合理性推論的兩個準則

Two Criteria of Reasonable Inferences

亞當斯和史東內克在傳統的有效性觀點之外，對指示條件句的推論提出了合理性這樣的概念，並對這樣的概念提出了兩個不同的準則。本文從可斷說性的概念分別來對兩者提出形式上的刻畫，以檢視這兩者的關係。本文把亞當斯支持的準則稱為「可斷說性的證成性」，而把史東內克支持的準則稱為「嚴格的證成性」，並認為這兩個準則可以幫助我們釐清條件句推論中的爭議。本文用這兩個準則來重新分析麥基對肯定前件律所提出的反例，試圖說明為何麥基提出的是一個可斷說證成性的反例，而不是嚴格證成性的反例。接著，本文對這個現象提出一個診斷，說明為何這兩個準則只會在條件句的推論中產生實質的分歧。最後，本文論證可斷說性的證成性太過於嚴格而難以成立，主張嚴格的證成性才是一個較好的準則。

Validity is an orthodox way to distinguish between good and bad inferences, which says that when inferences are valid, we can derive true conclusions from true premises. But when inferences involve conditionals, validity does not seem to be a good criterion. Unsatisfied with the traditional view of validity, Adams(1965) and Stalnaker(1975) propose their modified criteria, which are called “reasonable inference” in this paper, to evaluate inferences involving conditionals. In this paper, Adams’ criterion is called “justification of assertibility”, which can derive a highly assertible conclusion from highly assertible premises. Stalnaker’s criterion is called “justification of strictness” in this paper, which can derive a strictly assertible conclusion from strictly assertible premises.This paper examines these two criteria and their consequences. First, by reexamining McGee’s(1985) counterexample to modus ponens, I argue that McGee’s example is an assertibility-justified but not a strictness-justified counterexample to modus ponens. Second, I explain why these two criteria have such a substantial divergence when inferences have conditional conclusions. Third, if my diagnosis is correct and complicated conditionals are allowed in inferences, then the justification of assertibility is too rigid to be satisfied. I conclude that the justification of strictness is a better criterion than the justification of assertibility.