Scope's effect on simulink model operation

Hi, everyone. I just have this strange problem in simulink simulation. When I added a scope to observe the time domain waveform of an internal nodes, the signal on that node changed, and following stages give erroneous results. I have done some experiments: just duplicate the designed systems, and use scope to observe the internal nodes and final output of one system, while for other system just observe the final output. And the two results is different!! It seems the Quantum mechanics is affecting Matlab/Simulink: the act of observation affect the observed object. Does anyone have a clue on this?