For enabling the shift from 2D stimulus to 3D and VR, I first re-constructed Radboud Face Database emotional expressions through the state-of-the-art DECA model that uses deep learning for single image reconstructions. Next, using Meshlab and Blender for alignment and corrections, I created a novel 3D Emotion Database. In order to validate subjective perceptions of anger and fear emotions on the 3D faces from the created database, I setup an online experiment using PsyToolkit. The dataset is yet to be fully validated and published.

I then developed a prototype for combining emotional 3D face with the body pose using Blender. Further, I am collaborating for developing a novel experimental setup in Unity through Virtual Reality for neuroscientific experiments on approach and avoidance behavior in humans post validation of our 3D emotion database. During my stay with the Social Cognition Group, at the Laboratory of Cognitive and Computational Neuroscience, I also explored and implemented computational modelling of Reinforcement Learning in socio-emotional contexts from the behavioral data already available from previous lab experiments.