遭遇政策困境時，透過向外學習以產生自身方案的雛型，已是政策實務常見的作法。然而，政策學習要能成功，教材（lessons）/資訊的品質及適合度至關重要，學習前若能慎選學習對象，較能事半功倍。可惜的是，現有文獻對於選擇政策學習對象之機制為何？政府如何挑選學習對象？背後有無什麼認識論策略？並沒有系統性討論。
依此，本文初探我國政策學習現況，先回顧和政策學習對象選擇有關的文獻，納入我國地緣特徵因素後，整合歸納提出分類架構，再以行政院部會的資料進行內容分析，描述我國政策學習對象的來源分布，並推論背後的可能機制，與前述架構對話。結果發現，我國的學習對象選擇機制具有制度主義解釋的特徵，與組織現有網絡、認知限制與習慣有關，受非因政策學習目的而建立的關係（例如外交關係）影響，會考量習慣與便利而選擇容易取得資訊的地區（例如交通、語言易及），且因過往歷史而受霸權、區域主義牽引，幾乎所有政策領域都相同。
雖然本文無法評斷現有政策學習資訊的品質是「好」或「壞」、多樣性是否足夠，但在實務上，仍建議將資訊評估工作制度化，導入到機關中。研究上則針對政策學習所倚賴之資訊網絡的變遷，進行跨時分析，並深探地緣特徵因素所產生之影響。

Utilizing learning from others as a prototype for policy solutions to reduce the risk of policy experimentation and address unprecedented issues has become a standard tool in public governance. However, for policy learning to succeed, it is crucial to obtain useful "lesson" information that enhances policy quality. Unfortunately, research in the field of policy learning has not systematically addressed a fundamental epistemological question: Where do the sources of lessons originate (from countries providing helpful information or from friendly, neighboring, or linguistically similar countries)? Do different policy areas follow the same selection mechanism? These questions have yet to be systematically discussed.
This study conducts a textual analysis of written materials from the Executive Yuan of Taiwan to understand the distribution of sources for the government's policy learning over the past few years and to answer the abovementioned questions. The findings indicate that the selection mechanism for Taiwan's international policy learning targets predominantly follows an institutionalist approach, operating according to established organizational procedures and public servants' cognitive limitations and habits. This means that the selection often follows pre-existing networks established for other needs (such as political diplomacy, trade, etc.) or opts for regions where information is easily accessible (such as those that are easily reached or share the same language). This trend is consistent across various policy domains.
Although this study cannot determine whether the quality of current policy learning information is "good" or "bad" or whether it is sufficiently diverse, it is advisable to institutionalize the evaluation work of the current information network and integrate it into policy learning practices. In terms of research, it is recommended that changes in the information networks that the government relies on for policy learning over time be monitored and the impact of geographical factors considered.
 

政策學習、經驗汲取、資訊網絡、政府資訊行為

policy learning, lesson-drawing, information networks, government information behavior