Figure 1

Experimental Design. (A) Each participant was required to choose one of three redistribution rules: preserving the status quo of inequality, advancing intermediate redistribution, or achieving absolute equality. A total of 25% of the transferred income was lost as a cost. The equality rule incurred the largest cost, whereas the inequality rule was cost-free. During the games, these rules were termed Rules 1, 2, and 3, respectively, to avoid generating a decision bias. (B) When making a decision behind the VoI, class did not matter; thus, the rule was chosen in terms of equality. When the classes were “informed”, the experimental conditions were distinguished in terms of high-, middle-, and low-income classes. Therefore, we distinguished the decisions as an equality or non-equality decision in the VoI and a selfless or selfish decision in high- and low-income classes, whereas decisions were neutral in the middle class. (C) On repeated runs of each session (the Informed and VoI conditions), the participant chose a rule (red-boxed screen) and was shown the payoff that corresponded to her assigned class along with the rule chosen at the society level. Earnings from each round accumulated to a total income that was revealed at the end of game. Participants were told that the total income determined the payment for completion of the experiment. Before and after playing each session, participants were shown a photo of each partner and were asked to rate how much they (dis)liked each partner on a thermometric scale, i.e., a feeling thermometer (Affective Feeling I and II, respectively). The change in affection after playing was calculated by subtracting Affective Feeling I from Affective Feeling II. Next, we disclosed their own and their partners’ incomes over the entire session and asked the participant to again rate their feelings for their partners. We subtracted the rating immediately after playing (Affective Feeling II) from that after the disclosure of income (Affective Feeling III) and used it as a measure, namely, the change in affection after income disclosure. The relative income was calculated by dividing the participants’ total income by the partners’ income. In each session, a lottery game with the same format was randomly inserted. The redistribution-specific activation was calculated by subtracting the activation evident when drawing the lottery from that observed upon a redistributive decision.