以腦電圖為基礎探討十二週動手樂活動對於社區健康長者之成效

Abstract

背景：隨著台灣邁入超高齡社會，失智症防治已成為重要議題。近年來，認知介入被證實在維持健康長者的認知與日常生活功能及預防失智症方面發揮關鍵作用。然而，過往針對認知彈性的介入研究中，鮮少結合腦電圖技術探討介入後的神經機制，使得介入成效仍待釐清。本研究結合事件相關電位（ERP）技術，探討社區健康長者參與以主動推理（Active Inference）為基礎的「動手樂活動」後，對認知彈性與相關認知歷程之成效，並與傳統按部就班式教學的「控制動手樂活動組」進行比較。此外，研究亦納入多項認知與日常生活功能行為測驗，以檢視兩組在轉移效應上的差異。

方法：本研究採隨機對照試驗，共招募65名健康長者，篩選後納入48名受試者。分為兩梯次，每梯次隨機分配24人至動手樂活動組及控制動手樂活動組。兩組分別參與為期12週、每週2小時的活動，並於介入前後進行評估。主要成效指標為結合事件相關電位（ERP）（包括target-locked P2、N2及P3b）之交替任務測驗，以評估認知彈性。次要成效指標包括遠距聯想測驗（聚合性思考）、用途替代測驗（發散性思考）、彩色路徑測驗（注意力與執行功能）、魏氏空間與氏數字記憶廣度測驗（視覺與聽覺工作記憶）及操作型日常生活技巧評估-簡短版。本研究假設，參與動手樂活動的健康長者將比參與控制動手樂活動的長者在認知彈性及相關認知歷程上有更顯著的提升，並產生更廣泛的認知功能轉移效應。

結果：共41位完成行為評估後測（動手樂活動組22位，控制組19位），其中37位完成ERP後測（動手樂組19位，控制組18位）。兩組基線特徵中，控制組年齡顯著較高 (p = .020)。 在主要成效指標方面，動手樂活動組在特定轉換成本(Specific Switch Cost, p = .014，d = 0.33)與轉換後成本(Post Switch Cost, p = .002，d = 0.73)均呈現顯著下降，顯示其在規則轉換任務(switch trials)相對於轉換前任務(pre-switch trials)、以及轉換後任務(post-switch trials)相對於轉換前任務(pre-switch trials)的反應時間皆有明顯改善。而控制動手樂活動組則在轉換後成本(p = .029, d = 0.54)與規則切換任務錯誤率(switch trials error rate)上呈現顯著下降(p = .045, d = 0.72)。然而，兩組在主要成效指標的組間變化未達顯著差異。 在事件相關電位(ERP)分析方面，動手樂組於P2轉換效應(Switch Effect)峰值振幅差在Fz (p = .011, d = 0.75)、FCz (p = .027, d = 0.61) 與Cz (p = .034, d = 0.63)顯著減少，可能表示任務激活之認知歷程得到改善。此外，動手樂組在P3b的Fz: Pz對稱性顯著降低 (p = .006, d = 0.76)，可能反映對前額葉資源依賴性下降，工作記憶資源之神經分佈模式趨近年輕化。控制組則於P3b異質性重複任務(Heterogeneous Trials)中，Pz電極峰值潛值顯著提前 (p = .030, d = 0.47)，可能反映異質性區塊之重複任務中，更早啟動工作記憶資源分配歷程。 在組間比較方面，相較於控制動手樂活動組，動手樂活動組在Fz、FCz與Cz電極的P2轉換效應峰值振幅差呈現顯著較大的下降 (Fz: p = .008, d = 0.96；FCz: p = .022, d = 0.81；Cz: p = .044, d = 0.71)，以及在Fz: Pz對稱性亦呈現顯著較大的降低 (p = .034, d = 0.75)。可能反映動手樂組在任務激活的表現較佳，且對額葉區域在工作記憶資源分配上的依賴顯著降低。而控制動手樂組在P3b異質性重複任務(Heterogeneous Trials, p = .016, d = 0.86)與轉換前任務 (Pre-switch trials, p = .047, d = 0.76)的峰值潛值縮短顯著優於動手樂組，可能反映其在異質性區塊之重複任務中，能較早啟動工作記憶資源的分配歷程。 次要成效指標方面，動手樂活動組在視覺工作記憶(p = .034, d = .57)、發散性思考(p < .001, d = .52)及執行功能(p = .035, d = .20)方面呈現顯著的轉移效應；控制動手樂活動組則在發散性思考(p = .014, d = .31)、持續性注意力(p = .017, d = .46)及執行功能(p = .027, d = .49)表現出轉移效應。組間比較顯示，動手樂活動組在視覺工作記憶的轉移效果顯著優於控制動手樂活動組(p = .033, d = .70)。

結論：本研究初步結果顯示，主動推理導向之動手樂活動可能可以顯著提升健康長者之認知彈性，並改善與交替任務測驗有關之認知歷程，包括任務激活(P2)，及減少工作記憶資源分配中對額葉之依賴(P3b Fz: Pz對稱性)。相對地，控制組在P3b峰值潛值顯著提前，可能反映在重複任務中能更早啟動工作記憶處理歷程。 兩組皆展現不同的認知功能轉移效應，其中動手樂活動組的轉移效應涵蓋視覺工作記憶、發散性思考及執行功能，而控制動手樂活動組則展現在注意力、發散性思考及執行功能。此結果強調，以認知彈性為核心的介入方式對提升健康長者的認知彈性及廣泛認知功能具潛在價值。未來研究應納入更多元樣本，如主觀認知衰退或輕度認知功能障礙的長者，以進一步驗證此介入模式的成效。

關鍵字：認知彈性、主動推理、認知活動、腦電圖、健康老人、事件相關電位

Background: As Taiwan enters a super-aged society, dementia prevention has become a critical issue. Cognitive interventions have been shown to play a key role in maintaining cognitive and daily functioning in healthy older adults and in delaying the onset of dementia. However, few prior studies on cognitive flexibility interventions have incorporated electroencephalography (EEG) to examine post-intervention neural mechanisms, leaving the impact on cognitive processes unclear. This study employed event-related potentials (ERPs) to investigate the effects of active-inference-based “Tinkering” activity on cognitive flexibility and its underlying cognitive processes in healthy older adults, and compared these effects with a control condition using traditional step-by-step instructional Tinkering. Additionally, multiple cognitive and instrumental daily functioning assessments were included to examine transfer effects across both groups.

