A follow up to our discussion. I'm including a link to a paper. It's a bit thick, I recommend just reading the conclusions and then if you're ambitious delving into the details. I'm vaguely famaliar with nonlinear systems but I'm lost very quickly in this paper. The results can be understood though with a minimal knowledge of chaotic systems.

From the paper:

The agents producing the seizure tend to drive the brain activity toward a stable periodic motion. In such states, information processing would be impossible and recovery would be extremely difficult. However, the brain manages to remain on a chaotic attractor, although one of a very low dimensionality, in order to process reflex activities. (continued) Just to help out with the above paragraph. Attractors are mathematical objects needed for a chaotic state to exist (Chaotic is good when it comes to brain funtions). Dimensionality is refering to how complicated your chaotic structure is (the higher the better for the brain, to a certain extent). So their saying that all but a little of the chaotic nature of the brain's functioning disappear during an epileptic fit, except for the minimal ones needed for bare survival (breathing, etc.). The normal chaotic nature is replaced by a periodic nature (This is measured through brainwaves).

Here is the paper.