Most people have a nasty tendency to think that memory is something where you specify a memory location, the CPU applies that memory location to the address lines, then the CPU reads the results from the data lines. (Or you specify the memory location and some data, then the CPU applies the memory location to the address lines and the data to the data lines, leaving the RAM to make the data part of its state.) That is a very CPU-centric view of the world. It is also totally incorrect.
Once the address is on the address lines, the external circuitry can do whatever it wants with those electrical impulses. Usually it heads over to RAM, like it is supposed to. But it is also possible to have it do something else. It is easy enough to build a circuit that detects a certain address, then sends an electrical impulse to another part of the circuitry when that happens. If it needs to maintain state, it may go to a flip-flop. In that case, it could totally ignore the data lines and do it's work based upon what happens on the address lines.
I'm fairly certain that would be considered very bad form in this day in age, but it does tend to simplify things.
(When you look at digital electronics, it's possible to do all sorts of weird things. For example, ROMs can do many forms of computation that don't require you to save state.)