ABSTRACT
Since the year 2000, a number of researchers have suggested a developmental perspective on artificial intelligence (AI) and robotics. One of the fundamental methodological assumptions is that cognition is embodied, which means that it arises from bodily interactions with the world and that it is continuously meshed with them (Lungarella & Pfeifer, 2001). In other words, thinking emerges from real-life experiences, from sensorimotor coordinated interactions, and from exploration of the surrounding environment (Metta & Fitzpatrick, 2003). Since the complexity of the structure of living beings is much higher than the complexity of actual artifacts, only a few experiments can be done; nevertheless, the increase of this perspective is important. On one hand, psychologists can provide the detailed empirical findings and theoretical generalizations that can guide the implementations of robotic systems capable of cognitive development. On the other hand, these implementations can help clarify, test, and even develop psychological theories, which, due to the complexity of the interaction processes involved, often cannot be exhaustively tested (Berthouze & Prince, 2003). Robots can be used to instantiate and investigate models originating from developmental psychology, and it is necessary to design robotic systems with better autonomy, adaptability, and sociability by applying the insights gained from ontogenetic development studies. Psychological experiments can provide rich and subtle accounts of infant behavior as it changes with age, but it is more difficult to track and describe internal changes.
