jesus fucking christ i don’t understand neuroscience at all

i have to read a paper that’s basically about comparing how monkeys’ neurons fire/are connected when they do a basic matching/pattern-recognition task, versus a neural network trained to do the same thing, and they’re making some kind of conclusion based on–– well, i think they’re arguing that their neural network is a good model for the actual brains, and then based on that saying they can make a bunch of extended other predictions now using their neural network as a “good”/“accurate” model. which, yes, is just how all data-based model-building works. i have no idea what those conclusions are, but uh––they made them, maybe!

see, i can very roughly look at a figure like this 

and say, ok, the top is their biological data, the bottom is the model, and they’re arguing that the outcomes are similar enough that the model is doing a good job and can be studied on its own bc it’s potentially biologically relevant. i don’t know what any of these features mean (the x axis is milliseconds and the y axis is Hz, if you’re curious) or what it would look like for them to be disqualifyingly dissimilar, but sure. those look like they’re doing the same thing, why not.

but then they move into talking about the model on its own and like

apparently these graphs are of the “dynamical landscape” of neural trajectories.

my feeling on this: ‽‽‽‽‽‽‽‽‽‽‽‽‽‽‽‽‽‽‽