Is there a face-specific processing module which is cognitively and/or neurally dissociable with the module of object processing? And how is face processing system organized? Both of these questions are the core issues in the current cognitive neuroscience of face processing. With evidence from both neuropsychology and neuroimaging studies using ERPs, PET or fMRI, researchers had found a face-specific brain area, which was named fusiform face area (FFA). In this paper, we first review some critical studies investigating both the specificity and the multiple components of face processing. We then evaluate the popular cognitive and neural models of face recognition and speculate on the direction of future studies.