專題研討會 Symposium

• 鬱症之腦波與神經回饋研究／高雄醫學大學心理學系林宜美教授

在鬱症腦波生理病理機制，本研究檢驗鬱症患者是否符合「前額alpha波不對稱理論」，結果不支持鬱症患者之「前額alpha波不對稱理論」，而支持「全腦過度活化理論(brain hyperactivity; Bruder et al., 2007; Heller et al., 2010)」。本研究團隊以標準化加權低解析度電磁斷層(standardized weighted low-resolutionelectromagnetic tomography, swLORETA)分析腦波的電流源密度(current sourcedensity, CSD)，探討鬱症患者在休息狀態、憂鬱與快樂自傳式回憶作業之情緒相關腦區的活性與腦波連結性，結果發現鬱症組相較於健康控制組，在前額葉(prefrontal lobe)、扣帶迴(cingulate cortex)、楔前葉(precuneus)、杏仁核(amygdala)呈現過度活化現象，以及預設模式網絡(default mode network)過高的腦波連結性(coherence)。神經回饋療效方面，本研究團隊根據鬱症腦波之生理病理機制，發展每周二次、連續五周共十次的神經回饋方案。在表層神經回饋療效方面，前額alpha波不對稱神經回饋組、降低high-beta神經回饋組與控制組相比較，可顯著降低憂鬱與焦慮分數，其中以降低high-beta神經回饋組在情緒改善與腦波療效更佳，同時驗證神經回饋具可訓練性(trainability)、獨立性(independence)，以及可解釋性(interpretability)。在深層神經回饋療效方面，本研究以swLORETA結合Z分數常模資料庫，結果發現神經回饋組相較於控制組，可顯著降低憂鬱與焦慮症狀，降低異常的腦波Z分數。根據鬱症腦波生理病理機制發展的表層與深層神經回饋，可應用在鬱症之臨床心理介入，進而降低臨床憂鬱症狀。

• 以標準化加權低解析度電磁斷層Z分數神經回饋(standardized weighted low-resolution electromagnetic tomography Z-scoreneurofeedback, swLZNFB)在鬱症之療效／高雄醫學大學心理學系吳胤蓁、余汯恩、林宜美
  目的：先前研究發現鬱症(major depressive disorder)患者在額葉、顳葉與頂葉有異常的大腦活動。本研究探討標準化加權低解析度電磁斷層Z分數神經回饋訓練(standardized weighted low-resolution electromagnetic tomography Z-scoreneurofeedback, swLZNFB)對鬱症患者之憂鬱與焦慮症狀，以及腦波參數之療效。方法：48名鬱症合併焦慮症狀的患者被分配到swLZNFB組與控制組。swLZNFB組接受每周二次，每次1.5小時，連續五周共十次的swLZNFB，控制組則為常規醫療組。兩組參與者在前後測測量貝克憂鬱量表、貝克焦慮量表，以及閉眼5分鐘休息狀態的腦波。透過標準化加權低解析度電磁斷層(swLORETA)將腦波轉換為電流源密度(current source density, CSD)，包括delta、theta、alpha、beta、beta1、beta2、beta3、high-beta，並計算前額葉、前扣帶迴、後扣帶迴與杏仁核的腦波功率(power)、連結性(coherence)、延遲連結性(lagged coherence)、相位(phase)等參數，並與常模資料庫相比較，轉換為Z分數。以二因子混合設計變異數分析檢驗swLZNFB組與控制組在情緒與腦波參數之療效。

結果：swLZNFB組後測的憂鬱與焦慮分數、腦波異常數量，與腦波異常百分比均低於前測，且swLZNFB組在後測的憂鬱與焦慮分數低於控制組。swLZNFB組的後測相較於前測，顯著降低前額葉、前扣帶迴、後扣帶迴與杏仁核的beta1與beta3的CSD。

