- 學位論文
訓練基模生成與存取之解題環境的認知負載控制
Cognitive Load Control of Problem Solving Environments that Provide Trainings for the Development and Use of Schemas
摘要
近代心理學家藉由具有感官記憶、工作記憶、長期記憶等結構的訊息處理模型來了解人類對訊息從輸入到輸出之間轉換過程的心理事件。以訊息存取的角度來看記憶(REMEMBERING) ，訊息經由紀錄歷程編碼儲存在長期記憶達成學習新知的效果、經由提取歷程可以將先前習得的訊息活化至工作記憶中提供問題解決等心智活動使用。 基模是敘述性知識表徵的整合單位，基模的形成需要將不同例子同時放在工作記憶中比較異同，而基摸一旦形成可以幫助學習者新知的學習及問題的解決。但回憶先前習得的訊息會隨著時間過去而變得更加困難，這會導致之前儲存的基模例子無法順利的提取至工作記憶中而阻礙基模的生成。本篇論文我們以記憶視窗的概念來描述這件事。 在自然環境中事件是隨機出現的，反覆的作相同的事情也會導致厭煩感，因此我們的目的是希望能幫助學習者基模型成時在不超過他的記憶視窗範圍內盡量的拉大各基模例子間的距離。在人為的解題環境裡我們提出測量學習者學習努力變化的方法來偵測學習者提取各基模例子時是否正常，透過控制同時訓練的問題種類數N來導正各個基模例子能落在學習者的記憶視窗內。 在驗證的部分本控制方法是實作於原有的電腦輔助學習系統上，系統的設計理念為類比解題環境與鷹架輔助訓練，本論文擬定了一個實驗的前導設計，目前實驗正在進行中，我們預測使用我們方法訓練的受測者對已知知識的使用會較具有延伸性。
並列摘要
In recent years, psychologists have used the Model of Information Processing in trying to understand the changes of information from input to output in human mental activities. A typical information-processing model would consist of sensory memory, working memory and long-term memory. From this, remembering may be separated into coding processes and retrieval processes. New information is first coded by the coding processes in the working memory and then stored in the long-term memory. But this is only part of the learning process. Later, the retrieval processes retrieve the stored information from the long-term memory, “activate” it by bringing it into working memory, so as to be used by other mental processes such as problem solving processes. A schema is an internal form of declarative knowledge representation. The formation of a schema requires that we compare various similar “fragments of experience”, filter out their differences, and abstract their commonalities. Schemas form when the learner is actively constructing his/her knowledge. This happens when the learner tries to solve problems. It may also happen when new information comes in. As time elapses, the difficulty of recalling the stored information (the “fragments of experience”) would increase. Eventually, this may cause problems in retrieving previously stored schema instances into the working memory, thereby impeding schema formation. In this thesis, we use the concept of Memory Window to model this mental phenomenon. In the real problem-solving environment, many varieties of events occur. Also, we all know that if we are required to do the same thing over and over again, we will be fed up. And so, in helping the learner to do schema formation in our experiment, our purpose is to arrange things so that the width of the Memory Window can be as wide as possible in accommodating the recalled schema instances. Specifically, what we provide to the learner is an artificial problem-solving environment. In this environment, we use a method to detect whether the retrieval of schema instances is successful or not by measuring changes in the learner’s learning efforts. By controlling the number of problem domain types N, we hope to achieve the effect that each schema instance that the learner recalls would fall within the learner’s Memory Window. To test the effectiveness of our method, we implement it in a Computer Assisted Learning system. The system advocates analogical problem solving, and uses different kinds of hints to scaffold the learner. At the point of writing this thesis, the experiment is still on going. Our contention is that knowledge acquired through the proposed learning methodology would be more extensible in that the learner would be more capable of doing knowledge transfer.