逻辑与形而上学系列讲座第五讲丨Andreas Herzig:信念修正中的不相关性与词汇保持性

信念修正中的不相关性与词汇保持性

主题
信念修正中的不相关性与词汇保持性
活动时间
-
活动地址
锡昌堂727
主讲人
Andreas Herzig
主持人
马明辉

讲座信息

主讲人:Andreas Herzig

法国国家科学研究中心(CNRS)研究员、法国保罗·萨巴捷大学图卢兹计算机科学研究所(IRIT)教授、欧洲人工智能学会会士

主持人:马明辉 中山大学哲学系教授

时间:2026年4月3日 10:00-12:00

地点:中山大学锡昌堂727会议室

 

主讲人简介

Andreas Herzig is a researcher of the French National Research Center(CNRS) at the Toulouse Computer Science Research Institute (IRIT) of Université Paul Sabatier. He is a fellow of the Eur. Association for AI (EurAI). He is the Editor-in-Chief of the Journal Applied Non-Classical Logics (since 2015) and a member of the editorial Board of the J. of Philosophical Logic (since 2008). He was an associated editor of Artificial Intelligence (2017-2022). He was the head of IRIT’s AI Department (2009-2020) and of the Committee on AI and Data Science of the French National Research Agency (ANR, 2022-2025).

Andreas Herzig 是法国国家科学研究中心(CNRS)研究员,任职于法国保罗·萨巴捷大学(Université Paul Sabatier)图卢兹计算机科学研究所(IRIT)。他是欧洲人工智能协会(EurAI)会士。自2015年起担任《应用非经典逻辑》(Applied Non-Classical Logics)主编,自2008年起担任《哲学逻辑杂志》(Journal of Philosophical Logic)编委会成员。2017年至2022年期间,他曾任《人工智能》(Artificial Intelligence)副主编。2009年至2020年期间,他曾任IRIT人工智能部主任;2022年至2025年期间,任法国国家科研署(ANR)人工智能与数据科学委员会主任。

 

讲座简介

Two different notions of irrelevance have been much debated in the belief revision literature: irrelevance of syntax and irrelevance of a piece of information. The former is a basic postulate of the AGM framework but has been criticised as being too strong. The latter may be violated by AGM revision operations, and Parikh (followed by, among others, Makinson and Peppas) has argued that such a postulate should be added to the AGM framework. I counter the most important arguments against syntax irrelevance by showing that the most striking counterexamples can be accommodated if classical material implication is replaced by strict implication. I then show how this approach naturally leads to an account of revision operations for modal logics. I finally relate syntax irrelevance to an intuitively appealing postulate of vocabulary preservation that is strictly weaker than Parikh's and may also be violated by AGM revision operations.

在信念修正的相关文献中,两种不同的不相关性概念一直争论颇多:其一是句法的不相关性,其二是一条信息的不相关性。前者是AGM框架的一条基本公设,但它被批评为过于强。后者则可能被AGM修正算子所违反,而Parikh以及其后的Makinson、Peppas等人主张应将这样一条公设加入AGM框架之中。

本讲座通过表明如下事实,来反驳针对句法不相关性的最重要论证:如果将经典的实质蕴涵替换为严格蕴涵,那么那些最突出的反例都可以被容纳。并说明这一思路如何自然地导向对模态逻辑中的修正算子的一种说明。最后,将句法不相关性与一条在直觉上颇具吸引力的词汇保持公设联系起来;这条公设严格弱 Parikh的公设,但也同样可能被AGM修正算子所违反。