Yeah, there's some math that describes it.
A switch alternates between a very small resistance when closed and a very large resistance when open. So much so that you can usually ignore the resistances that do exist there because they're trivially bigger or smaller than would make a difference.
When you go putting blend pots in there, the resistances in the non-full-on positions are no longer negligible. They cause frequency rolloff and shelving effects that may or may not sound like a variable cap like you've described. This may be good or bad, but it's not all variations between two caps. Also, the non-negligible nature of the resistances mean that they interact with the source impedance driving the input and the input impedance loading it. Again, this may not be bad, but it is not all variations between two caps.
The math comes down to describing the driving signal impedance, the two caps and two resistances the pot makes depending on how you hook it up, and the input impedance loading it; the description is best done with complex (i.e. real+imaginary numbers) or s-transforms, and once you describe the impedance of the network, you can do some algebra to get the transfer of voltage and/or current as a function of frequency and plot that as either complex numbers or magnitude and phase.
Yes, there are simpler approximations. If the small and large caps are quite different, you can ignore the small one at low frequencies because it may be much higher impedance than either the resistor and large cap. At high frequencies, the big cap is long since much smaller than the resistor, so you only need to take into account the small one. Between the two points, or if the caps are near one another in size, or the resistor changes a lot, it gets messy. At least messy to quick and dirty approximations.
Your caps are about 20:1 different. That's awfully close for being able to ignore one or the other. You're right - if you have the pot set up as a variable resistor, the resistor set to zero (and not all pots actually go to zero; there's some end resistance in many of them) and the caps are paralleled effectively. With the resistor at max, the resistor plus large cap form a high pass which depends on the following driven impedance. This gets "shorted" at some frequency above the RC turnover point of the resistor and the small cap, but the driven impedance still determines the overall rolloff. The sweep between the two amounts to the changing turnover points of R*C1 and R*C2. It's linear in frequency with each of the caps. But the ear "wants" to hear exponential frequency changes as linear, so you may need to go either log or reverse log, depending on how you want it to act.
