Most studies examine holistic processing with respect to facial identity, but at least one study has also looked at holistic processing of facial expressions (Calder, Young, Keane, & Dean, 2000). However, this work used the partial composite paradigm, which is known to exhibit bias effects (Richler, Cheung, & Gauthier, 2011). The complete composite paradigm (Gauthier & Bukach, 2007) provides a bias-proof way to quantitatively measure holistic processing. In this paper, we perform the corresponding experiment in our face processing model (EMPATH, (Dailey, Cottrell, Padgett, & Adolphs, 2002)) to predict whether holistic processing of facial expressions will be found if the corresponding human experiments are performed, and our prediction is that it will. We also compared our model’s performance to the participants in recent experiments in facial expression recognition in humans (Tanaka, Kaiser, Butler, & Le Grand, 2012). Tanaka et al. (2012) concluded that expression recognition is not always holistic, but our results suggest that it is.