The requirement to notice differences poses a problem for psychology: The Duplication Problem: The states of two different agencies cannot be compared unless those agencies themselves are virtually identical. Otherwise, the system is swamped by irrelevent differences. Problem is that this appears to require "duplicate brains".
NB: Duplicate brains may be an advantage, not a problem. For instance, from the selectionist standpoint championed by some researchers (such as Changeux in Neuronal Man) you might want lots and lots of structures that are quite similar, but not identical. Is this a problem for Minsky because he is essentially instructionist in his position?