本研究結合記憶聯結錯誤的實驗典範與「記得/知道」程序，於登錄階段操弄處理層次與聯結性作業，比較重拾記憶(recollection)與熟悉感(familiarity)在不同登錄情境所產生之記憶聯結錯誤所扮演角色，並於記憶提取階段紀錄事件相關腦電位(event-related brain potentials, ERPs)，結合行為與電生理資料探討比較記憶聯結錯誤的「處理歷程解釋」以及「記憶表徵解釋」。實驗1採用受試者間設計，在深層處理（實驗1-1）與淺層處理（實驗1-2）的登錄情境下，操弄聯結性登錄的方式（聯結性處理 vs.非聯結性處理）：深層聯結組的受試者對學習項目進行造句難易度判斷；深層非聯結組則對學習項目進行造詞難易度判斷；淺層聯結組對學習項目中的兩個刺激字進行筆畫比較判斷；淺層非聯結組則需對學習項目中的兩個刺激字分別進行筆畫判斷。在隨後的測驗階段，受試者對舊項目(old item)、聯結項目(conjunction item)、部件項目(feature item)以及新項目(new item) 四類測驗刺激進行「新/舊」判斷，並對做出「舊」反應的項目進行「記得/知道」的判斷。實驗1結果顯示，在深層處理的情況下，聯結性登錄方式影響聯結項目的「記得」反應，此結果與「記憶表徵解釋」的預測相符。而在淺層處理的情況下，聯結性登錄方式則對聯結項目的「記得」與「知道」的判斷均未造成影響，表示聯結性登錄方式對於聯結錯誤引發重拾記憶的影響會隨著處理層次而有所不同。實驗2使用和實驗1-1相同的設計，並同時記錄受試者的電生理訊號。實驗2的行為結果與實驗1-1一致，亦即聯結項目的「記得」反應多寡會受到聯結性登錄方式的影響。然而，實驗2的電生理結果卻與行為結果並不一致：聯結錯誤所引發與重拾記憶相關的頂葉新舊效果，並未受到聯結性登錄方式的調節；而聯結錯誤所引發與熟悉感相關的中額葉新舊效果，卻會受到聯結性登錄的方式所調節，表示「記得/知道」判斷與相關電位新舊效果兩種測量方式在本質上是不相同的。儘管在行為與電生理結果的不一致，「記憶表徵解釋」較能對實驗1與實驗2的結果提出合理解釋。因此，記憶聯結錯誤的發生可能有部分成因來自於登錄階段時刺激部件的結合歷程。This study incorporated memory conjunction error paradigm and Remember/Know procedure to investigate whether depth of processing and associative encoding modulate the false recollection in memory conjunction errors. In Experiment 1 processing depth (deep vs. shallow) and associative encoding (associative vs. non-associative) were manipulated at study. At test, subjects made old/new judgments to old, conjunction, feature, and new testing items. The Remember/know procedure was also employed. It was found that the Remember response to conjunction item was modulated by associative vs. non-associative encoding (Experiment 1-1), a result that was consistent with the Representational account. The manipulation of associative vs. non-associative was however absent under the shallow encoding condition (Experiment 1-2). Experiment 2 adopted the same procedure as Experiment 1-1 and ERPs were recorded at test. The behavioral results shown in Experiment 1-1 were replicated in Experiment 2. However, there were inconsistencies between behavioral and ERP data. In spite of more Remember responses for conjunction items in non-associative condition in comparison to associative condition, the parietal effect thought to index recollection-based recognition was of similar magnitudes for conjunction errors in these two conditions. In contrast, the mid-frontal effect, thought to reflect familiarity processes in recognition, was different for conjunction errors in the two conditions. The discrepancy between Remember/Know data and ERP data suggests that these two measures may not index recollection and familiarity in the same way. Despite this discrepancy, Representation account still provides more reasonable explanation for the results of Experiment 1 and Experiment 2. Consequently, the failure of binding stimuli parts during encoding may be the partial cause for the occurrence of memory conjunction errors.
