Efficiently distinguishing the current feedback condition is crucial for the individual to make their following decision. An event-related potential named medial frontal negativity (MFN) has been demonstrated to be sensitive to error and gambling loss. In this study, we conducted the sample gambling game, in which subjects decide to choose one from two cards with different bets (big or small bet), and thereby two critical types of reward features: magnitude (small or large) and valence (loss or gain) were investigated. We analyzed the MFN in different feedback conditions to get insight into the differences in individual behaviors between magnitude and valence. Results of this study demonstrated that the MFN is sensitive to reward valence but insensitive to reward magnitude. Particularly, from the perspective of dynamic functional brain network, differences between small and large magnitudes were uncovered; compared to the large condition, the network pattern related to small reward involved the left central lobe (near electrode C3); whereas the large reward involved the left prefrontal lobe (near electrode Fp1) and right medial temporal lobe (near electrode T8).