There's pretty robust evidence for intelligence (at least as measured by IQ) being heritable. There've been a bunch of identical-twin studies showing heritability factors as high as 0.8, which is roughly as heritable as height and among the highest of personality traits:
As for why these intelligence genes wouldn't spread throughout the population: it could be because they have other effects with a negative survival/reproductive value as well. Intelligence is often inversely correlated with confidence, for example, since people's expectations of themselves rise faster than their abilities. Confidence is heavily correlated with social skill and the ability to attract a mate, so highly intelligent people often face severe deficits in the mating game (I'm sure we all know someone who is brilliant but completely socially inept, and many of us have been that person).
Also, remember that the ancestral environment that humans evolved in significantly less favorable to intelligence than the modern one. 15,000 years ago, intelligence might be nice, but the ability to outrun a lion or take down a buffalo was far more important. It's only within the last 3-5 generations that intelligence has become essential to surviving in modern society (that's the whole point of Flynn's thesis), and evolution doesn't work so well on a timescale of 3-5 generations.
http://en.wikipedia.org/wiki/Heritability_of_IQ
As for why these intelligence genes wouldn't spread throughout the population: it could be because they have other effects with a negative survival/reproductive value as well. Intelligence is often inversely correlated with confidence, for example, since people's expectations of themselves rise faster than their abilities. Confidence is heavily correlated with social skill and the ability to attract a mate, so highly intelligent people often face severe deficits in the mating game (I'm sure we all know someone who is brilliant but completely socially inept, and many of us have been that person).
Also, remember that the ancestral environment that humans evolved in significantly less favorable to intelligence than the modern one. 15,000 years ago, intelligence might be nice, but the ability to outrun a lion or take down a buffalo was far more important. It's only within the last 3-5 generations that intelligence has become essential to surviving in modern society (that's the whole point of Flynn's thesis), and evolution doesn't work so well on a timescale of 3-5 generations.