face 2
Photos from the Chicago Face Database used in PSA 001.


Over the last ten years, Oosterhof and Todorov’s valence-dominance model has emerged as the most prominent account of how people evaluate faces on social dimensions. In this model, two dimensions (valence and dominance) underpin social judgments of faces. To which world regions this model applies is a critical, yet unanswered, question. We will address this question by replicating Oosterhof and Todorov’s methodology across multiple world regions. When we used Oosterhof’s and Todorov’s original analysis strategy, the valence-dominance model generalized across regions. When we used an alternative strategy that allowed for a more optimal number of correlated latent factors, we observed much less generalization. These results underscore how each analysis strategy embeds substantive assumptions that can strongly influence theoretical conclusions.


Data collection is finished! We collected over 11,481 participants, in 41 countries, and 11 world regions. In October 2019, we released the data set for a Secondary Analysis Challenge and we had 8 submissionsThe paper was accepted at Nature Human Behaviour and was released on January 4, 2020! This marks to true completion of our first ever study at the PSA.