but that is making some big assumptions about how different parts of the brain are isolated from each other, isn't it? if my brain is one big neural net then f

but that is making some big assumptions about how different parts of the brain are isolated from each other, isn't it? if my brain is one big neural net then firing up a bunch of neurons that are trained in one language could also activate neurons trained in the associated culture. there's still no way to know that it's the choice of language, rather than the culture context, that is important. [edit:] traditionally you work around this kind of thing by finding subjects that differ only in one variable. identical twins, for example, might better show environmental differences, because they have common genomes. but while that works for nature/nurture i have no idea how you do something similar for culture/language.