For neuroscience to mean something to teachers, it must provide information beyond what is available without neuroscientific methods. It’s not enough to describe what’s happening in the brain, and pretend that you’ve learned something useful.
In general, if you are interested in describing effects at a given level of analysis, you are most likely to make progress by sticking to that level of analysis. If you’re interested in describing ways that students learn best, it makes sense to study classroom situations. To the extent that neuroscience will inform good teaching practice, it seems most likely that this influence will be funneled through the cognitive level of analysis: For example, neuroscience will help us better understand memory, and this improved understanding of memory might be used to improve classroom practice. It’s unlikely that leapfrogging the cognitive level analysis and going straight from the brain to the classroom will work out very often.