Abstract: Previous reasearch revealed that angry expressions from opponents inversed the differentiation pattern of FRN and diminished feedback P300 difference associated with losses and wins, Moreover, the modulation effect of angry feedback on outcome processing was more salient in females (Chen et al., 2017). But this finding was limited in the effects of supraliminal emotional feedback on outcome processing. In this research, we try to explore whether subliminal emotion feedback can also moderate the monetary outcome processing, and its gender difference.e participants were told that they would play as competitors in a gambling games, that is to say, a loss by the participant means a win by his or her opponent in the same amount, and vice versa. T -hey were told that the computer would determine performer and observer, while the performer made the choice, the observer gave the feedback after seeing the choice. A base payment of ￥40 was given to every participant for gambling. Participants were told that additional rewards or punishments were given based on their performances. The actual earnings for the participants ranges from ￥30 to ￥50. We used an interpersonal gambling game adapted from Gehring and Willoughby gambling task (Gehring & Willoughby, 2002) with detail procedure as follows: after a fixation period(500ms-800ms), the participants were told that the computer would randomly select the performer and the observer in each round of gambling. The person selected as the performer would view numeral 10 or 50 (cents), and make a choice by pressing the corresponding button as soon as possible. The choice was presented for 300ms with the rectangle compassed the choice becoming red. The participants were not told about their opponents’ emotion feedbacks, which were presented along with utilitarian cues for 16 ms on the forehead of the observer, and then changed to neutral face with utilitarian cues for 1000 ms. Likewise, the participants were required to send utilitarian cues on the forehead of their neutral faces, the “＋” on the forehead of the observer indicated that the performer won the points whereas a “－” lost the points,if they were chosen as observer. After the task, the participants were asked whether they could see emotional facial expressions. Nobody reported seeing emotional faces during the experiments. The whole experiment consisted of 384 trials divided into 8 blocks. In each block, the participant was selected as performer two thirds of the trials, while the confederate was selected as performer the remaining third. The behavior data showed the participants selected more high-risk options after losses (.59 ± .04) than after wins (.42 ± .03), [F(1,33) = 32.55, p＜.001, n_p^2=.50]. However, no significant main effect of emotion or any interaction involving emotion was observed [Fs(1, 33)＞2.71, p=.11] (See Figure 4 A). The analysis on reaction times showed a significant interactions of emotion×outcome valence [F(1,33) = 4.25, p＜.05, n_p^2=.11]. Simple effect analysis indicated that the reaction times were significantly longer after wins (841±32 ms) than after losses (793±29 ms, p<.05) when accompanied by angry expressions. But no significant difference was observed after wins (823±28 ms) and losses (821± 31 ms, p = .95) when accompanied by happy expressions (See Figure 4 b). The analysis of FRN amplitudes showed a main effect of outcome (lossee: 4.13±0.81μV,wins: 4.85±0.94μV, F(1,33)=7.74, p＜0.01, η2p=0.19), a marginal significant interaction of emotion×gender [F(1,33)=3.13, p=0.08, η2p =0.09]. Simple effects analysis indicated that FRN amplitudes were more negative for angry feedbacks (4.41±1.20μV) than for happy feedbacks (5.02±1.24 μV, p=0.03) in females, but no significant difference between angry (4.30±1.23) and happy (4.23±1.28 μV, p=.79) feedbacks in males. In P300, there was a significant main effect of outcome valence [F(1,33) = 20.9, p＜0.001, η2p =0.39], with larger P300 for wins (9.38±0.61μV) than losses (8.04±0.57 μV), a marginal significant interaction of outcome×gender [F(1,33) =3.30, p=0.07, η2p=0.09], Simple effects analysis indicated that P300 amplitudes were more positive for wins (10.79±0.84 μV)than for losses (8.91±0.80 μV, p<0.001) in females, but only marginal significant difference between wins (7.98±0.87 μV)and losse (7.17±0.82 μV, p=.06) feedbacks in males. Our study suggests that the subliminal emotional feedback may also influence the decision-making behavior and feedback processing of individuals, this effect mainly displays in the early phase of feedback processing which appraising motivtional significance of feedbacks autonomically, and impact on women more intensely.

Key words: Subliminal emotional feedback, risk decision, ERPs, interpersonal emotion

摘要： 本研究用互动的决策游戏和脑电技术考察阈下情绪反馈对结果加工的影响。结果表明输钱相对于赢钱诱发更大的FRN/RewP和更小的P300，同时阈下愤怒反馈可以延缓赢钱之后下一次选择的反应时，并在女性被试中诱发更负的FRN/RewP, 但对反馈相关P300没有显著影响。这些结果说明阈下情绪反馈可影响决策结果评价早期的对刺激动机性意义的快速评估，但对晚期的深度评估没有显著影响，且这种影响在女性被试中更为明显。

关键词: 阈下情绪, 风险决策, ERPs, 人际情绪