Ketosis for the body means fat-burning. For the brain, it means a lower seizure risk and a better environment for neuronal recovery and repair.
Ketogenic diets have been prescribed for seizures for a long time. The actual research diets used in the past were pretty dismal and seemed to involve drinking a lot of cream and eating a lot of mayonnaise. At Johns Hopkins, pediatric patients were admitted to the hospital for a 48 hour fast and then given eggnog (minus the rum and sugar, I'm guessing) until ketosis was achieved (usually took about 4 days). In addition, ketogenic diets were calorie restricted to just 75-90% of what would be considered a child's usual calorie intake, and often they were fluid-restricted too (1)! If we're talking soybean oil mayonnaise, you could see how someone could get into trouble with mineral deficiencies and liver problems pretty quickly

To understand "dismal," some of the latest research showed that a "modified Atkins protocol" was just as good as the classic ketogenic diet, and so much more liberating, as the patients were allowed up to 10 grams of carbohydrates daily, and they didn't begin with the fast, and they weren't calorie restricted (2)(3). While the classic ketogenic diet was 4:1:1 fat to carbs to protein. If you use MCT oil for 50% of your calories (have to add it in slowly though to prevent vomiting, diarrhea, and cramping!), you could increase the carbohydrates and proteins to a 1.2:1:1 fat:carb:protein and still get the same numbers of magical ketones circulating. And while "MCT oil" sounds nice and yummy when it is gorgeous coconut milk, this MCT Oil 100% Pure 32 fl.oz doesn't look quite as appetizing, especially when that is going the be half of what you eat for the foreseeable future (4). You can see why researchers consider ketogenic diets (especially the original versions) to be extremely difficult and unappetizing (they were), whereas seasoned low-carbers (who have a bit of a different idea what a ketogenic diet is) will find that attitude ridiculous, especially when you compare a ketogenic diet to the side effects of some anti-epileptic medications.

So it looks like modified Atkins (very very low carb, but not zero carb) and a preponderance of MCT is the same, ketone-wise, for the brain as the classic cream-heavy ketogenic diet. And what does it mean to have a ketogenic brain? Before, we talked about protons, but now I'm going to examine neurotransmitters and brain energy more closely. Specifically, glutamate and GABA (5).

If you recall, GABA is the major inhibitory neurotransmitter in the mammalian nervous system. Turns out, GABA is made from glutamate, which just happens to be the major excitatory neurotransmitter. You need them both, but we seem to get into trouble when have too much glutamate. Too much excitement in the brain means neurotoxicity, the extreme manifestation of which is seizures. But neurological diseases as varied as depression, bipolar disorder, migraines, ALS, and dementia have all been linked in some way to neurotoxicity.

Glutamate has several fates, rather like our old buddy tryptophan. It can become GABA (inhibitory), or aspartate (excitatory and, in excess, neurotoxic). Ketogenic diets seem to favor glutamate becoming GABA rather than aspartate. No one knows exactly why, but part of the reason has to do with how ketones are metabolized, and how ketosis favors using acetate (acetoacetate is one of the ketone bodies, after all) for fuel. Acetate becomes glutamine, an essential precursor for GABA.

Here's the confusing part. A classic ketogenic diet had three major components which were thought to contribute to the anti-seizure effect. One, it was calorie restricted. Just calorie restricting epileptic monkeys (no matter what the macronutrient ratios) reduces seizure frequency (and increases longevity). Secondly, it was acidic, and the extra protons themselves could block proton-sensitive ion channels, or the ketone bodies or fats themselves could affect the neuron membranes, making them harder to excite. (For the biochem geeks out there, ketones or fats seem to affect ATP sensitive K+ ion channels, making hyperpolarization easier to maintain). Thirdly, it lowered glucose levels. And lower glucose is associated with a higher seizure threshold (that's good - once doesn't want to easily have a seizure!) and less neuronal excitability. Gads. Doesn't sound to me like glucose really is the preferred fuel for the brain after all.

And now let's really get down to the mitochondrial level. Mitochondria are the power plants of our cells, where all the energy is produced (as ATP). Now, when I was taught about biochemical fuel-burning, I was taught that glucose was "clean" and ketones were "smokey." That glucose was clearly the preferred fuel for our muscles for exercise and definitely the key fuel for the brain. Except here's the dirty little secret about glucose - when you look at the amount of garbage leftover in the mitochondria, it is actually less efficient to make ATP from glucose than it is to make ATP from ketone bodies! A more efficient energy supply makes it easier to restore membranes in the brain to their normal states after a depolarizing electrical energy spike occurs, and means that energy is produced with fewer destructive free radicals leftover.

Umph. What does it all mean? Well, in the brain, energy is everything. The brain needs a crapload of energy to keep all those membrane potentials maintained - to keep pushing sodium out of the cells and pulling potassium into the cells. In fact, the brain, which is only 2% of our body weight, uses 20% of our oxygen and 10% of our glucose stores just to keep running. (Some cells in our brain are actually too small (or have tendrils that are too small) to accommodate mitochondria (the power plants). In those places, we must use glucose itself (via glycolysis) to create ATP. When we change the main fuel of the brain from glucose to ketones, we change amino acid handling. And that means we change the ratios of glutamate and GABA. The best responders to a ketogenic diet for epilepsy end up with the highest amount of GABA in the central nervous system.

One of the things the brain has to keep a tight rein on is the amount of glutamate hanging out in the synapse. Lots of glutamate in the synapse means brain injury, or seizures, or low level ongoing damaging excitotoxicity as you might see in depression. The brain is humming along, using energy like a madman. Even a little bit more efficient use of the energy makes it easier for the brain to pull the glutamate back into the cells. And that, my friends, is a good thing.

Let me put it this way. Breastmilk is high in fat. Newborns (should) spend a lot of time in ketosis, and are therefore ketoadapted. Being ketoadapted means that babies can more easily turn ketone bodies into acetyl-coA and into myelin. Ketosis helps babies construct and grow their brains. (Update - looked more into this specifically and it seems that babies are in mild ketosis, but very young babies seem to utilize lactate as a fuel in lieu of glucose also - some of these were rat studies, though - and the utilization of lactate also promotes the same use of acetyl-CoA and gives the neonates some of the advantages of ketoadaptation without being in heavy ketosis.)

We know (more or less) what all this means for epilepsy (and babies!). We don't precisely know what it means for everyone else, at least brain-wise. Ketosis occurs with carbohydrate restriction, MCT oil use, or fasting. Some people believe that being ketoadapted is the ideal - others will suggest that we can be more relaxed, and eat a mostly low sugar diet with a bit of intermittent fasting thrown in to give us periods of ketosis (though in general I don't recommend intermittent fasting for anyone with an eating disorder). Ketosis for the body means fat-burning (hip hip hooray!). For the brain, it means a lower seizure risk and a better environment for neuronal recovery and repair.