The optical emission of GRB 110205A is distinguished by two flares. We examine two possible scenarios for the optical afterglow emission. In the first scenario, the first optical flare is the reverse shock emission of the main outflow and the second one is powered by the prolonged activity of the central engine. However, we find that it is rather hard to reasonably interpret the late （t〉0.1 d） afterglow data unless the GRB efficiency is very high （~0.95）. In the second scenario, the first optical flare is the low energy prompt emission and the second one is the reverse shock of the initial outflow. Within this scenario we can interpret the late afterglow emission self-consistently. The reverse shock region may be weakly magnetized and the decline of the second optical flare may be dominated by the high latitude emission, for which strong polarization evolution accompanying the quick decline is possible, as suggested by Fan et al. in 2008. Time-resolved polarimetry by RINGO2-like polarimeters will directly test our prediction.
