This makes it a top down theory which means that the face is more important than the individual features when recognizing a face. When we recognise a face as a whole, we call it holistic processing. When you first see a face, you make a template of it in your memory. This template is called a schema. A dictionary definition of a schema is just a diagrammatic representation of the face. When you see the face again, you make a template and search through prior existing ones to see if there is a match. This is called structurally encoding. If there is a match, then you move onto the next stage. Here, your mind tries to fit the relevant FRU's into the template. If there is an exact match, then again, the next stage, person identification, is activated. In the person identification stage all the information you know about the person is accessed. This would include their job, where you know them from, their interests and whether or not you like them. The last stage to be accessed is the name generation. This is because it is stored separately to the other information. This is called the linear holistic form model. As it is linear each stage must be accessed in order. So, you cannot access a persons name before you have the information on them. As the FRU's must match the template exactly, global patterning must occur. This explanation of face recognition accounts for the whole face, not just the features. Other factors enter into the face such as spacing and complexion, which would not allow recognition in the feature analysis theory. .
However, in both theories different factors can affect recognition. Such things as the context you know them in, or the length of time since you last saw them can affect your recognition as well as changing their features. However, the more familiar the person is, the more chance there is of recognition. Other factors, such as you eyesight, distance from the person and the lighting can affect also, but tend to affect holistic processing more.