@albertcardona @UlrikeHahn This is a very interesting paper! I'm fascinated by this research space at the intersection of neuroscience and deep learning. Though, it's a little challenging to understand, given the unique way that methods and terminology blend together here! In particular, neurons with temporal dynamics are "nearly universal" in one field and "nearly unheard of" in the other, which is fun. From a deep learning perspective, it's a little disappointing that they didn't attempt to scale beyond CIFAR-10 in terms of task complexity, and that they didn't offer any analysis of the computational costs to train or run inference with their proposed architecture and training scheme. I realize the focus was informing neuroscience, and this study is already enormous, but I'm sad they neglected that side of things. I look forward to more on this front. Thanks for sharing!