Understanding of social interaction processes from the viw point of cognitive theories

这项研究的目的是获得一个人类思维的模型，在这个模型中，思维的各种功能按照应该的方式得到适当和系统的组织。这样的任务无疑是极其困难的，因为众所周知，即使是大脑的一项单一功能也是极其复杂的。然而，应该有人朝着这个方向迈出第一步，否则我们永远无法提出一个完整的思维模式。因此，我们雄心勃勃但不大的目标是首先发现重要的新问题，这些问题需要优先调查，如果我们希望达到上述目标，这些问题必须首先得到解决。考虑到大脑的功能主要是为了应对人类环境而进化的，而人类环境的主要组成部分是其他人，我们决定采用尝试建立一个能够与人进行日常对话的人工智能系统的研究策略，因为日常对话是人类社会互动过程中最具代表性的，需要大脑能够处理的各种技能，如认知、情感、学习等。我们四年的研究总体结果将总结如下：确定了许多新的研究目标，并在理解其本质方面取得了实质性进展。尽管人工智能系统本身还需要几年时间才能完成它的第一个模型。我们最突出的成就包括：交互过程的实验研究，情感和因果推理的理论化，日语会话解析器的开发，知识系统的灵活表示结构，构建会话系统的软件工具，以及一个这样的会话系统的实际实现。

The purpose of this study is to obtain a model of human mind in which the mind's various functions are properly and systematically organized as they should be. Such a task is undoubtedly extremely difficult to accomplish as even a single function of the mind is known to be tremendously complex. However, someone should take the first step toward this direction, since otherwise we should never be able to come up with an integrated model of mind. So our ambitious but modest aim has been first to discover important new problems which require priority in investigation, those which.have to be solved in the first place if we ever wish to reach to above-stated goal. Taking into account the fact that the mind's functions have been evolved primarily for the purpose of coping with human environment and the major constituents of the human environment are other human beings, we decided to employ the research strategy of attempting to build an AI system capable of engaging in daily conversation with people, since daily conversation was the most representative of the human social interaction processes, requiring all kinds of skills the mind could manage, those of cognition, emotion, learning, and so forth.The general outcome of our four-year research will be summarized as follows: Many new targets of research have been identified, and substantial progresses have been attained in understanding their nature, though the AI system itself requires a few more years for the completion of its first model. Examples of the areas where our most prominent achievements have been recorded are: experimental investigation of interaction processes, theorization of emotion and causal inferences, developments of parsers for conversational Japanese, a flexible representation structure for knowledge system, software tools for constructing conversational system, and the actual implementation of one such conversational system.