The paper read as if one group of researches conducted the study, and from that data delivered a lot of pure algebra to another set of researchers. The 2nd group of researchers then had to infer what the algebra said about the study. It was very hard to read. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2516255/ This paper was NOT about the accuracy of inference of traits, but rather how people inferred traits from looking at emotionally neutral faces. No part of the study examined the accuracy of these judgements. After numerous studies the researchers came to the conclusion that judgement of a face uses an overgeneralization of the same mechanisms used to read emotion on faces. Essentially, some people look angry even when showing an emotionally neutral face, so the participants in the study would describe them as being more dominant, or more angry.