工作記憶作業中的干擾程度和干擾比例對 alpha 振盪之影響

Abstract

alpha波為人類大腦8Hz至12Hz的神經振盪活動，其振盪的增強可能反映神經系統抑制無關訊息的歷程，同時也涉及對工作記憶的保護，故許多工作記憶研究常透過alpha波來切入（Jensen& Mazaheri, 2010; Pfurtscheller et al., 1996）。Bonnefond與Jensen（2012）為了瞭解alpha波抑制工作記憶中干擾的機制，透過視覺通道呈現改編的Sternberg範式，發現到強干擾的發生會伴隨alpha振盪的增強。但後續使用類似架構的研究，卻發現干擾程度對alpha波的影響不一致，強干擾的發生除了可能伴隨振盪的增強，卻也可能造成振盪的減弱（Weisz et al., 2020）或是無顯著影響（Sghirripa et al., 2021）。本研究推論可能是因為這些後續實驗設計在調整後出現干擾刺激和欲記憶刺激間的時間間隔，讓看似相同的實驗程序藏有潛在的混淆因素。除此之外，過去研究因干擾比例在操弄上的複雜性，而少去討論干擾比例對alpha波的影響。本研究欲從概念上改編Bonnefond與Jensen的Sternberg範式，操弄工作記憶作業中的干擾程度和干擾比例，提出一個新的實驗程序。本研究以登錄階段須忽略刺激的採樣點是否被打亂來操弄干擾程度（包含強干擾和弱干擾2個水準），利用登錄階段欲記憶刺激和須忽略刺激之比例來控制干擾比例（包含高干擾比例和低干擾比例2個水準），為一2×2二因子的實驗設計。22位參與者需記憶登錄階段的欲記憶刺激並忽略干擾，再於提取階段以按鍵判斷目標刺激是否為登錄階段的欲記憶刺激之一。結果發現在低干擾比例下，額葉、中央腦區、和枕葉的alpha波在強干擾出現前後增強，顯示出抑制強烈無關訊息時的alpha波可能同時反映了預備性抑制和反應性抑制。但該效果在高干擾比例下消失，推測是因高干擾比例促使alpha波變化達天花板效應，顯示出alpha波因應工作記憶作業中干擾時背後的複雜性。Alpha oscillations (8-12 Hz) are one of the most dominant rhythm in human electroencephalography (EEG). It is proposed to serve the function of inhibition to protect items in working memory from distracting information (Pfurtscheller et al., 1996; Jensen& Mazaheri, 2010). In a modified Sternberg paradigm, Bonnefond and Jensen (2012) found that alpha power increased at the anticipation of strong compared to weak distractors, reflecting the active gating of distracting information from interfering with the memory trace. However, there was a lack of evidence supporting the inhibition account of alpha oscillations in later studies using similar experimental design. While some research found that alpha power decreased at the anticipation of strong compared to weak distractors (Weisz et al., 2020), some research reported null results (Sghirripa et al., 2021). We speculated that the discrepancy might be related to the greater temporal disparity between the encoding phase and the presentation of the distractors in these later studies, which might have dampened the demands for inhibition. To conceptually replicate the study of Bonnefond and Jensen, here we designed a modified Sternberg paradigm where distractors were sandwiched between targets in the encoding phase to ensure that they compete for working memory resources. We manipulated distractor strength with 2 levels, including strong distractors (using original stimuli) and weak distractors (using scrambled stimuli). We additionally manipulated distractor ratio with 2 levels, including a context of less distractors and a context of more distractors. Participants (N = 22) were asked to remember the targets and ignore the distractors in the encoding phase, and to press a key to the probe in the retrieval phase to indicate whether it was one of the targets presented earlier. We found that alpha power increased for strong compared to weak distractors, from frontal to occipital electrodes. The effect was present throughout the encoding phase, not only upon the presentation of distractors but also before and after the presentation of distractors, providing evidence for both proactive and reactive inhibition of distractors. Meanwhile, the effect was restricted to the context of less but not more distractors; in the context of more distractors, it seems that both strong and weak distractors were properly suppressed. The results suggest an adaptive mechanism might underlie the generation of alpha oscillations to protect working memory against distractors.