Abstract
Feeling of Knowing (FOK) judgments refer to the feelings an individual has regarding his or her knowledge for a specific subject, more specifically whether or not that knowledge exists within memory. And the study of accuracy of FOK is to evaluate how accurate the feeling reflect the actual presence or absence within memory. Corresponding to the recall-judgment-recognition paradigm, psychologists used the statistical testing, the correlation coefficients, and the conditional probabilities as the index of the accuracy of FOK. However, none of these indexes can purely and stably reflect the accuracy of FOK. I review and analysis the deficits of these indexes from logical, methodological, and cognitive processing viewpoints, and propose a new probability model. In this model, the accuracy of FOK is viewed as a parameter of a latent variable. According to the meta-analysis, this parameter can differentiate normal and amnestic patient subject, and also duplicate several past researches. It shows that the modeling technic is a more proper way for analysis the accuracy of FOK. 「知感」是一種監控記憶的能力，意指人類對於其無法回憶的事物仍能有一種自己是否知道的感覺，然而這種感覺究竟能夠多麼準確地反映記憶項目是否存在記憶當中，就是「知感精確度」的主要研究議題。針對最常使用的回憶¬－判斷－再認實驗派典，心理學家曾經利用統計檢定、相關係數、條件機率等做為知感精確度的測量指標，但這些指標皆不能純粹地、穩定地反映知感精確度的概念。本文從邏輯、方法學與認知歷程的角度回顧這些指標的缺陷，並提出一個新的機率參數模型，將知感精確度視為模型中的一個潛在變項加以估計。根據以過去文獻資料進行統合分析的結果顯示，參數值可以有效區辨正常人與失憶症病人，並複驗多項知感研究的結果，這是一種較為合適的分析方法。
