Several brain areas are specialized or involved in the processing of faces, most importantly the fusiform face area (FFA). The FFA is also claimed to be involved in visual expertise, responding to any objects in which we are experts – not only faces. The question whether faces are processed in a special manner (domain-specific hypothesis), or if they are objects of expertise that obey the same rules as non-face objects (process-specific hypothesis) is still highly debated. In order to test various aspects of visual expertise and to compare them to the mechanisms involved in faces, a new class of stimuli was created: Greebles. Because they share the same visual properties as faces, Greebles have become a standard tool in the study of visual expertise. However, the main limitation with Greebles is that they are human-like. Not only do they resemble humans with a head-like shape, but they are also learned with genders and individual and family names during training, which might encourage participants to associate them with humans. In this study, we trained two groups of participants to become Greeble experts. One group learned Greebles with human names, while the other learned them with object names. Each participant was scanned in two fMRI sessions; one before the training, the second one after. We found that Greeble training influenced FFA activation only in the group that learned them with human names, not in the object names group. These results suggest that the method of expertise training can influence the way we encode a stimulus, and hence the way we process it. They also demonstrate that expertise training with non-face objects does not influence FFA activation, which is in favour of the domain-specific hypothesis.