機率證據與非常態證成

Probabilistic Evidence and Non-normical Justification

史密斯 （Martin Smith, 2016） 藉由著名的樂透悖論，指出「證成的最小化風險概念」以及「證成的多重前提封閉原則」兩者是不相容的。前者容許命題被機率證據所證成，後者則是關於證成性在信念間的轉換。同時接受兩者將導致我們擁有一對衝突但都被證成的信念。為了避免這樣的困境，以及更符應我們的日常直覺，史密斯選擇保留「證成的多重前提封閉原則」，並提出「證成的常態性理論」來取代「證成的最小化風險概念」。在這篇文章中，我首先論證史密斯的解決方案具有內在衝突而不令人滿意，接著，我將從拒絕「證成的多重前提封閉原則」出發，回應樂透悖論以及其他相關困難。

Smith (2016) holds that, by means of the lottery paradox, the risk minimization conception of justification is not compatible with the principle of multiple premise closure of justification. While the former allows justification to come from probabilistic evidence, the latter concerns the transition of justification among beliefs. Accepting both principles will then result in the inconsistency of two justified beliefs. To avoid this inconsistency and at the same time conform to our ordinary intuitions of justified beliefs, Smith holds on to the principle of multiple premise closure of justification, but replaces the other principle with the Normic Theory of Justification. In this paper, I will first argue that Smith’s proposal is still unsatisfactory for it has internal inconsistency. Second, I will further propose an alternative account to avoid the inconsistency from the lottery paradox based on rejecting the multiple premise closure of justification.