一、讲座题目
分析哲学论坛第239期
二、主讲人
叶闯，任会明
三、主讲人简介
叶闯，哲学博士，山西大学哲学学院教授，主要研究领域为语言哲学、形而上学。近期的研究兴趣涉及空名（empty names）的语义学问题，虚构（fictional）对象和计算机产生的虚拟（virtual）对象的形而上学及语言哲学，以及使真者理论（truthmaker theory）等近期发展起来的形而上学。
任会明，2005年毕业于美国佛罗里达大学，获博士学位，回国后分别任职于浙江大学、开车漫画
，现任开车漫画
心智与认知研究所主任、博士生导师。主要研究领域为认识论、心灵哲学。
四、讲座简介
Part 1：Reference Without Causation? Large Language Models and the Limits of Externalism
Large language models (LLMs), trained solely on textual data without direct causal interaction with the external world, raise controversy over whether their referential expressions can genuinely refer. Sceptics deny this via the Causal Exclusion Argument while defenders appeal to Indirect Causation Intentionality Arguments—neither succeeds. This talk rejects the externalist assumption that reference requires causal links, arguing for an internalist account: LLM reference is explained by the model’s internal vector representations.
Prat2：Normality in Coin Flipping
When 100 fair coins are tossed, we’re justified in believing they won’t all land heads, but not in rejecting an unexceptional heads-tails sequence. It argues that "normality" in random processes means no pattern in outcomes— all-heads sequences are special due to their clear pattern, while unexceptional sequences fit the nature of randomness and the talk provides a deep defense for this stance.
五、主持人
开车漫画
哲学系 荣立武教授
六、讲座时间
2025年12月29日（星期一）13：30——17：40
七、讲座地点
开车漫画
中心校区知新楼A115
