前置意圖與因果理解對幼兒觀察學習他人錯誤的影響

The Influences of Prior Intention and Causal Understanding on Preschoolers' Observational Learning from Others' Mistakes

過去研究採用陷阱管子作業發現，錯誤示範有助於三歲幼兒洞察工具的因果結構，因而更成功地以模仿方式來解決問題（Want & Harris, 2001）。本研究進一步排除示範階段的因果訊息，探討30個月和41個月大的幼兒（計116位）能否從實驗者的錯誤示範，察覺意圖達成的目標，而增進觀察學習工具使用的表現。一些幼兒觀看實驗者示範使用工具棒子從透明或不透明管中取出目標物─先誤將物件推入陷阱，再從正確管口插入棒子，進而成功取出目標物。在不透明情境下，幼兒沒有機會看到工具和管子的因果機制，只依賴實驗者的錯誤作為意圖指引。另一些幼兒在未觀看任何示範下，自發性地探索作業，或觀看實驗者以棒子示範與取物目標無關的控制動作。結果發現：不論觀看透明或不透明示範，幼兒雖傾向於以模仿方式取物，但無證據顯示他們對因果機制有所掌握；當考慮幼兒結合模仿方式與仿效方式成功取物的總表現，較大幼兒在不透明情境的總表現最為成功。錯誤對觀察學習的增進效果不僅是前置意圖或因果理解所致，較大幼兒能根據情境中的行為、心理及因果訊息能見度變化，選擇多樣的社會學習策略。

Previous research has shown that 3-year-old children benefit from observing a model's failure to retrieve a reward in a trap-tube task before they are shown a correct solution (Want & Harris, 2001). In this study, we investigated whether the tendency to benefit from mistakes is influenced by the availability of causal information or the model's intention. In two demonstration conditions, 30- and 41-month-old children (N = 116) observed an experimenter trap the reward before demonstrating how to obtain it with a tool from either a clear or an opaque tube. The clear tube made the causal properties of the task visible, whereas the opaque tube prevented them from being seen. In two other conditions, children were allowed to explore the task spontaneously or watched the experimenter manipulate the tool using actions irrelevant to retrieving the reward. Results indicated that both 30- and 41-month-old children imitated the demonstrated tool use regardless of the transparency of the task, but they did not show a bias in favor of the correct solution. However, when considering overall performance (reward retrieval achieved via both imitative and emulative strategies), 41-montholds were successful in retrieving, as opposed to trapping, the reward in the opaque condition. The facilitating effect of observing others' mistakes is unlikely to be due solely to prior intention or causal understanding. We suggest that children flexibly vary the strategy they used to reproduce a model's actions and action outcomes depending on what they perceive as relevant in the situation of the task.