It's very hard to tell. Clearly having infrastructure isn't direct evidence of present intelligence; the rate of innovation's probably related, but it's also affected by communications, culture, and scale effects, and then we've got the Flynn effect confounding things recently. This gets even harder when we talk about culture 20,000 years ago; that's before writing or large-scale societies, so ideas would propagate much more slowly and would be more prone to dying out under suboptimal conditions.

If we had a good idea of what might be driving the changes in brain size, we could look at present-day or at least recent cultures subject to similar forces and see what the deltas there are -- although that opens up the usual cross-cultural psychometric can of worms. But we don't have anything I'd consider compelling, and we certainly don't have any Paleolithic tribesmen handy that we can sit down with Raven's progressive matrices, so that leaves us with various proxies for intelligence -- of which braincase size is one, at around .4 significance if I remember right.