Not exactly. Additive and subtractive colour models are based on continuous physical phenomena, while XOR is a discrete operation lacking a real-world optical counterpart.

Subtractive colour models are based on mixing of pigments, which absorb certain frequencies of light and reflect others—red paint, for example, absorbs non-red frequencies from white (full-spectrum) light and thus reflects red hues. Additive colour models are based on mixing of light frequencies, as with a computer monitor.

The key thing to understand, though, is that additive and subtractive don’t refer to addition and subtraction in the algebraic sense, but rather in the frequency-response sense, which is more akin to audio mixing.

Additive and subtractive mixing can be defined in terms of formulas that approximate certain real-world properties (mixing of coloured light, mixing of paint, application of successive coloured filters etc.)

XOR doesn’t really make sense in this context, for two reasons:

• XOR works with binary values, 1 and 0 (or true and false). It isn’t normally defined on continuous values, which is what you need if you are dealing with colour spaces
• What real world property are you trying to model? To my knowledge, there is nothing in nature that displays XOR-like colour mixing behaviour.

You could, of course, define an artificial mixing function that exhibits XOR-like behaviour at extreme values (0.0 and 1.0) and does something else in between. You might even get some interesting visual effects. But it’s not a standard technique.