Desenvolvimento de um banco de fotografias e vídeos de expressões emocionais de crianças brasileiras de quatro a seis anos. Facial Affective Brazilian Set - children image and vídeo database (FABS)

• Ano de defesa: 2019
• Autor(a) principal: Negrão, Juliana Gioia
• Orientador(a): Schwartzman, José Salomão
• Banca de defesa: Seabra, Alessandra Gotuzo , Brunoni, Décio , Mecca, Tatiana Pontrelli , Mograbi, Daniel
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Presbiteriana Mackenzie
• Programa de Pós-Graduação: Distúrbios do Desenvolvimento
• Departamento: Centro de Ciências Biológicas e da Saúde (CCBS)
• País: Brasil
• Palavras-chave em Português: expressões faciais; criança; banco de estímulos faciais; emoção
• Área do conhecimento CNPq: CNPQ::CIENCIAS HUMANAS::PSICOLOGIA
• Link de acesso: http://dspace.mackenzie.br/handle/10899/24577

The recognition of facial expressions plays an important role in social interactions, especially during the childhood phase, where oral communications is not yet fully developed. One way to evaluate this skill in children is to use photographs and videos from emotional expressions databases. These materials are important to study the processes of social cognition and can serve as diagnostic tools and clinical intervention. Few studies, however, have focused on building specific photo databases for facial expressions in early childhood. Objective: The present work aims to develop and validate a database of photographs and videos of Brazilian children aged between four and six years, with facial expressions posed and induced for the following emotions: joy, sadness, anger, surprise, disgust, fear and contempt, and lastly the absence of emotion (neutrality). Method: Children were selected from a children's acting agency in Sao Paulo, SP, to participate in the study. A geneticist performed a morphological analysis of the children's photographs and ethnicity was defined only when there was a positive cross with the ethnicity declared by the guardians. In a pilot study, children's photos and videos were selected and tested to elicit emotions. After this stage, the built-in stimuli were applied to children as they were filmed and generated 29 hours of video (Mackenzie TV studios). The first judge (Judge 0) selected photo frames and movie clips that represented the seven emotions and neutrality. These were cut and inserted a web platform a professional specializing in image editing. A second judge (Judge 1) made an analysis of the already fragmented stimuli, and only the photos and videos in which there was 100% agreement between them as to the type of emotion were selected for the construction of the Facial Affective Brazilian Set – Children Image and Video Database (FABS) Database. Then, for validation of the bank, two other judges (Judge 1 and Judge 2) specialists in the Facial Action Coding System (FACS) method analyzed all the photographs and videos that made up the bank. Statistical analysis of agreement between the judges regarding emotions was performed using the Kappa Agreement Index. The accuracy of the judges was compared using the Two Proportion Equality Test. Results: A total of 182 children were selected to participate in the study, of which 31 (21%) were excluded for not achieving ethnic agreement between the geneticist and the child's 8 parents, and 3 (2%) for refusing to participate in the filming. Of the remaining 148, 16 of them participated in the pilot study and 8 children were excluded for lack of agreement between Judges 0 and 1. Thus, 124 children (55% girls) who generated 1985 stimuli participated in the construction of the FABS bank. Of these, 1014 (51%) were photographs, 971 (49%) videos, 904 (46%) were stimuli were posed and 1081 (54%) induced. The agreement analysis between the two judges to validate the bank obtained an overall Kappa index of 0.70 and an accuracy of 73%. There was greater accuracy in the evaluation of children of African descent compared to Caucasian children (76% vs. 72%, p = 0.026), and that the accuracy of judges in the evaluation of 4-year-olds was lower than in those of 6-year-olds (71% vs. 76%, p = 0.46). No significant difference was found in the agreement between judges regarding children's gender or type of stimulus (photo / video). Judges' accuracy was significantly higher when evaluating posed versus induced stimuli (78% vs. 69%, p <0.001) for both photos and videos. The emotions happiness, disgust, contempt and surprise had the highest percentages of agreement among the judges, while neutrality and surprise had the lowest agreement rates. Conclusion: This paper constructed and validated a set of infant facial stimuli that could be applied in future research on emotional recognition in early childhood individuals. In addition, the FBS database allows subsequent studies on the mechanisms involved in the recognition of facial emotions in childhood, and the investigation between facial expressions and psychiatric disorders