Models of face perception suggest that faces are perceived with reference to face 'prototypes'or 'norms'. Research has shown that race-contingent aftereffects can be simultaneously induced using faces of two different races, distorted in opposite directions, as adaptation stimuli. Subsequently, a test face of a certain race will appear transformed in a manner consistent with the adaptation images used for that race, suggesting the existence of multiple prototypes. We examined whether race-contingent aftereffect size is predicted by levels of perceived racial typicality or by dichotomous racial categorisation. In experiment 1, faces with a range of 'morph levels'(ie relative contributions of Asian/Caucasian faces) were either rated on a continuous scale for Asian/Caucasian typicality, or simply categorised as Asian/Caucasian. As expected, typicality ratings showed a shallow slope (observers were sensitive to morph level over a broad range), while dichotomous racial categorisation showed a steep slope (rapid switch from categorisation as Asian-Caucasian). In experiment 2, race-contingent adaptation was assessed using test faces with various morph levels. Aftereffect size showed a shallow slope, more closely resembling racial typicality than categorisation data. This suggests that faces are not exclusively coded against one prototype, but instead the visual channels processing faces are broadly tuned.