Program in Applied and Computational Mathematics, Princeton University, Princeton, NJ 08544;
Annenberg School for Communication, University of Pennsylvania, Philadelphia, PA 19104; and.
Proc Natl Acad Sci U S A. 2021 Dec 14;118(50). doi: 10.1073/pnas.2102148118.
Political theorists have long argued that enlarging the political sphere to include a greater diversity of interests would cure the ills of factions in a pluralistic society. While the scope of politics has expanded dramatically over the past 75 y, polarization is markedly worse. Motivated by this paradox, we take a bottom-up approach to explore how partisan individual-level dynamics in a diverse (multidimensional) issue space can shape collective-level factionalization via an emergent dimensionality reduction. We extend a model of cultural evolution grounded in evolutionary game theory, in which individuals accumulate benefits through pairwise interactions and imitate (or learn) the strategies of successful others. The degree of partisanship determines the likelihood of learning from individuals of the opposite party. This approach captures the coupling between individual behavior, partisan-mediated opinion dynamics, and an interaction network that changes endogenously according to the evolving interests of individuals. We find that while expanding the diversity of interests can indeed improve both individual and collective outcomes, increasingly high partisan bias promotes a reduction in issue dimensionality via party-based assortment that leads to increasing polarization. When party bias becomes extreme, it also boosts interindividual cooperation, thereby further entrenching extreme polarization and creating a tug-of-war between individual cooperation and societal cohesion. These dangers of extreme partisanship are highest when individuals' interests and opinions are heavily shaped by peers and there is little independent exploration. Overall, our findings highlight the urgency to study polarization in a coupled, multilevel context.
政治理论家们长期以来一直认为，扩大政治领域，纳入更多元的利益，将治愈多元化社会中派系的弊病。尽管过去 75 年来政治范围已经大大扩大，但极化现象却明显恶化。出于这种悖论，我们采取自下而上的方法，通过新兴的维度降低来探索多元（多维）问题空间中党派个体动态如何通过集体层面的派系化来塑造。我们扩展了基于进化博弈论的文化进化模型，其中个体通过成对交互积累收益，并模仿（或学习）成功他人的策略。党派偏见的程度决定了从对方党派个体学习的可能性。这种方法捕捉到了个体行为、党派介导的意见动态以及根据个体不断变化的利益而内生变化的交互网络之间的耦合。我们发现，虽然扩大利益的多样性确实可以改善个人和集体的结果，但越来越高的党派偏见通过基于党派的分类促进了问题维度的减少，从而导致极化程度的增加。当党派偏见变得极端时，它还会促进个体之间的合作，从而进一步加剧极端极化，并在个体合作和社会凝聚力之间造成拉锯战。当个人的利益和意见受到同伴的强烈影响，而独立探索的机会很少时，极端党派偏见的这些危险就最高。总的来说，我们的研究结果强调了在耦合的多层次背景下研究极化的紧迫性。
