None of those arguments claim "LLMs could not possibly be good models of some cognitive capacity"
What about the cognitive capacity of understanding?
What about the cognitive capacity of understanding?