Infants Attribute Goals to Acts Produced by Human and Object Agents

A debate has been going as to whether infants’ ability to represent acts in terms of goals is sensitive to motions performed by people only (the mirror neuron systems theory) or to a variety of cues through which infants apply the principle of rational action to both human and inanimate agents (the teleological stance theory). Solution to the above issue is timely and important (for notable examples, see Sommerville, Woodward and Needham, 2005 ; Biro and Leslie, 2006). Robotic scientists have begun to use such knowledge to design robots that mimic human biomechanical movements, to which infants are likely to attribute goals or intentions (Kamewari, Kato, Kanda, Ishiguro and Hiraki, 2005). Using an eye tracking technique (Tobii 1750), the study reported here presents an attempt to evaluate the mirror neuron systems theory (Gallese, Fadiga, Fogassi and Rizzolatti, 1996; Gallese, Keysers and Rizzolatti, 2004) and the teleological stance theory (Gergely and Csibra, 2003; Gergely, Nadasdt, Csibra and Biro, 1995). The key question is whether 12-month-old infants were as likely to predict the goal-state of an action performed by a human agent as by an object agent under contexts of goal selection.