Comment by ilaksh

I think what that shows is that in order for the fast reactions to be useful, they really have to incorporate holistic information effectively. That doesn't mean that slower conscious rational work can't lead to more precision, but does suggest that immediate reactions shouldn't necessarily be ignored. There is an analogy between that and reasoning versus non-reasoning with LLMs.