Using item response theory to select emotional pictures for psychophysiological experiments.

Standardized picture databases such as the International Affective Picture System (IAPS; Lang et al., 2008) and Emotional Picture Set (EmoPicS; Wessa et al., 2010) facilitate the study of emotional scene perception, and have the potential to increase replicability and comparability of results within and across labs. However, with the availability of large numbers of pictures comes the challenge of selecting subsets of pictures for inclusion in experimental paradigms. Typically, researchers rely on expert consensus or normed ratings to select emotional pictures, but these methods may favor pictures with high agreement over those that provide the most information or best differentiate individuals. Here, in n = 297 individuals, we demonstrate how item response theory (IRT), which provides information on psychometric functioning at both the item and test level, can be used to select negative and positive pictures for eliciting the late positive potential (LPP), a measure of emotional attention. We present results for 50 negative and 50 positive pictures, and show how pictures with higher discrimination values improve differentiation between individuals with different levels of emotional attention. Moreover, "strong" modulators of the LPP - i.e., erotic and mutilation pictures - provided the most information about individuals with low levels of emotional attention, whereas, "weak" modulators of the LPP - i.e., affiliative and exciting pictures - provided the most information about individuals with high levels of emotional attention. Results demonstrate how IRT can inform emotional picture selection and improve the psychometrics of psychophysiological tasks, which can ultimately increase the replicability of findings based on standardized pictures.