[bduerst]

The problem with evolutionary behavior hypotheticals is that it's too easy to craft them to fit a narrative - even with yours the inverse is still true.

The next generation of the species is all PB-seeking high risk ants, which expend more energy & die off faster, reproducing less. Diversity shrinks over generations as natural selection picks off most of the high risk individuals, leading to a uniformity.

This still doesn't apply to humans, because even food preferences weren't really subject to natural selection (outside food neophobia in children), because pre-agricultural humans ate whatever they could get their hands on.