論瑣碎性結果與對條件化的限制

On the Triviality Results and the Restriction on Conditionalization

路易士的「瑣碎性結果」論證被視為能有力地拒斥對於條件句採行的真值條件語意論，因為如果接受古典的機率理論、條件化，以及史東內克對條件句的論點，我們可以推導出一個荒謬的結果──對任意的命題A與C而言，p（C/A）=p（C）。本文試圖回應他的「瑣碎性結果」論證，並指出：無論我們以條件機率如何定義自然語言的任何二位連接詞，瑣碎性結果仍會出現。因此，放棄史東內克對條件句的論點不是避開瑣碎性結果的唯一方式；相反的，我們可以對「條件化」做適當的限制來避免瑣碎性結果。

Lewis's arguments for the triviality results are considered as a powerful rejection of the truth-conditional accounts of conditionals: the absurd consequence that for any propositions A and C, p(C/A)=p(C) is derived from the classical probability calculus, conditionalization, and Stalnaker's semantics for conditionals. In this paper, it is argued that the triviality results need not be a threat to Stalnaker's semantics, for we can derive a generalized triviality result from the classical probability theory and any thesis about conditional probability. The lesson, I suggest, is that we should reconsider the classical probability theory or set a restriction on the rule of conditionalization such that the triviality results may be avoided.