>>14389038At least with NNs, the model that comes at the end of training is an optimal path built by trial, error, and randomness. Assuming you're good with the model and don't want to update or retrain, most of the components of the network usually wont be needed since only a particular highlighted path will be hit anyway.
It could be possible that the brain behaves the same way. Then the key thing to watch for is not loss of functionality, but inability to gain completely new functionality, or even to learn new ways of doing old things.
