Intergenerational sustainability dilemma game
We first describe in detail the intergenerational sustainability dilemma game (ISDG). In this game, a chain (which represents a “society”) consists of five distinct generations, each of which comprises of three participants.Footnote 2 Three participants in one generation are required to choose between Option A and Option B (Fig. 1). These options entail the pie (i.e., money) for the generation and, thus, each generation has to discuss and decide how to redistribute it among themselves, in addition to the choice from Options A and B.
An essential feature of the ISDG is that the choice of the current generation affects the size of the next generation’s pie (Fig. 1). Option A brings a larger benefit to the current generation, but it is detrimental to the benefit of the next generation. This is interpreted as exploiting the future generations or refraining from investing in the future. In contrast, Option B brings less benefit to the current generation, but preserves the size of the pie as it is. Therefore, Option B is a sustainable choice. For example, as shown in the last column on the left of Fig. 1, the first generation chooses between obtaining 3600 JPY (Option A) and 2700 JPY (Option B). When the first generation chooses Option A, the second generation’s pie decreases in size by 900 JPY; they have to choose between 2700 JPY (Option A) or 1800 JPY (Option B). In contrast, when the first generation chooses Option B, the size of the second generation’s pie is not affected (i.e., 3600 JPY vs. 2700 JPY). In a similar way, the choice of the second generation affects the size of the third generation’s pie and so on (see Fig. 1).Footnote 3 Thus, all generations obtain 2700 JPY when they continue to choose Option B, but their pies shrink gradually (3600 for the first generation, 2700 for the second, 1800 for the third, etc.) if they continue to choose Option A.
While the equality, utilitarian, and maximin principles suggest that all generations should choose Option B, the self-interested choice of each generation is Option A. Thus, there is a conflict between the intergenerational rationality and the single-generational rationality, like in the well-known prisoner dilemma, where the collective rationality conflicts with the individual rationality. However, the ISDG game differs from the prisoner dilemma on a number of key aspects. First, in the ISGD game, the payoff for people in a given generation is fixed as a function of their own decision; the decisions of future generations do not influence the payoff obtained by the original generation. Consequently, direct reciprocal behavior of between present and future generations is impossible; choosing the sustainable choice cannot be explained by reciprocal altruism (Trivers 1971). Second, each generation can only select Option A or B one time, and are therefore unable to exert influence the decisions of future generations beyond their one selection. Consider that even if the current generation chooses Option B, there is no guarantee that the next generation will also choose Option B, nor is there any way for the current generation to intervene in the next generation’s decision-making process.
There are a few studies that experimentally investigate the sustainability of a resource across generations. Fisher et al. (2004), Hauser et al. (2014) and Sherstyuk et al. (2016) (henceforth, FHS) carried on an experiment of dynamic games across generations, where members of a generation individually deicide their level of consumption of the inter-generational resource. In the FHS models, the larger the consumption of the resource by the members of some generation, the greater their benefit and the worse the situation of the subsequent generations. Thus, similar to the ISDG, past generations unilaterally affect the situation of future generations.
The ISDG has two specific features compared to the FHS models. First, the experimental task the participants work on is simple enough to eliminate the possibility of mistakes or misunderstanding of the participants. In particular, in the ISDG, the participants face a binary choice problem between the sustainable and the self-interested options, while the FHS considers a rather complex dynamic problem with multiple choices, wherein a certain level of cognitive ability is required to understand what the best options are with regard to self interest and total welfare. Second, while people in the same generation should discuss and take a decision as a group in the ISDG, participants in the same generation take individual decisions separately in the FHS, and the combination of their choices determine their own payoff, as well as the situation of the next generation (i.e., how much resources remain in the future). Therefore, participants in the FHS choose considering not only the choices of the future people, but also the choices of others in the same generation. In particular, the over-consumption or the free-riding behavior of members of the same generation becomes important for the sake of sustainability. In contrast, eliminating the effect of intra-generational conflict, the ISDG directly considers the problem of intergenerational resource allocation and focuses on the moral dilemma of the current people between self-interest and sustainability.
Introducing an imaginary future generation
The difficulty associated with a generation’s selection of Option B derives from the inability of future generations to communicate and negotiate with the current generation. The absence of voices from future generations makes it impossible for the current generation to consider their hopes and preferences.
We, thus, suggest introducing a person who acts on behalf of people of the future generation into negotiations (i.e., the imaginary future generation). The imaginary future generation communicates and negotiates with individuals in the current generation, on behalf of the future generation. Note that, because the imaginary future generation is a part of the current generation, their delegate receives the benefit based on the decision of the current generation.
