Categorical Perception of Facial Expressions Is Not a Homogeneous Effect

Categorical Perception of Facial Expressions Is Not a Homogeneous Effect Olga A. Korolkova (olga.a.kor@gmail.com) Moscow State University of Psychology and Education, 29 Sretenka Street, Moscow, 127051 Russia Moscow Institute of Psychoanalysis, 34 Kutuzovsky Avenue, Moscow, 119334 Russia 1992), and later also on gray-scale images (Calder et al., 1996; de Gelder et al., 1997; Young et al., 1997). Its specific aspects were revealed in children (Cheal, Rutherford, 2011; Kotsoni et al., 2001), in patients with amygdala damage (Adolphs et al., 1999), schizophrenia (Kee et al., 2006) and autism spectrum disorders (Teunisse, de Gelder, 2001). Although, several studies did not show the CP (Fiorentini, Viviani, 2009; Schiano et al., 2004) or reported the factors reducing this effect (namely, verbal interference: Roberson, Davidoff, 2000; Roberson et al., 2007). The concept of CP implies that there is a number of discrete emotional categories with qualitative differences between each other, and that each entity of facial emotion can be placed by an observer into one of these alternative emotional categories. Another way of describing emotion and facial expression are continuous spatial models, where the perceived differences between faces are measured as distances in a face-space (Valentine, 1991), or the diversity of emotions is located in a multi-dimensional space with the axes corresponding to physiological properties such as arousal, sleepiness or valence (Russell, 1980). A possible solution of the long-standing debate between categorical (discrete) and dimensional (continuous) understanding of emotion and face perception is by incorporating the both notions into consolidated hierarchical model of higher-level categorical system that interacts with low-level perceptual, non-categorical one (e.g., Fujimura et al., 2012; Roberson et al., 2010). In line with the discrete emotions theory, at least seven basic emotions, universal among cultures, were revealed by P. Ekman (Ekman, Friesen, 1976), and 21 emotional continua, representing transitions from one basic emotion to another, can be constructed between each pair of emotions to test the categorical perception hypothesis. Although, the majority of the facial expressions CP studies explored the perception and differentiation of face images belonging to only a few of these emotional continua. In particular, CP has been shown on continua between happiness and six other emotional expressions, but results on CP of happiness / surprise and happiness / sadness are controversial (see Calder et al., 1996; de Gelder et al., 1997; Etcoff, Magee, 1992). The continua most frequently used in studies are those between anger, sadness, fear and happiness. Though, even on these expressions, the results are ambiguous and do not always show the CP (Calder et al., 1996; de Gelder et al., 1997; Fiorentini, Viviani, 2009). Other continua were rarely included in the stimuli for discrimination task, and for 8 out of 21 continua, no experimental data is available. An extensive study of facial expressions CP was conducted by Young et al. (1997) and included all possible Abstract We studied the categorical perception on transitions between seven basic emotional facial expressions and explored the influencing factors. In Experiment 1, participants performed a multiple-choice emotion labeling task while observing basic or morphed (blended between a pair of basic emotions) facial expressions. In Experiment 2, other participants completed AB-X discrimination task. They observed pairs of images adjacent in a morphing continuum, and matched the test image to one of the pair. The results of Experiment 1 revealed influence of emotional context, formed by the presented expressions, on perception of surprise, anger, disgust, and neutral face. The “categorical field” of morphed expressions includes not only the two relevant emotions (morphing basis) but a number of additional ones. Based on the data of Experiment 1, we selected the pairs of stimuli crossing the categorical boundary, and pairs falling within the category, to predict the discriminability obtained it the Experiment 2. A generalized linear mixed model was fitted to the data. We show the main effect of within/between category pair, type of continuum and continuum/category interaction on the probability of correct discrimination. Overall, our results showed the categorical perception, but its strength depends on particular pair of emotional categories. Keywords: face, facial expression, emotions, categorical perception, generalized linear model. Introduction When we want to test whether the perception of a particular class of objects is categorical, we generally compare the ability to perceive physical differences in pairs of objects belonging to the same and to different categories. Assuming that these differences are equal, the enhanced discriminability in cross-boundary pairs compared to within-category pairs reflects the categorical perception (CP). If categories do not influence the discriminability, we refer to continuously perceived differences. The classic experimental paradigm for testing the CP was proposed by Liberman et al. (1957) in the studies on phonemes perception. It included an identification (labeling) task aimed at defining the boundary between categories, and a discrimination task, in which subjects were asked to differentiate objects in both cross-boundary and within- category pairs. Since then, the CP has been extensively studied on a variety of objects, including facial emotional expressions (Calder et al., 1996; Cheal, Rutherford, 2011; Etcoff, Magee, 1992; Fugate, 2013; de Gelder et al., 1997; Herba et al., 2007; McCullough, Emmorey, 2009; Roberson et al., 2007; Roberson, Davidoff, 2000; Suzuki et al., 2005; Teunisse, de Gelder, 2001; Young et al., 1997; etc.). CP has been shown first on faces presented as linear drawings (Etcoff, Magee,