Abstract: The ability of the brain to absorb and incorporate within itself new ideas implies that it is a metastable system. It must continually change and yet not devolve into randomness. How does it do this, and what are the consequences? Brain activity can be represented in terms of a large collection of excitable bodies that possess both spontaneous activity and that can stimulate other bodies to become excited. I discuss a simple computer model of such a system and study (1) what is necessary for such a system to learn, and (2) what is necessary for it to maintain itself in a state capable of further learning. It turns out that the highest performing brain models that are able to maintain stable learning also show self-organized criticality. Unfortunately, the homeostatic constraints that maintain optimal brain performance also predispose the brain toward neurological disease. The relationship to epilepsy is presented, and an approach to its cure is proposed.