[OscarCunningham]

The code itself also contradicts that sentence. Notice that the first roughness graph doesn't have any local minima at rational numbers. It's only when the overtones are added to the notes (at integer multiples of the fundamental frequency) that the minima appear.

So the code thinks that human ears don't detect integer multiples specifically, they just detect sounds whose overtones line up with each other.

[sideshowb]

Good point, but our ears definitely do perceive the 2 to 1 ratio. Maybe that particular phenomenon is better analysed through the usual pschoacoustic approach of "at what point do we stop perceiving one sound and start perceiving two "?