It all has to do with predictability and subjectivity. The goal is always to predict the future based on the present.
Any animal signaling is abstract. Math is, however, the most reliable and predictive form of verbal signaling. Mainly because it is a non-local language.
The function of most language/signaling/talk seems to be to signal (over and over) loyalty to, and membership in, the local group(s). Have to check songbirds – they are great template for verbal signaling.
So the conflict is between predictive talk vs. in group, self-referential talk. By definition, math/evidence-based talk feels like it attacks in-group cohesion, because itignores local social norms — thus:”war.”
It must be a shock similar to having someone say repeatedly “I’m with you on your side”, over and over and then making a factual statement, thus no longer saying soothing things. Our brains code that as an attack.
Apparently, our brains don’t care about facts so much as social cohesion. The main goal of language/talk/verbal signaling.
It seems like verbal signaling’s main job is continual reassurance in social settings, to oneself and others. Pretty useless, and in fact counter to the discovering of facts