Development of the Facial Emotion Recognition Task (TAREFE): methodological aspects and initial psychometric indicators

Abstract Background Facial Emotion Recognition (FER) is a core component of social cognition, supporting theory of mind processes, empathy, and social reasoning. Deficits in FER are associated with functional impairments across multiple domains of typical and atypical development. Measurement instruments with sound psychometric properties are therefore essential to support clinical assessment in different contexts. Objective To present the development process of the “Facial Emotion Recognition Task” (TAREFE), a computer-based test designed to assess FER in Brazilian adults, with emphasis on the methodological procedures and the initial psychometric indicators. Methods A set of stimuli depicting Brazilian faces was used (n = 170, with diversity in sex and racial background). The stimuli represented the six basic emotions and a neutral expression and had been previously validated in a general population sample. Stimulus selection was based on Item Response Theory (1PL and 2PL), with analysis of infit and outfit indices, S-X 2 analysis, item difficulty (b) and discrimination (a). Results Twenty-eight stimuli were selected, showing excellent discrimination (46% moderate; 54% high or very high) and varied levels of difficulty (43% easy or very easy; 39% moderate; 18% difficult). None of the selected stimuli showed a misfit to the model. In TAREFE, participants are required to identify the emotion expressed by each face by choosing one of the seven response alternatives (Stage 1). Accuracy rates and response times are automatically recorded. To enhance task discriminatory power, an additional optional stage (Stage 2) was introduced, in which stimuli are presented at lower intensity levels (50% and 70% of the target emotion, generated using morphing techniques) to increase recognition difficulty.