在經濟學與行為科學裡，「學習」是個重要的主題。經由適當的學習，人類個體的行為往往能夠從不理性趨向於理性。而在這當中，「頓悟學習」是一項特殊且值得探究的學習法則。不同於一般循序漸進的過程，頓悟學習指的是人類在一段醞釀期後突然習得新行為的現象。相較於更熱門的「強化學習」，此類突發性的行為相當不易研究，也常被學界忽視。換言之，這裡面還有廣大的研究潛力亟待開發。綜觀迄今為止的行為研究雖已在多種實驗任務中驗證了頓悟學習的存在，我們依舊尚未得知頓悟學習的現象有多普遍，也不知道預測頓悟的方法。另一方面，神經科學中頓悟學習所使用的腦區或神經網路的證據大多為相關性而非因果性，且其研究往往受限於特定實驗任務，無法確定其推廣性，並缺乏與頓悟學習的計算模型連結的證據。因此，我希望能設法彌補這些不足。本計畫預計結合行為、眼動、跨顱直流電刺激以及功能性磁振造影等多種實驗方法，以進一步了解頓悟學習的一般性、解碼其神經機制，並用我提出的計算模型來擬合頓悟學習的神經數據。我預計將組成一個結合經濟學家、心理學家與神經科學家的跨領域研究團隊以進行上述研究。 ;People often fail to act fully rationally, but learning will in general lead to more rational behavior. Therefore, learning is an important theme in economics and behavioral science. Epiphany learning (EL) is one of the major learning rules, in which people will suddenly learn a new behavior after a period of incubation. However, due to its abrupt nature, EL is not easy to study and thus heavily understudy compared to the more popular reinforcement learning rule. In behavioral study, past research have shown the existence of EL in various experimental tasks, but we still do not know how general EL is, and how to predict epiphany. Moreover, past neuroscience evidence on the EL brain region or neural network were mostly correlation, not causation, subjected to specific types of experimental task and lacked a computational model to describe EL. This proposal takes on the challenge to find the robustness of EL behavioral study, decode the neural mechanism of EL and fit the neural data with the computational model I proposed. I plan to build a multidisciplinary research team consists of economists, psychologist and neuroscientist and combine research methods such as behavioral, eye-tracking, tDC and fMRI experiment to fulfill this challenge.
