We have discussed EL2 metastable mechanism combining with the theoretical calculations for the EL2 metastable "actuator" performed by Baraff and Schluter. On the basis of the experimental data for the Electronic Paramagnetic Resonance (EPR) "AsGa" photoquench-ings performed by Goltzene et al., we suggest that the AsGaVAsVGa atomic model, in which VAs and VGa are situated in the 1st and 2nd neighboring layers around AsGa antisite atom respectively, can be used for interpreting not only the capacitance but also the EPR "AsGa" photoquenching. This model seems more reasonable than AsGaVGaVAs, in which VGa and VAs are situated in the 2nd and 3rd neighboring layers around AsGa antistite atom respectively, as suggested by Wager and Van Vechten. In addition, our discussion has further supported the identification of the EPR "AsGa" defect ia n-irradiated GaAs as an isolated AsGa antisite atom, and has also shown a slight difference between EL2 defect configurations in LEC and HB GaAs crystals, which would be associated with the EL2 family phenomenon.