He isn't claiming that this is how the brain works, he is just proposing a model. He describes in some of his talks that his work isn't about replicating the brain, it's about distilling algorithms from brain function/anatomy.