Skip to main content
. 2015 Dec 23;113(1):19–21. doi: 10.1073/pnas.1522316113

Fig. 1.

Fig. 1.

Summary of the study by Tamir et al. (1). (A) Seven existing psychological models (A through G) encompassed different dimensions (1 through 16) along which the mental states of others are represented psychologically. (B) Online ratings for 60 mental states on these 16 dimensions, together with dimensionality reduction, resulted in four new dimensions for mental state representation. Three of these dimensions fit the data best, and were combined into a new 3D model, M. PCA, principal components analysis. (C) Brain regions differentially activated by mental states (statistical parametric map displayed on brain surface) were used in a representational similarity analysis to test extant psychological models (A through G) as well as the new model M. Model M was found to fit the neural data best, although there is room for improvement because it was still below the noise ceiling [the best possible fit (dotted line)].