作　者：徐英瑾
关键词：儒家; 德性伦理学; 击靶德性论; 神经计算; 隐喻
摘　要：

一般而言，“德性伦理学”就是指通过道德主体的行为倾向，来对其道德输出进行评价的伦理学路数。儒家的道德学说在很大程度上可以被归类为德性伦理学，并完全可以在一个自然主义框架中得到“祛魅化”处理。这种处理的第一步，便是按照扎格泽布斯基的“行为者德性论”与思汪敦的“击靶德性论”的话语方式，厘清儒家德性论的基本框架；而在第二步中，我们不妨深入探讨经由某种可计算的平台来进一步刻画“德性熏养”机制的技术可能性，并由此彻底地将儒家的传统话语方式顺化为当代认知科学的话语方式。文章认为，丘奇兰德的神经计算模型对于德性熏养机制的刻画，将会不可避免地遭遇到“描述层次崩塌”的难题，因此并不适合成为刻画儒家德性伦理学的恰当工具；而一种更符合儒家精神的技术刻画模型，将诉诸种种基于“儒家道德样板库”的隐喻投射机制，使得此类刻画得以“可计算化”的技术平台，则是“非公理化推理系统”。

Confucian Virtue Ethics,Neural Computation and Cognitive Metaphors
 

Xu Yingjin（Fudan University）

Abstract:Virtue ethics is an umbrella label for any positions in normative ethics according to which moral outputs of a moral agent is evaluated in terms of his dispositions of delivering such outputs.Confucianism can be perceived as an instantiation of contemporary virtue ethics for many reasons,especially due to the similarities between Confucianism and Linda Zagzebski’s agent-based virtue ethics or Christine Swanton’s target-based virtue ethics.And Confucianism can be further naturalized as an ethical position which can be fleshed out via some computable model on how an artificial moral agent should work.The first computable model evaluated in this paper is provided by Paul Churchland,who proposes to train a neural computational network to acquire “virtue”.However,such proposal is believed to be defective since it cannot handle the problem of “the collapse of the hierarchy for levels of representations”.A more preferable approach is going to be appealing to the metaphorical/analogical reasoning capacities based on theso-called “corpus on Confucian moral exemplars”,and the technical realization of the whole approach is expected to be based on Pei Wang’s “Non-axiomatic Reasoning System”.
 

Key words:Confucianism virtue ethics; target-based virtue ethics; neural computation; metaphor

[PDF]（下载数：1474）