It's an interesting theory, but I'm not sure it really explains anything. E.g. that parts of the brain grow with use is certainly a surprising and interesting result. But it's not different than muscles growing with use, I don't think it requires agency to explain.

Split personality is interesting, but also incredibly rare and may not really exist at all. Schizophrenia appears to be the result of physical damage to brain tissue, not merely healthy brains developing independent agents. Addiction is easily explainable with reinforcement learning gone wrong.

There are theories of neural learning algorithms that are agency. Like the bucket brigade algorithm. But as far as I know that never worked very well. It's difficult or impossible to determine the "reward" for neurons, and having them be agenty doesn't really help anything.

Cool idea, though I'd sooner suspect most "supernatural" phenomena are more easily explained-- with fewer assumptions and suppositions-- by simple psychology, misinterpretation, fallacy, drugs, sleep paralysis, fear, rationalization, or attention-seeking, without any need to appeal to multiple consciousnesses or neurons with agency.

I described a similar theory of neuronal agency in 2005: http://thinkinghard.com/blog/HowDoesANeuronKnowWhatToDo.html.

Interesting theory. I could see the main idea, individuality of brain segments, to be extended to cover other dimensions than just the spatial. Especially with schizophrenia, time perception and illusions are quite common, so a temporal schism of brain as a process could be an alternative approach. The combinatorial or additive emergence of agency is not as given as the article implies it is, though.