A study introduces innovative ways to test whether neural population activity exhibits structure above and beyond that of its basic components.
Suppose a fancy new analysis method reveals an (apparently) surprising form of population-level organization in your large-scale neural data set. How can you tell if the observed pattern is truly surprising? Is it the hallmark of a population-level mechanism that reveals the circuit’s true function, or is it merely an expected byproduct of things we already knew about neurons contained in the population? To put it bluntly: when are findings of population-level structure ‘new science’ and when are they merely old knowledge dressed up in new clothes? In this month’s issue of Nature Neuroscience, Elsayed and Cunningham propose new methods for resolving this question1.