結論：10次swLZNFB降低鬱症患者的憂鬱與焦慮症狀，降低腦波異常參數與深部腦區的異常活動，可作為未來鬱症患者之臨床心理介入方案。

• 神經回饋在額葉-邊緣迴路的療效／高雄醫學大學心理學系黃泯翰、林宜美

目的：鬱症患者在額葉-邊緣系統迴路(fronto-limbic circuit, FLC)呈現異常腦波同步性(coherence)，本研究針對鬱症患者進行標準化加權低解析度電磁斷層Z分數神經回饋(standardized weighted low-resolution electromagnetic tomography Z-score neurofeedback, swLZNFB)，檢驗swLZNFB對鬱症患者的FLC腦波同步性之療效。

方法：51名鬱症患者，分派至swLZNFB組(n = 25)與控制組(n = 26)。swLZNFB接受藥物治療與一周兩次，共十次之swLZNFB，控制組僅接受常規藥物治療，並進行前後測五分鐘閉眼休息狀態腦波測量。在刪除腦波雜訊後，與NeuroGuide常模資料庫進行比對，輸出各頻帶腦波同步性的Z分數(包括delta、theta、alpha、beta與high-beta)，在絕對值轉換後，以二因子混合設計變異數分析檢驗swLZNFB組與控制組在治療前後腦波同步性之療效。

結果：在背內側前額葉(dorsal medial prefrontal cortex, dmPFC)與杏仁核(amygdala, AMY)的theta腦波同步性、腹外側前額葉(ventral lateral prefrontal cortex, vlPFC)與杏仁核的beta腦波同步性之Z分數絕對值呈顯著的「時間×組別」交互作用。組別單純主效果顯示swLZNFB組在後測dmPFC-AMY的theta腦波同步性、vlPFC-AMY的beta腦波同步性之Z分數絕對值顯著低於前測；控制組在後測vlPFC-AMY的beta腦波同步性之Z分數絕對值顯著高於前測。時間單純主效果顯示swLZNFB組在前測vlPFC-AMY的beta腦波同步性之Z分數絕對值顯著高於控制組，共變數分析發現swLZNFB組在後測vlPFC-AMY的beta腦波同步性之Z分數絕對值顯著低於控制組；swLZNFB組在後測dmPFC-AMY的theta腦波同步性之Z分數絕對值顯著低於控制組。

結論：swLZNFB可改善鬱症患者異常的FLC連結，提升前額葉皮質對於杏仁核之調控。

• 神經回饋之個案報告／高雄醫學大學心理學系張儷馨、林宜美
  目的：探討標準化加權低解析度電磁斷層Z分數神經回饋訓練(standardized weighted low-resolution electromagnetic tomography Z-score neurofeedback, swLZNFB)對鬱症患者在憂鬱、焦慮情緒、睡眠困擾與腦波之療效。

方法：個案為28歲單身女性，大三休學中，高三時因感情議題引起情緒低落，被診斷為鬱症(major depressive disorder)，個案自述常感到情緒低落、低自我價值感、明顯失眠問題，在服用安眠藥後仍入睡困難，在焦慮時感到心跳太快、喘不過氣。醫師評估後安排每周二次，每次1.5小時，共十次的神經回饋訓練，本研究以NeuroGuide軟體的symptom checklist，比對常模資料庫轉換為Z分數後，訓練憂鬱與焦慮情緒相關腦區的Z分數，包括前額葉、扣帶迴、楔前葉、杏仁核之腦波之功率(power)、連結性(coherence)、延遲連結性(lagged coherence)、相位(phase)等，在訓練前後各進行一次臨床症狀與腦波測量。

結果：個案完成十次神經回饋訓練後：(1)憂鬱與焦慮症狀、睡前身心警覺症狀下降；睡眠品質提升；(2)前額、頂葉、顳葉的beta波之Z分數顯著下降，全腦的在high beta波之Z分數顯著下降；(3)整體腦波異常率由42.20%下降至13.86%。

