Nope.
IIRC it was mainly that it would be hard to implement in a sane way on systems that don't have a (full) colour display. Having four static colours allows for manual intervention, whereas having unlimited dynamic colours does not.
For example, when going to four level grayscale: if the program selects red and green with equal luminosity, how would the interpreter know how to map the colours without losing legibility? For four static predefined colours a human can intervene and configure the program to use colours that make sense.