I just stumbled across this paper by Jung et al. (1999). The abstract:

Proton magnetic resonance spectroscopy (1H-MRS) offers a unique non-invasive approach to measurement of N-acetylaspartate (NAA) and choline (Cho), putative markers of neuronal and glial integrity. Previous studies revealed that these neurochemicals predict cognitive impairment in diseased subjects, although little is known about their relationship to cognitive functioning in healthy people. We measured the concentrations of NAA and Cho in the left occipitoparietal white matter of 26 healthy adults and compared them with intellectual performance assessed by the Wechsler Adult Intelligence Scale-3. We found that NAA (b = 0.6, p up to 0.01) and Cho (b = -0.42, p up to 0.01) were independently associated with the Full-Scale Intelligence Quotient. Together, these metabolites accounted for a large proportion of the variance in intelligence (r2 = 0.45). Possible mechanisms underlying these correlations, such as mitochondrial function and myelin turnover, are discussed. 1H-MRS is a sensitive new tool to assess the neuronal underpinnings of cognitive function non-invasively.

Jung and his colleagues have replicated this result in two other samples, although I apparently do not have access to the journals where the reports appeared. It is implied that NAA is likely a mere marker of some other brain state causally responsible for enhanced cognition, but it seems that this is not known for certain (Barker, 2001).

Curious about what exactly N-acetyl-aspartate is, I looked it up in Wikipedia. I got only one hit: Canavan disease.

Canavan disease is caused by a defective ASPA gene, responsible for the production of the enzyme aspartoacylase. This enzyme breaks down the N-acetyl aspartate acid, which is a toxin. With decreased levels of aspartoacylase comes an increase in N-acetyl aspartate, which interferes with growth of the myelin sheath of the nerve fibers in the brain. The myelin sheath is the fatty covering surrounding nerve cells that acts as an insulator, and allows for efficient transmission of nerve impulses.

Hmm. Recall that Canavan disease is one of the neurological diseases identified by Cochran et al. (2006) as a a likely side effect of selection for IQ among the Ashkenazi Jews. So maybe we have something like:

one defective ASPA allele -> reduction of apartoacylase -> increase in NAA -> white matter gets a boost -> big increase in IQ

two defective ASPA alleles -> larger reduction of apartoacylase -> larger increase in NAA -> white matter gets overclocked -> Canavan disease

Probably not so simple. But worth keeping in mind as we go forward.