結論：在swLORETA ZNFB後可改善鬱症患者之情緒、睡眠困擾、大腦過度活化之腦區與異常腦波參數。

• 使用眼動追踪探討科學教育領域中認知科學的研究／國立陽明交通大學教育研究所陳聖昌助理教授

隨著認知神經科學的發展，近年來眼動追蹤技術（eye movement tracking technique）已經被廣泛地使用科學教育領域，並且做為探討學生認知行為的重要工具。多項研究也表明，學習者的凝視持續時間、掃視距離和回視行為與學生學習成效有關。本研究區分為三個面向進行科學教育領域中認知行為的研究。第一個面向從記憶歷程進行科學概念辨識與提取的研究；第二個面向利用三維動畫教材探討眼動對於學生建構心智模型的研究；第三個面向運用不同教學模式整合不同表徵方式之電學概念的數位學習內容，來探討眼動與類比推理之間的關係。三個面向研究的結果均指出，視覺注意力是調節訊息處理的重要關鍵。當外界訊息進入大腦時，大腦也同時會因為當下情境或訊息的熟悉度，同步提取大腦儲存的長期記憶，進入大腦的工作記憶區進行訊息的整合與理解。因此在科學教學上，教師需要提供學生生活情境相關且熟悉的事物，讓學生使用類比推理的方式，逐步形成各種類比的心智模型，並且進而利用各種適當的教學媒體、策略或工具提供良好的學習刺激，使其心智模型轉變成科學心智模型，並且最終形成科學概念。因此科學教師的教學，除了要注重學生的先備知識之外，能夠在適當的時機，提供適合學生學習的刺激物，此點更顯得重要。

• 大腦腦區間的協同作用在科學學習歷程中的功能與角色／國立彰化師範大學生物系周文己助理教授
  科學學習中的認知歷程在近年來被廣泛討論，越來越多的研究將認知神經科學的方法融入科學學習相關的研究議題中，以探索學習歷程中所隱含的訊息處理模式以及大腦不同腦區所涉及的認知功能。本研究之目的是藉由回顧我們過去一系列結合認知神經科學方法和科學教育理論之研究成果，藉由腦波儀收集學生進行科學學習時的腦波動態變化情形，深入探討科學學習歷程中訊息編碼、大腦功能性連結以及關係性推理這類型複雜的認知功能時，大腦不同腦區的腦波活化以及協同合作情形。

在科學概念編碼任務的研究中，研究結果顯示額葉、頂葉以及枕葉在科學概念編碼任務中具有不同的功能及角色，額葉主要功能為認知控制與維持訊息處理，而頂葉及枕葉則是互相合作進行訊息表徵的處理。而在大腦功能性連結的研究中，結果顯示中央頂葉與枕葉的腦波具有明顯的同調性及相位一致性，這表示中央頂葉與枕葉具有功能性連結並透過同步化作用以完成工作記憶任務。而在關係性推理這類複雜的認知功能上，我們發現學習者在解決難度較高的關係性推理問題時，大腦前扣帶迴θ波、右頂葉α波、右枕葉α波以及μ波間具有較強的同調性，顯示學習者進行較高難度的光學成像關係性推理時，需要大腦不同腦區間進行協同合作才能完成。綜合上述研究成果，我們認為在科學學習歷程中，無論是科學概念的訊息編碼、功能性連結以至於較複雜的認知功能如關係性推理等，都需要大腦不同腦區之間的協同合作，期望這一系列的研究成果可幫助我們了解如何將認知神經科學方法應用於科學學習的相關研究中，並期待能從另一角度更加深入了解科學學習的歷程而從中獲得不同的啟發。

