Abstract: 
High motivation and positive emotions are important prerequisites for effective knowledge transfer, while negative emotions can impede the acquisition of new knowledge. This talk presents our current state of research in SCIoI’s project 06 that aims at identifying adaptive teaching strategies of humans, and to synthesize them with the goal to test whether these strategies can be applied successfully by synthetic agents. To this end, we conduct 3 experiments in which (a) we aim to build a model of adaptive teaching using an emotionally-and performance adaptive intelligent tutoring system (ITS) and to compare it to a cognitively-adaptive ITS in its effects on human emotional experience and learning performance; (b) we integrate the model of the emotionally-adaptive ITS in embodied agents (physical robots: Pepper and Cozmo) and compare its effectiveness in knowledge transfer to the traditional ‘on-screen’ form by examining humans’ learning outcomes across conditions; c) we test the adaptive mechanism in the social interaction of two robotic agents – a learner robot (Nao) and a tutor robot (Pepper) – in a scaffolding situation were both are equipped with simple emotional expression / recognition mechanisms. The talks provides an overview about our recent results, their implications and perspectives for future research.