Although EEG signals are highly complex,there are still some neural mass models that describe its systematic dynamics.In this paper,we presented an empirical exploration to the theoretical prediction of such a model by fitting the actual EEG signal to the model,with a particular concern on the influence of the extrinsic input p(t) to EEG signal.The results suggest that the neural mass model can produce good approximation to the actual EEG signal,and the mean of estimated input signal fell well within the interval for the simulate study recommended by previous reports.However,it is also found that the variance of the input signal commonly accepted for simulating EEG rhythmic activity is far higher than the value in real data,thus may lead to the generation of unreliable synthetic data,and bias the analysis result.