本研究結合來源記憶作業和記得知道程序，比較受試者在不同的登錄以及提取情境下，對於特定與非特定來源訊息的提取以及事件相關腦電位效果的影響，藉此探討特定與非特定來源訊息在記憶中的表徵形式，是符合來源監控架構或是模糊痕跡理論的預測。在實驗一學習階段，刺激字出現的同時由，一男性或女性的聲音將該刺激字唸出，測驗階段中，受試者要在作出性別判斷後，再作一個記得知道的反應。結果顯示正確的來源判斷主要伴隨著記得反應。實驗二使用受試者內設計，操弄學習階段中唸出刺激字的講者數量，分為提供兩男兩女四種聲音的學習四聲音情境，以及提供一男一女兩種聲音的學習二聲音情境，分別要求受試者記憶唸出刺激字的講者或是性別訊息，在測驗階段都要求受試者回憶項目的性別訊息。結果顯示在學習二聲音情境下，受試者的性別回憶正確率，比在學習四聲音情境下高，支持模糊痕跡理論的想法，以細節性痕跡登錄的訊息導致較好的記憶表現。實驗三使用和實驗二類似的設計，並同時記錄受試者的電生理訊號，發現在學習二聲音情境下，受試者引發較大的頂葉新舊效果，代表受試者能夠重拾出較多的性別訊息，然而在學習四聲音情境下，在兩側後方頭皮似乎有個嘗試將講者訊息轉換成性別訊息的Late Negativity效果。實驗四在學習階段提供兩男兩女四種聲音，並要受試者記憶唸出刺激字的講者訊息，測驗階段依照不同的提取需求，區分為需要回憶講者訊息的回憶講者情境，以及只需要回憶性別的回憶性別情境。結果顯示回憶性別情境的來源回憶表現比回憶講者情境好，代表細節性和概要性痕跡的提取，並非是完全獨立的。根據實驗二、實驗三和實驗四的結果，特定與非特定的來源訊息在登錄的過程中分別以細節性和概要性痕跡的形式儲存在記憶中，而兩種記憶痕跡在提取的過程中並非完全獨立。The characteristics of specific (verbatim) and non-specific (gist) information are debated between the Source Monitoring Framework (Johnson, Hashtroudi and Lindsay, 1993) and the Fuzzy-trace theory (Reyna and Brainerd, 1995). The main purpose of the study is to test the representation of specific and non-specific information. Experiment 1 incorporated source memory test and Remember/Know procedure to address this issue. Subjects listened to words spoken by one male voice and one female voice at study, and made source judgments as well as Remember/Know judgments at test. We found that correct source judgments were primarily associated with Remember response. In experiment 2, we increased the variability of source attributes by presenting words spoken by either 4 speakers (2 males and 2 females) or by only 2 speakers (1 male and 1 female). At test, subjects made gender judgments followed by Remember/Know judgments. In accordance to the Fuzzy-trace theory, the gender recognition rate in the 4-voices condition was worse than the 2-voices condition. This result may due to the reliance on gist memory in the 4-voices condition hence decreasing the source accuracy. In experiment 3, we used ERPs under the similar procedure as in experiment 2 and found that the 2-voices condition elicited larger left parietal old/new effect, which is sensitive to the amount of information recollected. However, the Late Negativity effect, which reflects processes that form and retain integrated task-relevant information when needs continued evaluation (Johansson and Mecklinger, 2003) was larger in the 4-voices condition than in the 2-voices condition. Experiment 4 presented words spoken by 4 speakers (2 males and 2 females). At test, subjects made either gender or speaker judgments followed by Remember/Know judgments. The results showed that the retrieval of specific and non-specific information may not be independent. Consequently, the encoding of specific and non-specific information is more likely to accommodate by the Fuzzy-trace theory, but the retrieval of specific and non-specific information is not dissociated.
