Effective, hyper-personalised learning within MaTHiSiS relies on the adaptation of parameters in the learning experience, based upon each learner’s affective state. The Affect Recognition Software serves as a basis of the recognition of the learners’ affect states. Its goal is to gather behavioural cues of the learner from the Platform Agents (PAs) - devices and apply innovative machine learning techniques in order to interpret them into comprehensive affective cues that tell the story of the learner’s uptake of the learning objective(s).