• 探討科學概念學習的記憶與決策：EEG研究／國立彰化師範大學科學教育研究所黃莉郁助理教授

記憶是人類認知的基石，透過訊息的輸入與輸出，我們不斷累積經驗與知識，擴展認知基模，以因應周遭環境中多樣的問題與挑戰，而由於科學知識通常涉及相互關聯且具階層性的結構，也更加凸顯出訊息處理和記憶品質的重要性。因此，本研究綜整過去研究團隊運用腦電圖 (electroencephalogram, EEG)探討與科學概念學習有關的三篇研究 (Huang et al., 2013, 2018; Liang et al., 2020)，透過訊息處理及工作記憶相關理論，探討科學概念登錄 (encoding)、提取(retrieval)與科學決策 (decision-making)歷程中，大腦各區域相互連結(connectivity)的動態變化，並透過腦波之時間與頻率分析，深入討論學生成功獲得及使用科學概念的大腦模式。研究結果發現前扣帶回θ波在工作記憶中的認知控制與表徵維持上扮演了重要角色，其功率高低之調控與相應之時間點，對任務正確性有關鍵性的影響；此外，前後腦區之不同頻率腦波的出現模式和時間上的變化，亦反映了其所擔負的認知功能不同，並顯現於不同的表徵呈現上，這些動態變化也顯示出與科學概念學習相關的訊息處理，很可能便是透過不同腦區、不同頻率腦波之間的合作與分工加以調控，這些結果有助於我們更深入地了解學生科學學習的的認知歷程與可能遭遇的學習困難，為未來實時回饋與腦機介面等應用提供參考，以期提高科學學習成效和實現個性化教學之目標。

• 運用fMRI探討科學與迷思概念的運作及其改變之機制／國立陽明交通大學教育研究所佘曉清教授
  許多研究顯示迷思概念很難經由傳統的科學教學來克服，並阻礙科學概念的理解與建立。為促進學生概念改變，研究者發展雙重情境學習模式(DSLM) (She, 2002, 2003, 2004a, b)，提供給學生產生不和諧與心智結構，且成功促進學生科學概念改變成功。為探討科學概念改變成功之大腦運作機制，我們利用功能性磁振造影functional magnetic resonance (fMRI)來探討學生在提取科學概念和迷思概念時的大腦活動。 結果顯示提取正確的科學概念時，大腦hippocampus (HIP)的活化顯著高於迷思概念。此外 Granger causality (GC) 分析發現，MTG-HIP、和IN-ACC-HIP網絡的連通性(connectivity)在處理正確的科學概念時比處理迷思概念時更加顯著(She et al., 2023)。進一步運用fMRI來研究學生科學概念改變時腦區的活化。結果顯示概念改變成功時，大腦hippocampus (HIP)的活化顯著高於概念改變失敗。同時也有較強的superior medial frontal, superior temporal gyrus, inferior frontal gyrus, mid frontal的活化。初步研究結果提供DSLM促進科學概念改變成功的可能機制之解釋。

• Reading Emotion in Context: the Reciprocal Relationship between Feelings and Words／國立臺灣大學語言所李佳霖副教授（Online）
  Emotions play a crucial role in various aspects of our lives and are fundamental to human communication. While the ability to express emotions through language is relatively recent in evolutionary terms, it is vital for psychosocial well-being. As such, the interaction between language and emotion in the brain holds significant implications for addressing societal challenges. In this talk, I will present Event-Related Potential (ERP) data collected from healthy young and older adults reading sentences containing emotional context and/or emotional final words. Our data from young adults showed that emotion words in a neutral context could trigger immediate and robust neurophysiological responses, indicating early attention allocation to emotion words (indexed by the P200 effect) and more controlled and continued deeper assessment of emotion (indexed by late positive responses). Moreover, emotional context induces similar emotion-related attention processes from the incoming word. Critically, regardless of the words’ emotionality, greater late positivity was observed in words following emotional context relative to neutral context. These language-induced emotional effects are less available in healthy older adults and were more readily observed with emotional context when more time was allowed to build up emotionality. These findings thus provide evidence for contextualized emotion responses during coherent sentence comprehension, in line with the view of the constructionist theory of emotion. Our findings also showed that lexically unexpected continuation following an emotional context elicited an anterior positivity in older adults. As the lexical unexpectedness effect indexed by the anterior positivity was usually absent in older adults, these findings suggest that emotion-induced attention can potentially enhance certain aspects of language processing. Together, these results provide support for contextualized emotional responses during the comprehension of congruent sentences and provide an initial understanding of how language and emotion work together in sculpting information processing.

