The idea that only the last few reps of a set, the ones performed close to failure, drive most of the growth from that set.
It follows from motor unit recruitment. Early reps in a set of 15 use only the small units; by rep 12 or 13 the large ones are working, so the final three to five reps are the ones the muscle has not already been doing all week.
Treat it as a heuristic, not a law. It is not established that a rep at 4 reps in reserve contributes nothing, and the model breaks down at very high reps where cardiovascular limits arrive before muscular ones. What it does explain well is why sets taken to RPE 5 build very little and why a set of 6 near failure and a set of 15 near failure can produce similar growth.
The practical rule that comes out of it: fewer sets taken close to failure beat more sets taken nowhere near it.