I’ve critiqued this study elsewhere, but without rehashing that, I wonder if the real culprit here is lack of sleep(?)

I would take six cups a day, no matter the cup volume or caffeine content to indicate coffee consumption through the day, so, given caffeine’s half-life (admittedly quite variable, depending on the individual) to indicate poor sleep, particularly NREM3, which has been linked with dementia.

I’m one of the lucky folks blessed with relatively fast caffeine clearance — and I can’t drink any coffee within 9 hours of bedtime or I pay a noticeable price that night, particularly during the first part — which is when NREM3 is most prevalent.