Reviews: Neural system identification for large populations separating "what" and "where"

I am no expert in this field, so right now I can not fully judge on the potential impact of the paper at hand in this field of science. So first I want to ask a few clarifying questions to the authors to give me a clearer picture. I will also give a preliminary judgement at the end. Introuction 34-47: I think it should be mentioned that using more expressive models such like CNNs will come at a cost. Simple models such as GLMs are very interpretable, it also easier to argue for biological plausibility than with neural network trained by backpropagation.