The difference is that CPUs, unlike those other machines, can be used to model/simulate things that are similar to brains. There is impedance in the translation, of course, but that impedance can be measured as a sort of “distance” between the architectures; just like one might measure the “distance” between two Instruction Set Architectures.