As already mentioned, in the present study, there are two conditions: the treatment and the control condition. In both conditions, three participants made a choice through discussion between Option A and Option B. In the treatment condition, one of the members was told to negotiate with other members as a representative of the later generations, whereas there was no imaginary future generation in the control condition. It is also explained that the payoff of the imaginary future generation is determined by the choice of the current three participants, including this person, and how they allocate the amount of money from their choice among the three. Comparing these conditions, we investigated whether the presence of the imaginary future generation helps people make sustainable choices in the context of an ISDG.
We performed this experiment in two waves, respectively, occurring in February and June of 2014. We recruited subjects from a subject pool based at Kochi University of Technology in Japan. In total, we recruited 210 graduate and undergraduate students (90 in February and 120 in June) to participate in the study.
The data from five generations from twelve chains (N = 180, 55 women, 125 men; mean age = 19.47) were submitted to the analyses reported below. The other 30 participants were assigned to the sixth generations, who only received benefits following the decisions of the former generations. Five chains were assigned to the control condition, whereas seven chains were assigned to the treatment condition.
Upon arriving at the reception desk, they drew a card that indicated which chain and generation they belonged, as well as their identification numbers (i.e., 1, 2, and 3 in the control condition or 3α in the treatment condition).Footnote 4 They then were introduced to separate rooms, depending on whether they were in the treatment or control conditions. In each room, a member of the research team distributed instructions and explained the experimental procedures to participants (see Appendix for the specific instructions). Specifically, participants were told that each generation would make a decision between Option A and Option B and would receive a reward based on their choice. They knew all branches of the game tree (i.e., they saw Fig. 1), but did not know the total number of generations involved in the game. In the treatment condition, participants were also told that one of the three participants (i.e., the person who drew a card indicating 3α) should discuss with other members on behalf of later generations. In the instruction, the role of α participant is explained as follows: “Subject α will negotiate with the other two members of the subgroup, not on behalf of him/herself, but on behalf of the people in the subgroups who follow the current subgroup. However, the reward of Subject α will be determined by how the subgroup allocates its money”. The instructions did not refer to the context of the intergenerational resource allocation problem and did not allude to salient research objectives. For instance, rather than use the word “chain” and “generation” in the instructions, we instead used the word “group” and “subgroup”. After receiving the instructions, the first generations were led to small rooms with respect to each chain, where they engaged in discussions. After arriving at their decisions, participants moved to another room to complete a questionnaire that measured social value orientation (Van Lange et al. 1997) and demographics (e.g., sex and age). Participants then received their payouts and were dismissed. The procedure was repeated five times.
Each generation in a chain used the same discussion room, in order. In each discussion room, there was a research assistant, who handled the flow of subjects (i.e., letting subjects who finished the decision move out and inviting the next participants) and followed the group discussion. The discussion was carried on orally and recorded through a voice recorder. The discussion was required to finish within 10 min; otherwise, the generation’s reward regarding this task would be zero. In the treatment condition, at the beginning of the discussion, subject α (an imaginary future generation) had to inform the other two members that he/she drew the α card.
The group decisions were all written on a whiteboard.Footnote 5 Therefore, subjects were aware of the former generations’ decisions. For example, members of the third generation could see the choices of the first and the second generations in the same chain, like “B, B.” Each generation could not face and communicate with the former generation, as they came into the discussion room only after the former generation moved out. Also, they could not know the decisions of the other chains.
On average, the experiment took approximately 90 min. Since subjects were dismissed right after receiving the payout, the subjects assigned to the first generation were dismissed in 30–40 min, whereas those assigned to the fifth or sixth generation were dismissed in 90 min. For their participation, all subjects received a flat rate of 900 JPY, plus additional money as they decided in the ISDG.
To explore whether and the degree to which the presence of an imaginary future generation influenced the decision-making process, we transcribed all recordings of the negotiations. In total, participants produced 3034 statements.Footnote 6 We employed three coding types. The coding schema is shown in Table 1. Specifically, the coding took into account whether a statement was in support of or against Option A or Option B, neutral between the two, or about payout or not (Coding 1), whether each participant’s final, pre-decision opinion was in support of Option A or Option B (Coding 2), and how the generation’s decision was taken (Coding 3). For each statement (Coding 1), each individual (Coding 2), or each generation (Coding 3), two trained assistants applied a code. When these two coders disagreed on or missed the code to be assigned, one of authors made the determination.