Modeling Mental Representations
In the past few years, together with Matthias Keller, I developed and validated a novel method to extract individuals’ mental representations of others and themselves. The paradigm works as follows: Individuals are presented with dozens of pairs of faces that are all based on the same base face and only randomly vary on the dimensions of a multidimensional face space. Participants are repeatedly asked to answer a specific question, for example, which of the two persons better represents a target of social exclusion or a trustworthy person. Participants are asked to intuitively decide and click on the respective portrait. By averaging all the faces selected by participants we extract their mental representation, for instance, of a target of social exclusion or a trustworthy person. The advantages of this novel method compared to the classical reverse correlation method (Figure Part A) all arise from the fact that we model faces instead of face images (Figure Part B). For example, the same random vectors can be applied to various faces and changes the appearance of the different faces in the same way. In other words, the same random vector has the same meaning regardless of the base face it is applied to. Similarly, an extracted prototype can be applied to novel faces in order to transfer its characteristics to these faces. Moreover, the extracted prototypes can be objectively and statistically compared to other meaningful vectors in face space (e.g., agency and communion), even if they are derived from different base faces.
Walker, M. & Keller, M. (2019). Beyond attractiveness: A multi-method approach to study enhancement in self-recognition on the Big Two personality dimensions. Journal of Personality and Social Psychology, 117(3), 483-499. http://dx.doi.org/10.1037/pspa0000157
Rudert, S.C., Hales, A., Keller, M.D, Walker, M., & Greifeneder, R. (2019). Who gets ostracized? A personality perspective on risk and protective factors of ostracism. Journal of Personality and Social Psychology. Advance online publication. http://dx.doi.org/10.1037/pspp0000271