文献类型：期刊文献

英文题名：Asymmetric biased assimilation in the Deffuant-Weisbuch model: Stability, phase behavior, and network effects

作者：Peng, Haotian[1];Zhou, Xijin[2];Jin, Li[1]

机构：[1]Shanghai Jiao Tong Univ, Global Coll, Shanghai, Peoples R China;[2]East China Univ Sci & Technol, Sch Informat Sci & Engn, Shanghai, Peoples R China

年份：2026

卷号：700

外文期刊名：PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS

收录：;EI(收录号：20263421341676);Scopus(收录号：2-s2.0-105047817924);WOS:【SCI-EXPANDED(收录号:WOS:001857107600001)】；

基金：This work was in part supported by the National Natural Science Foundation of China under Grant 62473250.

语种：英文

外文关键词：Opinion dynamics; Biased assimilation; Asymmetric bias; Stability analysis; Social networks; Opinion polarization

摘要：Biased assimilation is a psychological mechanism behind opinion polarization, but most existing models treat bias toward the two opinion extremes symmetrically. We introduce an asymmetric biased-assimilation Deffuant-Weisbuch (DW) model on social networks in which each agent can have different bias strengths toward extreme opinions 0 and 1. For the main equilibrium families, we establish local stability and stochastic convergence results. Extreme equilibria are locally stable under positive relevant boundary biases and become locally exponentially convergent when the same-extreme active graph covers all agents. Intermediate equilibria with negative-bias parameters are locally contractive at the pairwise level and locally exponentially convergent on connected networks within a local all-active neighborhood. Matched baseline simulations show that asymmetric biased assimilation can generate directional dominance beyond the polarization produced by classical and symmetric biased DW dynamics. We then map the phase behavior on Erd & odblac;s-R & eacute;nyi, Watts-Strogatz, and Barab & aacute;si-Albert networks. The results show how confidence threshold, bias asymmetry, open-mindedness, and topology shape polarization, fragmentation, directional drift, and neutral-phase formation. The analysis and simulations together provide a compact framework for studying how asymmetric evidence processing organizes macroscopic opinion patterns in bounded-confidence social networks.