| 摘要: | 摘要
工作記憶是個能同時儲存與處理複雜認知功能所需訊息且有結構的系統。我們對工作記憶有興趣是因為其容量與流體智力有高度相關性。近年有研究指出藉由適當的訓練，工作記憶容量的進步會反映在作業表現或智力測驗分數上。儘管如此，在相關的行為研究中尚未獲得一致的結果，仍然有學者抱持相反的意見。其中一個關鍵原因可能在於，認知訓練的成效也許不是以智力測驗成績的提升來表現之，換句話說，行為表現(智力測驗成績)是間接的量測，可能沒有完全反應神經生理的改變，因此，本研究希望能從事件相關電位P300中找出能客觀量化的指標，用來評估工作記憶訓練的成效。
本研究招募二十個中央大學的學生並隨機分配成兩組，實驗組與對照組各十人。實驗組被要求進行一個與spatial span作業相似的益智遊戲，為期三週，每週五日，每日三十分鐘。對照組則進行與記憶無關的動作控制遊戲，訓練頻率與實驗組同。於訓練期間，每位受試者都將進行五次(訓練前、訓練一週、訓練兩週、訓練三週、結束一週後)四種難度的空間n-back作業(n=0, 1, 2, 3.)，並同時記錄腦電波。反應時間和準確率等行為數據將會在實驗結束後進行離線處理。除了P300振幅，高記憶負荷(3-back)與低難度(0-back)之間於頂葉區的振幅比(以R表示) 與振幅比隨訓練的變化(以R表示)也會被計算出並加入統計分析。
重複量數變異數分析的統計結果顯示，組間差異在行為數據、P300振幅和比值R中皆未達顯著，可能是由於微弱的訓練成效無法被線性的方法找到。在統計結果中只有R(每週減去訓練前)有顯著的組間差異：實驗組的平均R在訓練後大於對照組，表示此記憶訓練確實會影響神經活化，並且在訓練停止後維持至少一週。在未來的工作中，我們會繼續找出此訓練成效之下的神經機制。; Working memory (WM) is the structured system which can simultaneously maintain and manipulate information for complex cognitive behavior. The capacity of WM is of interest because it was thought to be related to fluid intelligence. Recent studies have shown that working memory capacity can be improved by adequate training as reflected in greater task performance / standardized intelligence scores (IQ scores). Neuroimaging studies have further shown that there are greater activities after WM training in the prefrontal and parietal cortex and these higher activities are associated with the facilitatory effect on striatal dopamine release. Nevertheless, results from behavioral studies were inconsistent. In this study, we aimed to find a direct estimate of WM training effect by P300.
Twenty healthy college students were recruited and randomly divided into two groups (Experimental and Control, 10 each). Experimental group was asked to play a spatial span task-liked puzzle game 30 minutes a day, 5 days a week for 3 weeks while control group underwent a movement related game in the same training frequency. EEG data from all subjects were recorded during n-back tasks (n=0,1,2,3) at different training phases (pre-test, 1 week, 2 weeks, 3 weeks, follow-up). The behaviour data during the task like reation time and accuracy were computed offline. The P300 amplitudes ratio, the P300 ratio at Pz between high (3-back) and low (0-back) working memory load (denoted as R) and the changes of R in response to training (denoted as R) were calculated. Both the behaviour and EEG data and entered repeated measurement ANOVA for statistic test.
The statistic test on the behaviour data, P300, and R shows no significant difference between two groups，implying a subtle training effect that may not be captured by linear methods. The only statistical result confirmed there is a significant difference of R between two groups (each week minus pre-test) is R: The mean R of the experimental group became greater than that of the control group after memory training, suggesting an effective training and this training effect could stay at least one week after training. In a future study, we will be looking for the mechanisms underlying these training changes. |
