The map of the sample
The appeal of the PNAS study is that it does not stand in one place. The Carnegie Mellon team tested olive baboons and rhesus macaques, children aged three to six, seventy-nine adults from Bolivia's Tsimane' community and twenty-one Americans on the same touchscreen shape-matching task. That is good design: if a finding repeats across species and across cultures, its odds of being one lab's caprice fall. My 'in how many?' is here answered more than once, in more than one population.[1]
Still, fairness demands: 'continuity' is a strong word, 'identity' is not. The authors themselves say this is a foundational capacity children later turn into mathematics, and that they do not address how education builds on it. What is measured is a shared beginning; not a shared end.[1]
Until it replicates, a hypothesis
Yesterday I praised a study for being 'finally a paper with numbers'; today my praise is conditional but sincere. This is a result that earns being taken seriously not because it is striking but because it is soundly built. Even so, as a referee my job is to want the good result to be real — which is why I try to break it.[1], [2]
My forecast: this finding's fate rests on an independent lab re-running the same task with different primates and different cultures. The samples are mid-sized; the touchscreen task is an abstraction, not geometry in the wild. If the signal beats again in a second team, the sentence 'humans and monkeys share a language of shapes' turns from a headline into knowledge. If it does not, we are still left with an elegant hypothesis — which is no bad outcome. Until it replicates: the line between news and knowledge is still right there.[1]