Abstract：Human beings are surrounded by social relationships often described by social distance. Generally speaking, our strongest social relationships are with those who are “closest” to us, and individuals with weaker relationships are more “distant”. Social discounting refers to the phenomenon that the subjective value of another person’s gain or loss decreases as the social distance increases. Different from temporal discounting and probability discounting which have received more attention in the realm of behavioral decision making, very few researches on social discounting have been conducted. In 2006, Jones and Rachlin studied this phenomenon for the first time under the frame of “I will forgo some gains in order to let another person obtain $75”. It was found that the amount of money an individual was willing to sacrifice declined as the social distance increased. That study also revealed that the function of social discounting fitted the hyperbolic model. However, there may exist some other frames in our daily life. Therefore, in the present research, the phenomenon of social discounting is explored in four frames, including “I will forgo some gains in order to let another person obtain ¥100” (NG-G), “I will forgo some gains for another person to avoid losing ¥100” (NG-NL), “I will suffer some losses in order to let another person obtain ¥100” (L-G) and “I will suffer some losses for another person to avoid losing ¥100” (L-NL). Meanwhile, this research also explored the social discounting function in the context of Chinese culture. Four experiments, including two pre-experiments and two formal experiments, were conducted. In the first pre-experiment with 117 undergraduate and post-graduate subjects, social distances between oneself and mother, a friend and a mere acquaintance were measured. The results show that the social distance between one and his/her mother is the shortest, while that between one and a mere acquaintance is the longest. Some Chinese subjects in the first pre-experiment could not understand the choice of the traditional titration procedure well, which was used by Jones and Rachlin in 2006. So a new situational choice titration procedure was tested in the second pre-experiment. The results indicate that the new titration procedure is more understandable for Chinese subjects, and therefore, is more applicable to this research. Based on the pre-experiments, two formal experiments were carried out. The first one compared social discounting between the frames of NG-G and NG-NL, and the second made comparison between the frames of L-G and L-NL. In both experiments, we conducted a 4 social distance (mother, friend, a mere acquaintance and a stranger) ×2 frames (NG-G and NG-NL in the first experiment, L-G and L-NL in the second experiment) within-subject design. In addition, the subjects’ altruism was also measured by Self-report Altruism Scale. Valid subjects recruited in the two experiments were respectively 78 and 77. The results show that: (1) in all of the four frames, exponential model is more suitable for describing social discounting than hyperbolic model among Chinese subjects; (2) task frames influence social discounting degree, which is lower in frames of NG-NL and L-NL than in frames of NG-G and L-G; (3) the interaction between social distance and task frames also has impact on social discounting degree. As social distance increases, the difference between social discounting degree in NG-G and NG-NL frame becomes smaller. This trend also exists in frames of L-G and L-NL; and (4) the interaction between social distance and altruism influences social discounting degree. Only when the social distance is long enough is there an obvious difference in discounting degree between subjects with high altruism score and those with low. In conclusion, this research suggests that people not only avoid his/her own losses, but also try to avoid others’.