A helmet is only needed if you intend to spend significant time in traffic.
The worst wreck I’ve ever had on a bike was without a single car in sight. Pinch flat while carrying speed through a steep downhill curve. I split an expensive MIPS helmet in two and still hit hard enough that I had a minor concussion, road rash up one side of my body, and cracked the face of a week old watch just to pour salt in the (metaphorical) wound. I mostly landed on my head and that helmet is the reason I didn’t have drastically more severe head injuries.
Helmets aren’t just for traffic.
Which would mean ~1 in 5 have absolutely nothing to do with a motor vehicle. That’s significant.
Then that should be an easy [citation needed] for you because my searches are coming up blank for actual studies. Lots of assertions of it, but I’m not finding anything in terms of actual data.
It’s very easy, on the other hand, to find comprehensive meta analyses on the efficacy of helmet use.
It’s also worth noting that the introduction makes a point of calling out another common online assertion that you repeated – that helmets make people engage in more risk-taking behavior – as false:
I don’t think they’re stupid. I think they’re bad at risk analysis. That’s a pretty inherent feature of humans. It’s the reason I want to see actual data.