Methods: A randomized controlled trial was conducted with 65 older adults, of whom 17 were excluded after screening. A total of 48 participants were randomly assigned across two waves into either the Tinkering group or the Control-Tinkering group. Both groups participated in a 12-week program with two-hour sessions per week. Pre- and post-intervention assessments included EEG recordings and cognitive tests. The primary outcome was cognitive flexibility, assessed using a Task-Switching Paradigm with ERP components (target-locked P2, N2, and P3b). Secondary outcomes included the Remote Associates Test (convergent thinking), Alternate Uses Test (divergent thinking), Color Trails Test (attention and executive function), Wechsler Spatial and Digit Span Tests (visual and auditory working memory), and the Brief UCSD Performance-Based Skills Assessment (UPSA-B, Traditional Chinese version). We hypothesize that healthy older adults participating in Tinkering activity based on active inference theory will show significant improvements in cognitive flexibility and related cognitive processes, along with a broader enhancement in cognitive functions compared to those in the Control-Tinkering group.

Results: A total of 41 participants completed post-intervention behavioral assessments (Tinkering: n = 22; Control-Tinkering: n = 19), and 37 completed post-intervention ERP assessments (Tinkering: n = 19; Control-Tinkering: n = 18). The Control-Tinkering group was significantly older at baseline (p = .020). In the primary outcome measures, the Tinkering group showed significant reductions in Specific Switch Cost (p = .014, d = .33) and Post-Switch Cost (p = .002, d = .73), indicating a decrease in response time for tasks requiring rule changes before and after the switch trials, respectively. In contrast, the Control-Tinkering group showed a significant reduction in Post-Switch Cost (p = .029, d = .54) and error rates during switch trials (p = .045, d = .72). with no significant between-group difference. Regarding ERP analyses, within-group comparisons revealed that the Tinkering group exhibited a significant reduction in P2 switch effect amplitude at Fz (p = .011, d = 0.75), FCz (p = .027, d = 0.61), and Cz (p = .034, d = 0.63), which may reflect improvements in task-set activation processes. Additionally, a significant reduction in P3b Fz: Pz asymmetry was observed (p = .006, d = 0.76), possibly indicating decreased reliance on frontal regions and a shift toward a more parietal neural activation pattern, similar to that seen in younger adults. In the Control-Tinkering group, a significant shortening of P3b peak latency at the Pz electrode during heterogeneous trials was found (p = .030, d = 0.47), which could suggest earlier engagement of working memory resources during repeated tasks within heterogeneous blocks. Between-group comparisons showed that the Tinkering group had significantly greater reductions in P2 switch effect amplitude at Fz (p = .008, d = 0.96), FCz (p = .022, d = 0.81), and Cz (p = .044, d = 0.71), as well as a significantly greater reduction in P3b Fz: Pz asymmetry (p = .034, d = 0.75), which may suggest relatively enhanced task-set activation and reduced reliance on frontal-lobe resources for working memory allocation. In contrast, the Control-Tinkering group showed significantly greater shortening of P3b peak latency in both heterogeneous trials (p = .016, d = 0.86) and pre-switch trials (p = .047, d = 0.70) compared to the Tinkering group, which might reflect earlier initiation of working memory processing during repeated tasks. In secondary outcome measures, the Tinkering group showed significant transfer effects in visual working memory (p = .034, d = .57), divergent thinking (p < .001, d = .52), and executive function (p = .035, d = .20). In contrast, the Control-Tinkering group demonstrated transfer effects in divergent thinking (p = .014, d = .31), sustained attention (p = .017, d = .46), and executive function (p = .027, d = .49). Between-group comparisons indicated that the Tinkering group demonstrated significantly greater transfer effects in visual working memory (p = .033, d = .70) compared to the Control-Tinkering group.

Conclusion: This study provides preliminary evidence that engaging in active-inference-based Tinkering activities may enhance cognitive flexibility and support improvements in task-related cognitive processes in healthy older adults. Specifically, these changes might include enhanced task activation (reflected by reduced P2 switch effect amplitude) and decreased reliance on frontal regions during working memory allocation (reflected by reduced P3b Fz: Pz asymmetry). Conversely, the Control-Tinkering group exhibited earlier P3b peak latencies during repeated trials in heterogeneous blocks, which might reflect faster engagement of working memory allocation processes during repeated trials in the heterogeneous blocks. The intervention effects of both the Tinkering and Control-Tinkering groups transferred to different cognitive domains, with the Tinkering group showing transfer effects in visual working memory, divergent thinking, and executive functions. In contrast, the Control-Tinkering group showed effects in sustained attention, divergent thinking, and executive functions. These results emphasize the potential of interventions based on cognitive flexibility in improving cognitive flexibility and broader cognitive functions in healthy older adults. Future research should include a more diverse sample, such as individuals with subjective cognitive decline or mild cognitive impairment, to further validate the effectiveness of this intervention.

Keywords: Cognitive Flexibility, Active Inference, Cognitive Activities, Electroencephalography (EEG), Healthy Older Adults, Event-Related Potential (ERP)