Modern women are statisticaly less interested in programming. That's because boys 20 years ago got c-64 or PC because they wanted to play games, and girls 20 years ago got barbies.
Turns out barbie doesn't have the same side effects.
That's what most of games were about in 8bit and 16bit times, so boys were interested in them more than girls (because of natural inclination and social norms), so there's more boys than girsl in IT right now.
This will change because there's more gaming girls now, and there are other reasons to have computer.
Natural inclination is unnecessary to the story. The burden of proof is on you to prove that it exists in the first place.
Though while this might be large enough to be considered statistically relevant, nothing I've seen suggests a natural bias strong enough to account for more than a fraction of the mis-balance we see in later life in the science & technology arenas, so social pressure (even "accidental" pressure due to, as you suggest, the "toys" parents and other family buy), both at a young age and as kids progress through school, are presumably the larger factors by quite some margin.
There are natural gender biases, mental as well as physical, that are not caused by external cultural pressure, and to completely disregard that would be a mistake. But I think (caveat: going by my experiences here rather than any scientific study) in most cases that the difference between the average man and the average woman is smaller than the range of differences within each gender (so if you consider only the bulk of the population it might be valid to consider us identical overall).
Research is of varying quality. We see some terrible research comin from actual professors in real universities. Here's one example of someone at a prestigious institution who has anti-science viewpoint: http://wjh.harvard.edu/~jmitchel/writing/failed_science.htm
So, when companies like Mattel (or whoever) do research into the colour pink and girl's toys we should probably be a bit wary about the conclusions they draw, especially when they don't release the full data.
Ex communist countries worked differently.