Infants’ perceptual abilities allow them to distinguish faces of different races and genders from an early age (for a review, see Pascalis et al., 2011). However, it is still unknown when infants begin using these perceptual differences to represent faces in a conceptual, kind-based manner. The current dissertation examined this issue by testing whether 12- and 24-month-old infants represent faces of different races and genders as distinct ‘kinds’ or instead as variations of a single broader category (e.g., ‘human face’). The current dissertation included two experiments each with a different type of violation-of-expectation individuation paradigm. Experiment 1 used a passive viewing looking-time measure to establish infants’ baseline response to human- vs. non-human faces as well as their response to male versus female faces. Results from Experiment 1 replicated previous looking time measures and found evidence that 12-month-old infants have a representation for an ontological or ‘human’ kind. Using a manual search paradigm, infants’ face individuation based on gender and race was assessed in Experiment 2. Twenty-four-month-old infants, but not younger infants, displayed reaching behaviors that indicated they individuated faces based on the kind ‘human’ as well as face gender. There was no evidence of individuation based on a face’s race group. The current findings help to determine how infants begin to conceptually represent gender and race differences as identity-defining variations that might serve as a starting point for socio-cognitive biases observed later in development.