There are two sorts of extreme IQs I’d like to talk about: melon and thal. There may be others, but I don’t have a theory to explain those yet. I’m going to start with melons because it’s a quicker explanation.
Beyond a baseline healthfulness and capacity for growth of new neurons, axons, synapses, etc., IQ is not a very useful measure for general ability (as in Spearman’s g). After that, IQ scores grow according to the healthy production of special modules that happen to be useful for answering problems on IQ tests, like having a large working memory, high spatial manipulation ability, etc.
These modules don’t always play nice. For instance, a person with mixed melon and sapiens genes may have a functioning southbridge (we’re playing loose with analogies here) but no memory management controller. This is effectively “retardation” in the truest sense, even though the brain will make use of the extra space due to the miracle of neuroplasticity, and the person will still have a fairly high IQ. (This is another reason I often describe thals as thards: because we’re all probably poorly mixed.) However, the brain can’t make better use of the space than its original design for a full melon brain: a southbridge cooperating with an MMC is a powerful combination.
Now, consider somebody like Goethe. If we believe that intelligence is strictly distributed according to a bell curve (or approximation), then Goethe is one of those guys who should not exist. Estimates of his IQ vary, but they are all north of 200-205. This statistical rarity essentially disproves the hypothesis that genetic correlation will tell us which switches to turn on/off for everyone to have the highest possible IQ: there is competition among them to determine two or more conflicting archetypal designs for the brain.
In Goethe’s case, there is no such conflict (or much less than usual) so that all of the modules are working together in a holistic fashion. It’s not that his brain is greater than the sum of its parts, it’s that everyone else is retarded because they are either pure homo sapiens, or they are trying to put two different chipsets on the same motherboard.
Something else this indicates is that a full melon is going to have an IQ north of 200. Goethe’s head is not unusually large, although it obviously grew continuously over his lifespan like a turtle’s shell. (Go here and scroll down for the gallery of portraits over time.)
Now, I mentioned that neuroplasticity plays a big role here. That’s true: if a special module exists in the brain somewhere, then the rest of the brain will hack together a connection and find some way to make use of it. But this will not necessarily be the best use…unless the brain is following the optimized genetic blueprint laid out by one of the high-IQ racial archetypes. This is especially true for melonheads, whereas the situation for neanderthals gets a little weird due to cognitive style.