I also like "Refactoring your Wetware": https://pragprog.com/book/ahptl/pragmatic-thinking-and-learn...
However, from another angle, it's basically about using the idea of language learning to teach basic cognitive psychology.
(My phd was in human memory, and I can't stop thinking about ways this book could be used as part of a cool learning course).
RE networks, I agree that there is likely evidence for distinct patterns of activation in various neuroimaging studies, but having worked in memory + neuroimaging, I think there's a serious risk that people will take something like "statistically significant difference in brain activity" and use it as a substitute for "substantial differences in learning behavior / retention". (this is a well known problem in imaging).
I'm not too familiar to L2 acquisition research, though, but those are my impressions from thumbing through some of the field. Would def love to hear some study recommendations :).
Greg Wilson (who founded Software Carpentry) has a great collection of thoughts on learning to program in general: http://third-bit.com/
https://www.coursera.org/learn/learning-how-to-learn
Learning at scale via MOOCs seems to be enormously effective. EdX alone issued 250K certificates for 2.5M registered users. Mostly in CS.
I'd be interested to see YC Startup Schools own results as well. Do at least 10% of Startup School 2017 grads go on to full time work on their companies?