It's as if signals from a face recognizing network in the brain are misdirected into the visual cortex where they inappropriately influence perception of basic

It's as if signals from a face recognizing network in the brain are misdirected into the visual cortex where they inappropriately influence perception of basic shape. Because I think here's the kicker. Semantic info about faces has to be correlated back to their position in your visual field. If you're looking at a crowd of people, your brain can't just come up with a list of the ones you know. The correct info has to be attached to and track each respective face. There's kind of a tagging going on which has to merge into the representation. Normally the tag is not visible: it is in the semantic layer that has no color or shape, so it doesn't influence what you're seeing. But it is pinned to something you're seeing. Imagine some wires get crossed there. Information is pinned in the wrong way, corrupting the visual.