Janet Hyde and Janet Mertz manage to show a significant correlation between the percentage of girls on a country's International Mathematical Olympiad Team, and that country's World Development Indicator Gender Gap Index. The emerging pattern is quite clear: The greater the gender parity in a country, the more girls go to the Math Olympiad; thus indicating a significant role - who could have doubted it - of social equality in girl's performance on this (and other) indicators of mathematical achievement.This fits into the broader discussion of genotype versus phenotype that plays out elsewhere in economics. A person's phenotype is what we observe, but too often we attribute this observable characteristic to the underlying genotype, or inherited characteristics. In reality, not only can social forces during one's lifetime change one's achievement, but also tiny changes in utero can impact gene expression. For example, this article on iodine deficiency in Tanzania shows that correcting maternal iodine deficiency in the first trimester of pregnancy led to large cognitive gains for daughters of affected mothers.
While it's certainly possible that there are differences in the way male and female brains process numerical information, or develop with regard to this ability (some people say girls simply learn math later), I would say attributing the bulk of observed differences to genotype when we know so many social differences exist is just lazy.
[Hat tip Larry]