• Cognitive Assessment of Autism Spectrum Disorder with Artificial Intelligence in a Game-Based Social Platform／

國立中山大學電機工程學系曾乙立助理教授

Numerous serious games have been developed to offer personalized training and cognitive assessment, particularly in areas involving emotions and social interactions. However, these games have traditionally been assessed primarily through behavioral questionnaires and verbal feedback. Recent advancements in human-machine interfaces have created opportunities to integrate physiological markers into upcoming gaming interfaces. In this research, we designed a game-based social training platform that captured eye-tracking and electroencephalography (EEG) signals to provide neurophysiological indicators for cognitive assessment in children and adolescents with autism. We recruited both typically developing (TD) participants and those with autism spectrum disorder (ASD) to engage in the social training game while recording eye-tracking and 12-channel EEG signals during their sessions. The EEG signals underwent preprocessing, and the study analyzed and compared event-related potentials (ERPs) and event-related spectral perturbations (ERSPs) between the two groups. We utilized six different machine learning frameworks to aid in the computer-based assessment of ASD. The results demonstrated that the support-vector machine achieved the highest balanced accuracy, reaching up to 90%, when EEG features were used for evaluating autism. These findings indicate the potential of incorporating physiological features into a game-based social training interface for cognitive assessment in individuals with ASD.

• The cognitive mechanisms of worry and the intervention approach／國立中山大學教育研究所馮雅群助理教授

Worry is a stream of negative thoughts about future events and can be maintained by the consistent tendency to generate interpretations (i.e. interpretation bias). Interpretations can be generated at the moment the ambiguity is first encountered (online) or later on reflection (offline). At present, it is unclear which specific types of interpretation are related to worry, and which can be modified by cognitive bias modification for interpretation (CBM-I) and therefore reduce the levels of worry. To address these questions, high worriers (n=28) and low worriers (n=27) were compared on different interpretation bias measures. "Offline" tasks assessed interpretations after the opportunity of reflection; while "online" tasks assessed initial interpretations when encountering ambiguity. Event-related potentials were also employed to exam neural activity when encountering ambiguous information. In line with our prediction, high worriers lacked benign interpretation biases that can be found in low worriers across different types of interpretation. In our second study, we investigated whether enabling high worriers to generate more positive interpretations can reduce the levels of worry. High worriers were randomly allocated into single session CBM-I training (n=35) or an active control group (n=31). Both online and offline interpretation biases were assessed post CBM-I. Consistent with our predictions, the training group showed greater online and offline benign interpretation biases and lower levels of worry compared to the control group. However, only the offline interpretation bias mediated the relationship between groups and levels of worry. These studies show that the benign interpretations high worriers lack can be promoted by CBM-I.

• Understanding emotion words in late childhood／高雄醫學大學心理系葉佩文助理教授

Recent studies suggest that word is important for building emotion concepts, aggregating inducing events (e.g., fight), sensory-motor experience (e.g., pain), and internal feelings (e.g., aggression). However, fewer Taiwanese studies have explored the development of emotion words. In this presentation, I will outline our behavioral and electrophysiological approach on how typically developing children/adolescents understand emotion words. First, I will discuss a series of event-related potential studies on adolescents (11–14 years) and adults who were required to categorize valence for emotion words. Results indicated that lexical processing was modulated by emotion connotations and valence in both age groups, particularly for negative valence (indicated by N400). Notably, the younger group exhibited a more sustained emotional processing for all emotion words than the older group (indicated by late potential component, LPC). Follow-up studies showed that facial expressions may restrictively enhance understanding emotion from words. Behavioral data complementally suggest that valence is key to acquisition of emotion concepts for words. Finally, I will present recent work showing that internal states (e.g., depression and anxiety) may influence the processing of emotion words. Taken together, these studies enrich our understanding of how emotion concepts connect to language in late childhood.