A lot of the things we sort of know are also related to studies that look at a very small subsystems attempting to isolate variables. Like take a slice of neuro

A lot of the things we sort of know are also related to studies that look at a very small subsystems attempting to isolate variables. Like take a slice of neurons, apply a certain chemical and check how it changes action potentials. Over time a bunch of that kind of data can be pieced together in larger systems analysis. That kind of things relies on extrapolation from that lower-order data though, ideally with confirming studies from subject animals, but the data is really clean. The media reports on research is usually bad too, usually taking whatever speculative impact the research might have that is suggested for funding or future work ... but wasn't actual the results of the paper just something tacked on as basically informed speculation.