I know muscle confusion has no scientific basis, but I am starting to wonder if it is a good way to combat over training and CNS fatigue.

I heard of a routine but cant remember what it is called that required you to change exercises every 4 weeks.

People have explained this concept as simply muscle confusion which sounds like nonsense. But perhaps it is worth looking into from a different angle.
When you change exercises your CNS needs to learn the movement again, for example if your only doing DB bench for chest for a couple of months and you overtrain and cant gain anymore due to accumulative CNS fatigue, and then you switch to cable cross over, you start making gains again, although this is not muscle gains but just the CNS adapting to the movement for a few weeks, which could be considered time wasted and hence why "muscle confusion" is considered debunked.

But lets say you alternate within a week or 2, not months and not within a workout, DB bench with cable crossover with decline BB bench on a 2x or 3x a week routine. Your body never has a chance to detrain from the movement because it takes 5-12 days for the CNS to recover, and because its not the same repetitive movements on the CNS each workout, it SHOULD based on logic not science or assumption, maximize muscle stimulation through lessening the fatigue on the CNS.

Now you could say this is already implemented into once a week routines, for example, following a compound movement such as bench press with an isolation movement like pec deck or DB flyes, but the reason for this is to continue the sets on the larger muscle group(pecs) while not allowing the smaller muscle groups(anterior delt and triceps) to prematurely fatigue.
This theory does not have to do with big muscle groups or small muscle groups, its simply lessening one of the barriers to muscle stimulation which is cumulative CNS fatigue over a period of weeks.

I am wondering if anyone has experience with this, or if this is implemented in some of the popular routines and perhaps been explained in a similar way as I have explained it, or better yet implemented it in a similar fashion as to not allow CNS detraining of the movement.