長期記憶情緒脈絡對注意力導引與搜尋歷程的影響

Abstract

近期研究顯示長期記憶可以引導視覺注意力。然而，目前仍不清楚長期記憶中的情緒與中性脈絡在引導知覺和記憶搜尋時是否會有不同的影響。本實驗透過行為實驗（實驗一，24位參與者）及腦電圖實驗（實驗二，30位參與者），檢驗長期記憶中負向脈絡對記憶導引的注意力與長期記憶表現的效果，相對於中性脈絡是否不同。兩項實驗皆包括第一天的學習階段和第二天的測試階段。在學習階段，參與者學習在中性脈絡或負面脈絡場景圖片中嵌入的目標物體位置，及該物體與場景之間的聯結，並重複學習回合五次。第二天，參與者在測試階段完成注意力導引及知覺和記憶搜尋作業，這些作業當中包括新場景和舊場景。實驗二將在測驗階段同時記錄受試者的腦波訊號。行為結果顯示，記憶引導注意力的效果顯著，且場景情緒性對注意力導引的影響在新舊場景圖片有所不同。與中性場景相比，在新場景中，參與者在負面場景中正確偵測目標的反應時間較快、d’分數較高。相反地，在舊場景中偵測目標時，兩種脈絡的行為反應無顯著差異。此外，負面脈絡對新舊場景中的搜尋任務有不同影響。參與者在新場景中搜尋目標時，知覺錯誤在兩種情境中無顯著差異。然而，參與者在舊場景中搜尋目標時，負面場景的記憶錯誤顯著高於中性場景。群體層次的分析顯示，學習階段時搜尋時間的進步幅度與記憶錯誤之間存在顯著的負相關，即學習時搜尋時間的進步越大，記憶錯誤越小。腦電圖結果顯示，負向脈絡對新舊場景中的事件關聯電位有不同的影響。在注意力作業及記憶搜尋作業當中，對於已學習及新場景，在提示出現後的時間窗口，觀察到負面場景相比中性場景在左右後部電極上出現顯著更高的正向振幅。另外，我們也觀察到情緒相關的事件關聯電位LPP的效果在已學習及新場景圖片不同。當受試者受到學習過的場景提示，負向脈絡圖片相對於中性脈絡出現更強的LPP效果；相對地，當受試者受到新場景提示，兩種脈絡在LPP的效果上沒有出現此差異。整體而言，我們的行為結果及腦電圖結果皆顯示負面脈絡在長期記憶導引的注意力及視覺和記憶搜尋中扮演不同角色。

Recent studies have revealed a close relationship between long-term memories (LTM) and visual attention. While these studies demonstrated that LTM can guide attention, it remains unclear whether the emotional context from LTM may have a different impact on attention to guide visual and memory search when compared to the neutral context from LTM. We address this issue in a behavioral experiment (N = 24) and an electroencephalography (EEG) experiment (N = 30). Both experiments consisted of a learning session and a testing session. In the learning session, participants learned the locations of a pre-defined object within neutral or negative scenes and the object-scene associations across five blocks. On the following day, participants performed orienting attention, visual and memory search tasks with new and old scenes in the testing session. EEG data were only recorded on day 2. Our behavioral results with pooled data from two experiments (N = 54) demonstrated the effects of memory-guided orienting attention. The scene context emotionality (negative and neutral) showed different impacts on the scenes that were learned (old) or not (new). Compared to neutral scenes, participants showed faster RTs and higher d’ score when detecting the target in negative scenes in new scenes. On the contrary, when detecting the targets in old scenes, the two contexts did not show different behavioral responses. Moreover, the negative context showed different impacts when searching the target in old and new scenes. When the participants searched the targets within the new scenes, perceptual errors did not show significant differences between two contexts. In contrast, we found a larger memory error for negative scenes compared to neutral scenes when the participants searched the targets within the old scenes in the memory search task. The group-level analysis showed a significant correlation between the search time improvement and memory error. This inverse correlation reflected the larger the learning improvement, the smaller the memory errors. In addition, event-related potential (ERP) results (N = 30) showed the different impacts of emotional contexts on new and old scenes. We observed a significantly higher positive amplitude in response to negative scenes compared to neutral scenes over both left and right posterior electrodes during the post-cueing interval for both old and new scenes on both tasks. Moreover, we observed a larger late positive potential (LPP) for negative scenes compared to neutral scenes, and this effect was stronger when participants were cued with old scenes compared to neutral scenes. Overall, our experiments provide novel evidence that negative and neutral contexts play different roles in LTM-guided attention and search processes.