研究生姓名 : 陳奕秀
學年度 : 107
論文題目 : The individual difference and the neural correlates of visual working memory load effects revealed by EEG alpha/ beta desynchronization
英文論文題目 : The individual difference and the neural correlates of visual working memory load effects revealed by EEG alpha/ beta desynchronization
指導教授 : 阮啟弘
系所名稱 : 國立中央大學認知神經科學研究所
論文頁數 : 86
中文關鍵字 : 視覺工作記憶、個體差異、腦電波
英文關鍵字 : visual working memory、individual difference、EEG
中文摘要 :
視覺工作記憶(VWM)是我們日常生活中至關重要的認知功能。通常可以通過使用特徵綁定方法來測量視覺工作記憶能力容量,其中參與者必須將對象的多個特徵集成為單一表示。幾個電生理學證據揭示了潛在的神經機制及其與視覺工作記憶能力的相關性。過去研究顯示,在視覺工作記憶測驗中,記憶保留期(retention period)中的α synchronization與記憶負載(memory load)的增加呈現負向的關聯性。這項結果符合對於α波的抑制假設(inhibition hypothesis),這項假設說明,藉由α波的能量上升,能夠抑制腦中與記憶無關的訊息。然而,近來有更多研究注意到desynchronization在記憶中扮演著重要的角色,像是α波的能量下降與注意力狀態有關。另一方面,一些研究表明,在頂葉的α和β desynchronization,與長期記憶編碼(encoding)和檢索(retrieval)的處理密切相關。過去研究提出一項假設,α和β desynchronization與訊息豐富度(information richness)有關,當被記憶的物品訊息含量越高(記憶負荷越重),則α和β desynchronization的程度越大。然而,desynchronization與視覺工作記憶之間的關係仍然不明確,需要更多證據來闡明β desynchronization在視覺工作記憶中的作用。過去的研究常以側向化的工作記憶測驗(lateralized visual working memory task) 來探討在頂葉和枕葉的對側延遲活動(contralateral delay activity, CDA component) 和α波神經震盪與負荷差異的神經關聯性,並且顯示出這兩種神經相關性(neural correlates) 和個體差異高度相關。然而,傳統方法的潛在限制是處理空間訊息和特徵綁定的過程可能會產生混淆。此外,非側化的神經振盪(像是β波)可能受到此實驗設計的限制。因此,本研究採用了不同版本的改變偵測測驗(change detection task),同時結合了顏色和形狀的特徵。並且將空間訊息設計成與測驗不相關的因子(irrelevant factor),以便分離空間注意和特徵綁定的過程。另外,先前的研究表明,神經相關的模式在高記憶容量和低記憶容量個體之間有所差異。因此,為了觀察個體差異,採用Pashler’s K(Kp)來表示視覺工作記憶的能力,Kp代表受試者所能記憶的物品個數。本研究採用Kp = 2作為區分高低表現者的客觀依據。並且進一步利用希爾伯特-黃轉換(Hilbert-Huang Transform, HHT)中遮罩的經驗模態分解法(Masking Empirical Mode Decomposition, Masking EMD) 進行時頻分析以闡明視覺工作記憶容量個體差異背後的神經機制。本研究旨在探討α/ β desynchronization與視覺工作記憶的個體差異。我們假設視覺工作記憶中的表現與α/ β desynchronization有關,並且,α/ β頻率範圍內的電生理訊號在高低表現者之間會有所不同。
本研究的結果顯示在頂葉和枕葉的β 神經震盪與視覺工作記憶的行為表現有正向的關聯,並且,此相關從刺激呈現前即被觀察到,說明了這種相關性與偵測改變測驗無關而是個體之間原有的差異(baseline difference),為了排除與測驗無關的干擾,我們選取刺激開始前的-599 ms 到 -200 ms 作為比較的依據(baseline)。結果中能清楚地呈現出α/ β desynchronization,並且在高記憶負荷的情況下,α/ β desynchronization的程度顯著地比低記憶負荷的情況下大,而這種記憶負荷差異(load difference)主要出現在在枕骨、右頂葉中,與先前α/ β desynchronization反映訊息豐富度的假設相互呼應。另外,我們觀察到在β頻率範圍左前額葉的記憶負荷差異與視覺工作記憶容量之間呈現負相關,表明在β頻率範圍中高低負荷的差異越大反映出越高的視覺工作記憶容量。我們分別比較了高低表現者的記憶負荷差異,進而探討這種記憶負荷差異在不同個體上的表現,結果顯示,只有高表現的個體表現出,除了在枕骨、右頂葉,在左前額葉區域也能觀察到顯著的記憶負荷差異,但在低表現者則觀察不到。綜合上述結果,我們認為左前額葉的β神經震盪很有可能是用來區分個體差異的潛在性重要指標。
英文摘要 :
Background: Visual Working Memory (VWM) is the crucial and complex cognitive function in our daily lives. Typically, estimates of VWM capacity can be measured by using feature binding method in which participants have to integrate multiple features of an object into a unitary representation. Several electrophysiological evidence has revealed the underlying neural mechanism and its correlation with the capacity of VWM. More specifically, the synchronization of alpha power has been found to be associated with the increasing load in the retention period of VWM tasks. It is in line with the inhibition hypothesis that synchronization of alpha power suppresses the irrelevant information in disengaged regions. However, more attention has been shifted to the role of the decreased neural activity in memory. It was mostly proposed that the decrease of alpha power is correlated with the heightened attentional state. On the other hand, several studies have shown that alpha and beta desynchronization, especially from the parietal sites, was strongly correlated with the processing of long-term memory encoding and retrieval. It was proposed that the desynchronization of alpha and beta power is related to the information richness. Therefore, the relationship between the desynchronization and VWM still remains unclear and more evidence is needed to clarify the role of decreased beta power in VWM capacity. Previously, studies have shown lateralized parieto-occipital negative slow wave (i.e. Contralateral Delay Activity, CDA) and lateralized alpha power modulation as neural correlates of VWM capacity that are highly correlated with the individual difference. However, the potential limitation of the traditional approach utilizing lateralized VWM task is that the process of spatial information and the binding process might be confounded. In addition, non-lateralized neural oscillatory processes might be constrained by the experiment design in which the lateralization of alpha frequency band is more pronounced. Hence, the present study adopted a modified version of the change detection task which combined the features of color and shape simultaneously. Moreover, the spatial information was considered as an irrelevant factor in order to dissociate the process of spatial attention and the binding process. In spite of this, previous studies have shown that the pattern of the neural correlates is different between high and low capacity individuals. Therefore, to observe the individual difference, Pashler’s K (Kp), which represented the number of items retained in memory, was adopted to estimate the capacity of binding-VWM. Also, an objective and stable criterion of Pashler’s K value as 2 (i.e. Kp = 2), rather than the traditional median, was utilized into the classification of the high and the low groups. To elucidate the neural mechanism behind the individual difference in binding-VWM capacity, Masking Empirical Mode Decomposition (Masking EMD) of Hilbert-Huang Transform (HHT), an enhanced EEG signal processing algorithm was further utilized for data analysis to achieve high temporal resolution in time-frequency domain. The current study aims to reveal the relationship between the alpha/beta desynchronization and the individual difference in binding-VWM. Based on previous findings reviewed above, we hypothesized that the performance in VWM is related to the alpha/ beta desynchronization and the pattern of electrophysiological activities in alpha/ beta frequency range would differ between low and high capacity subjects.

