test task. In conclusion, turning faces upside down does not affect which facial features are stored in working memory. However, turning faces upside down weakens face recognition and short-term memory.