Infants’ Observational Learning through Ghost Conditions: The Roles of Agency and Affordance Learning

發展心理家發現嬰兒在沒有人只剩下物體運動的示範情境中，也能有效的學習物體操作，因此認為意圖模仿並非其中必要的歷程，嬰兒可以透過結果狀態仿效、物體效應學習、物體運動重演……等一些非模仿的社會學習歷程學習物體操作。然而，越來越多證據指出，嬰兒有理解非人類物體行為目標的能力：根據線索基礎論的學者指出，當主事者生命性的動作形態具備自我發動性、變異性等行為線索，以及鮮明的行為效果，嬰兒就有能力歸因其目標。目的推理論也指出，嬰兒根據主事者的行為、環境的限制、行為的結果狀態表徵主事者目標導向的行為，推理主事者行為的目標。就此理論角度而言，單純觀看物體運動示範的嬰兒，也可能透過目標歸因，或目標推理的歷程，重演目標動作。本實驗設置障礙情境，在無人示範情境中，操弄自發運動體的行為線索，將自發運動體分為具有行為動因，且行為合理的主事者與不具行為動因，且行為不合理的主事者，觀察16-18個月的嬰兒分別觀看兩類自發運動體的完整示範或失敗嘗試後重演目標動作的表現。沿續Huang與Charman（2005）的研究，進一步探討嬰兒在無人示範情境中的模仿表現是否受自發運動體的動作形態與行為效果影響，亦或是單純因為物體動力特質引發的物體效應學習。我們分析嬰兒觀看示範後第一個與20秒內的動作反應是否重演物體示範的目標動作，並且記錄嬰兒達成目標動作所花費的時間，與達成目標動作前探索物體的動作反應，結果發現，自發運動體的似生命性的運動線索與行為效果，都有效提升嬰兒重演目標動作，而且只有當主事者的動作展現生命性並且伴隨鮮明結果組態時，嬰兒重演目標動作的表現會最好。顯示，在無人示範的情境中，歸因主事者的目標是嬰兒觀察學習物體操作的關鍵歷程。Development psychologists have shown that infants can learn how to manipulate objects in ghost conditions where only the object made to move during demonstration are observed. Ghost conditions provide a variety of information that induces infants to engage in non-imitative social learning of different kinds, such as end-state emulation, affordance learning and object movement reenactment. However, there is increasing evidence that infants can infer goal-directed action produced by non-human object agents. According to the cue-based theory, infant can attribute a goal to an object from several behavioral cues, such as self-propelledness, equifinal information, and the salient action effect. As stated in the recent theory of teleological reasoning, infant can infer a behavior goal-directedness based on object’s ability to adjust its own behavior when adapting to environmental constraints. Under this interpretation, infants engage in goal emulation by attributing a goal to the behavior of the self-propelled object. In the present study, we examine how behavior features of a self-propelled objects and its action effect influence infants’ observational learning of object manipulation. The data was seem to show that infants reproduce the target action more often when behavior cues could be used to identify the object as an agent. The animate motion and the salient action effect can both improve their performance in reproduce the target action. This in turn suggest that goal emulation plays a role in the observational learning in a ghost condition