当“形似真人”适得其反：虚拟主播拟真度与期望违背效应

When “Looking Human” Backfires: Virtual Streamer Realism and Expectation Violations

摘要: 人工智能技术正在重塑传统的电商直播，虚拟主播为直播营销注入新活力。然而，现有研究关于虚拟主播形态和行为拟真度如何营销消费者购买意愿，缺乏系统的理论框架和实证支持。本研究基于期望违背理论，提出形态锚定期望，行为提供证据的认知评价模型。通过2 × 2的情境模拟实验，考察不同拟真度组合对购买意愿的影响。研究发现，形态和行为拟真度存在交互效应，在高形态拟真度下，低行为表现违背了由高形态设定的高期望基准，引发负向能力期望违背，导致购买意愿降低；反之，在低形态拟真度下，高行为表现因超预期的互动能力触发正向违背，提升购买意愿。中介效应检验证实，能力期望违背是连接拟真度特征和购买决策的核心心理机制。本研究揭示了拟真度效应的非单调性，为企业提出重行为表现的策略，以规避盲目追求超写实外观引发的高期望陷阱。

Abstract: Artificial intelligence technology is reshaping traditional e-commerce live streaming, with virtual streamers becoming the new norm in marketing. However, existing research lacks systematic theoretical frameworks and empirical evidence on how the form realism and behavioral realism of virtual streamers influence consumers’ purchase intentions. Based on expectancy violation theory, this study proposes a cognitive evaluation model in which form anchors expectations while behavior provides evidence. Through a 2 × 2 scenario-based experiment, the study examines the impact of different realism combinations on purchase intentions. The findings indicate a significant interaction between form and behavioral realism. Under high form realism, low behavioral performance violates the high expectations set by the form, triggering negative ability expectancy violations and reducing purchase intentions. Conversely, under low form realism, high behavioral performance exceeds expectations, leading to positive violations and increased purchase intentions. Mediation analysis confirms that ability expectancy violation is the core psychological mechanism linking realism features to purchase decisions. This study reveals the non-monotonic nature of the realism effect and advises firms to prioritize behavioral performance to avoid the high-expectation trap caused by blindly pursuing hyper-realistic appearances.