When asked to judge or react to a facial emotional display of a person, people do not only take the emotion into account, but also other socially important features of the face, such as, for example, ethnicity (Kozlik & Fischer, 2020; Paulus & Wentura, 2014). Importantly, the emotion-related and nonemotion-related features are seemingly not (or not always) processed in a simple, additive manner, but are—in a more functional manner—integrated to provide an “amalgamated signal” on which individuals base their judgment and responses. Whereas Paulus and Wentura (2014) put forward a social-message account of this amalgamated signal, Kozlik and Fischer (2020) recently proposed a processing-conflict explanation. The empirical evidence regarding this issue is, however, mixed. In three experiments, we aimed at replicating and extending Kozlik and Fischer’s central experiment to gain further insight into the validity of the social-message versus the processing-conflict account. However, we failed to replicate their findings. The implications of the new evidence for the two accounts are discussed.