If you are only, exclusively, concerned about frequency
domain and everything is perfectly linear then the
frequencies are unique and idealized theory applies.
But if your circuit is by nature time-domain, or has some
nonlinear elements, then it may indeed respond to (say)
zero-crossings from either "tone", or may make new ones
by mixing, etc. And the behavior may also depend on the
relative phase of the two tones if it has to do with zero
crossings, crest amplitude or so on.
A lot of work goes on in RF-land to make the assumptions
of linearity and independence of tones, real.
"In theory, there is no difference between theory and
practice. But in practice, there is".