Une approche computationnelle des sciences de l'éducation : modélisation de la résolution créative de problèmes dans l'interaction enfant-robot par le biais de traces d'apprentissage de type comportamentale

International audience; In computational neuroscience and bioinspired artificial intelligence (AI), different studies aim at understanding the mechanisms of perception such as shape recognition and sensori-motor coordination such as grasping or pinching. A step further, as in the Mnemosyne research program, cognitive mechanisms are analyzed to better understand brain circuits responsible for reasoning and problem-solving by a biological or algorithmic agent. The ANR CreaMaker and AIDE research programs seek to improve understanding of creative problem solving via behavioral analysis of the relationship between brain processes and problem-solving activities. The CPS behavior (intentions + CPS regulation) is analyzed using learning analytics created automatically as well as human analysis based on coding schemes. #CreaCube is a large dataset of creative problem-solving challenges (n>1300 experiments). Each video is coded in accordance with a behavioral learning analytics coding schema. #CreaComp is a computational learning sciences research program dedicated to the improvement of the study of creative problem solving.