摘要：
所谓人工情绪研究，就是指在人工智能研究所提供的技术环境内，以计算机工程可以实现的方式，去模拟人类情绪的部分运作机制，以便提高整个人工智能产品的工作效率。之所以说人工情绪对于一种完整意义上的人工智能研究来说是必不可少的，是因为在人工智能体自身运作资源（包含运作信息）有限的情况下，“情绪”可以为智能体对外部环境的应激反应提供快速通道。然而，主流的认知科学所提供的情绪模型（如OCC模型）的基本思路，是将情绪视为对于智能体所处现状与所求目标之间差值的评估所形成的衍生物，并没有像胡塞尔的现象学所期望的那样，提供对于使得相关目标得以产生的“生活世界”的刻画。同时，海德格尔的“向死而生”理论对于智能体自身的唯一性的提点，也没有得到人工情绪研究者的重视。从这个意义上说，欧陆现象学对于情绪的研究成果，远未被当下的人工情绪研究所消化。
关键词:
人工情绪,
人工智能,
现象学,
生活世界,
向死而生,
OCC模型
Abstract:
Artificial emotion (AE) refers to the modelling of emotional mechanisms within appropriate computing environments, and the resultant systems are expected to be able to efficiently respond to environmental contingencies. The inquiry into AE is indispensable for artificial intelligence (AI) in the sense that emotions play a critical role in the process of forming stress reactions, which are indispensable for producing appropriate behaviors, especially when the agent is constrained by limited time budget and relevant inputting information. However, according to the model of emotion provided by the mainstream cognitive scientists, especially the “OCC-model”, emotion is viewed as something derived from the agent’s evaluation of the gap between its current state and its goal, and hence lacking a reference to a wider web of beliefs underpinning such evaluations. And the necessity of referring to this deeper web is addressed by phenomenologist Husserl in this theory of lebenswelt. Moreover, Heidegger’s notion of “being-towards death”, which indicates the uniqueness of very existence of agents’ operating histories, is also widely neglected by mainstream AE-studies. Hence, AE-researchers fail to do justice to phenomenologists’ treatment of emotions.
Key words:
artificial emotion,
artificial intelligence,
phenomenology,
lebenswelt,
being-towards-death,
OCC-model
徐英瑾. 欧陆现象学对人工情绪研究的挑战[J]. 探索与争鸣, 2019(10): 137-147.
Xu Yingjin. The Challenges that Continental Phenomenology Poses to Artificial Emotion[J]. Exploration and Free Views, 2019(10): 137-147.