I see it like natural sciences strive to do replicable experiments in outside world, while math strives to do replicable experiments in mind. Not everything is transferable from one domain to the other but we keep finding many parallels between these two, which is surprising. But that's all we have, no foundational truths, no clear natural/unnatural divide here.