因果性的反事实条件分析与因果模型理论探析  

Counterfactual Account of Causality and Analysis of Causal Model Theory

作　　者：史红继 张建军[1] SHI Hong-ji;ZHANG Jian-jun(Department of Philosophy,Nanjing University,Nanjing 210023,China)

机构地区：[1]南京大学哲学系,南京210023

出　　处：《自然辩证法研究》2023年第3期46-54,共9页Studies in Dialectics of Nature

基　　金：国家社会科学基金重大项目“广义逻辑悖论的历史发展、理论前沿与跨学科应用研究”(18ZDA031)。

摘　　要：刘易斯开拓了从事件之间的反事实条件依赖关系来定义因果性的研究进路。然而,该进路面临着可能世界相似性标准存在理论困境、因果性定义不恰当等问题。朱亚迪·珀尔将因果性的反事实条件分析运用于人工智能领域的因果推理,提出了因果模型理论,该理论在应对“回溯性”和“共伴性”及多元联结问题上具有相对于刘易斯理论的独特优势。但因果模型理论在刻画假设性思考之选择机制上存在不足。借鉴交互信念可计算理论等新成果,可为因果模型论提供优化方向。David Lewis pioneered the approach of defining causality in terms of counterfactual dependencies between events. However, this approach is faced with the theoretical dilemma of the world similarity standard and the inappropriate definition of causality. Judea Pearl applied the counterfactual condition analysis of causality to causal reasoning in the field of AI, and proposed the causal model theory, which has a unique advantage over Lewis theory in dealing with the problems of “backtracking” and “co-occurrence” and multiple connections. However, the causal model theory is inadequate in describing the selection mechanism of hypothetical thinking. It is possible to optimize causality model theory by referring to new achievements such as mutual belief computability theory.

关 键 词：因果性 反事实 因果影响力 假设性思考 

分 类 号：N031[自然科学总论—科学技术哲学]