Configural, coordinate, and holistic representations have all been proposed to explain why face recognition is more disrupted by inversion than other stimuli. The current study attempted to determine the nature of the representation that causes the face inversion effect. Experiments 1 and 2 compared the planar rotation functions for face, animal, and object recognition in order to determine whether the rotation function for faces was qualitatively different than the rotation functions for animals and objects. Experiment 3 examined the inversion effects produced by manipulating the number of features present in a face. Experiment 4 tested whether face like inversion effects could be found for houses that, like faces, shared the same structural description. The results of Experiments 1 and 2 showed that the planar rotation function for faces is qualitatively different (steeper) than the rotation functions for animals and objects. Experiment 3 found inversion effects for features of faces removed from the context of a whole face that grew larger as the number of features in the face increased. Experiment 4 found inversion effects for house stimuli that also increased as the number of features to be coded increased. The current set of experiments suggests that the face inversion effect is due to the precision required for the discrimination of objects, the amount of visual information to be coded by a coordinate representation, and the amount experience one has with forming a coordinate representation of an object from a particular orientation.