According to the multidimensional space framework, faces are represented as locations in a psychological face-space. Our aim was to test whether the locations of veridical, caricatured, and anticaricatured face stimuli in face-space, derived from multidimensional scaling analyses, could account for identification accuracy and distinctiveness ratings for these stimuli. Caricatures were identified more accurately and rated as more distinctive than veridical images, significant less distinctive than veridical images. In the face-space derived from multidimensional scaling, caricatures were further from the origin and less densely clustered than veridical image, whereas anticaricatures were closer to the origin and located in denser regions of the space. A quantitative model successfully predicted identification performance from the spatial locations of the stimuli. In general, the physically transformed faces were systematically mapped in the psychological space.