Method: 68 neurologically normal participants were recruited from National Central University for the current study. Owing to technical discrepancies such as more than two broken channels either with zero as the value of signal or consistently showed the unreasonable power which was 10 times greater than other channels and the insufficient of trials (less than 10 trials) in the high load condition after the process of artifact rejection, 10 subjects were excluded. The binding-VWM task with electroencephalography (EEG) measurement was employed. The binding paradigm was a modified change detection task adopted from Tseng and colleagues (2016). To identify the neural correlates underlying different load in high and low capacity individuals, we manipulated the load by changing the set size within a memory array. Set size four (SS4) and two (SS2) were corresponding to high and low load condition respectively. Participants were instructed to memorize the items presented in the study array. After a short-term interval, the test array was presented and participants were requested to judge whether the items in test array were identical to those in the study array with a key response. To estimate the capacity of binding-VWM, Pashler’s K value (Kp) was adopted in the current task. We utilized the value of Kp = 2 as the criterion for separating the subjects into groups of high or low capacity.

Results: In the present results, positive correlation between raw power of beta oscillations and performance was observed across time at occipital and parietal regions, indicating a task-irrelevant relationship which might be a baseline difference of the neural activity between individuals. To rule out the task-irrelevant interference in the following comparisons, we re-scaled the data from the baseline (-599 ms to -200 ms relative to stimulus onset). The current results presented the desynchronization of alpha and beta power. Importantly, the level of desynchronization within alpha and beta frequency bands were significantly lower in high load condition when compared to low load condition at right temporo-parietal, frontal and occipital regions in the binding-VWM condition. These results confirm the hypothesis that the information richness is reflected by the desynchronization of alpha and beta power. Furthermore, the decreasing alpha/beta power might reflect the reactivation of sensory features and correlated with the successful recall. The results then showed a negative correlation between beta activity and VWM capacity among left frontal regions, suggesting that the higher capacity was reflected by the lower amplitude of beta oscillations. Furthermore, these results demonstrated that only high capacity individuals but not the low capacity ones showed the neurophysiological pattern that the power of alpha/ beta oscillations significantly differed between load conditions across occipital, right temporo-parietal, and left frontal regions.

Conclusion: We argue that the observed desynchronization of alpha/beta oscillations reflects the information richness of a presented scene and is negatively correlated with the capacity of binding-VWM. Our results reveal the neural signatures pertaining to the set size difference circumscribed to posterior brain region whereas the individual differences refrain to the left frontal regions. According to current findings, we suggest that the general difference between loads among occipital and right temporo-parietal regions was more pronounced in high performers, indicating that the performance will improve when the left frontal regions were more engaged in the binding process.
 
地址:320 桃園市中壢區五權里2鄰中大路300號 服務信箱: ncu5200@ncu.edu.tw
電話:(03) 422-7151轉65200 傳真:(03) 426-3502
中大認知所 版權所有 © 2005 Institute of Cognitive Neuroscience. All rights reserved.
請以IE6.0或更新版本,以及1024 x 768以上之解析度瀏覽以取得最佳效果。