[CST-2] Neural Computing - face recognition

[Matthew Richards]

In the lecture on face recognition, Dr Daugman showed some slide (not in
the notes) that showed various faces arranged in a circle, the idea being
that if you followed a line towards the centre of the circle the faces
would look more and more like the "typical" face in the centre, and then
once past the centre they become more like the "opposite" of the original
face - I think he called these "antifaces".

Now, I rather got the impression that this circle was part of some
experiment that supported the theory that our brains encode faces by their
differences from an average face.  However, I didn't gather what the
experiment was, or what the results were.  Did anyone manage to take in
enough of what he said to know the answer?

Thanks,
Matthew