NARX MIMO feedback identification

Hi.

If my NARX has 4 inputs & 4 independent targets, W X Y Z & A B C D. The targets are each a function of only the 4 inputs W X Y Z. Will NARX recognize that the 4 targets are independent when they are fed back into the input layer via the feedback line, i.e. will target A be recognized as an input for target B, instead of just inputs W X Y Z?

[Greg Heath]

You seem to be implying that if

 x = f(a,b) and y = g(a,b)

that x and y are independent.

Not so.

Greg