基于恐怖谷效应的数字虚拟角色设计研究

Research on Digital Virtual Character Design Based on the Uncanny Valley Effect

摘要: 本研究基于恐怖谷理论观点，探讨数字虚拟角色形象设计中，拟真度高低与原型物种差异对虚拟角色认同感、亲和力和喜爱度的影响。采用量化研究方法，以2 (拟真度高/低) * 2 (是否人类)二因子线上实验法进行数据收集，共收集有效样本342份，使用SPSS进行数据分析。研究结果发现，在数字虚拟角色的认同感上，高拟真度组与低拟真度组没有显著区别，但人类形象组与非人类形象组两者的平均分数有显著差异。针对人类形象角色，高拟真度的数字虚拟角色，其认同感、喜爱度均显著高于低拟真度的数字虚拟角色。然而，针对非人类形象角色，低拟真度的数字虚拟角色，其亲和力、喜爱度均显著高于高拟真度的数字虚拟角色。文章同时测量了恐怖谷效应研究中经常混用的因变量，数据结果表明人们对于认同感、亲和力和喜爱度的理解和评分存在明显区别。研究表明，人们会受到拟真程度高低以及原型物种差异的影响而对数字虚拟角色产生不同的认同感、亲和力和喜爱度感知，此前的恐怖谷效应研究因变量存在混淆使用和模糊表意情况，此实证研究厘清了恐怖谷效应测量变项，同时为数字虚拟角色设计与相关数字产品开发提供理论支持和设计实践建议。

Abstract: Based on the Uncanny Valley theory, this study discusses the influence of verisimilitude degree and prototype species on the acceptance, affinity and likeability of digital virtual characters. Using quantitative research method, data collection was carried out by 2 (high/low verisimilitude degree) * 2 (whether the prototype species is human) two-factor online experiment method. A total of 342 valid samples were collected and SPSS was used for data analysis. The results show that there is no significant difference between the high verisimilitude group and the low verisimilitude group in the acceptance of digital virtual characters, but the average scores of the human group and the non-human group are significantly different. For human image characters, the digital virtual characters with high verisimilitude have significantly higher acceptance and likeability than those with low verisimilitude. However, for non-human image characters, the affinity and likeability of low verisimilitude digital virtual characters are significantly higher than those of high verisimilitude digital virtual characters. At the same time, the paper measures the dependent variables that are often mixed in the study of the Uncanny Valley effect. The data results show that there are obvious differences in people’s understanding and rating of acceptance, affinity and likeability. People will have different feelings of acceptance, affinity and likeability for digital virtual characters due to the verisimilitude degree and the difference of prototype species. Previous studies on the Uncanny Valley effect have always confused the use of dependent variables with vague meanings. This empirical study clarifies the measurement variables of the Uncanny Valley effect, and provides theoretical support and design practice suggestions for digital virtual characters design and related digital product development.