In this thesis, we study the development of linguistic concepts (corresponding to a subset of nouns, verbs and adjectives) on a humanoid robot. To accomplish this goal, we use affordances, a notion first proposed by J.J. Gibson to describe the action possibilities offered to an agent by the environment. Using the affordances formalization framework of Sahin et al., we have implemented a learning system on a humanoid robot and obtained the required data from the sensorimotor experiences of the robot. The system we developed (1) can learn verb, adjective and noun concepts, (2) represent them in terms of strings of prototypes and dependencies based on affordances, (3) can accurately recognize the concept of novel objects and events, and (4) can be used for tasks such as goal emulation and multi step planning.

This study investigated children's social learning of prosocial behaviors in a transgressor context. Two-to three-year-olds (24-47 months, N = 54) saw videos of an adult help another adult in distress by performing a novel prosocial action. Children were then led to believe that they had transgressed to cause their parent's pain and sadness. It was hypothesized that children in the experimental condition who watched the video and then transgressed would be more likely to perform the novel action (imitation) and to display non-demonstrated prosocial behaviors (goal emulation) relative to children in two control conditions: (a) children who did not view the video but transgressed and (b) children who viewed the video but witnessed a neutral interaction. Children in the experimental condition were no more likely to imitate or emulate than children in the control conditions, suggesting that children have difficulty applying socially learned prosocial behaviors in a transgressor context.