This paper presents recent developments within the INSIGHT project, which aims to provide causal explanations for perceptual events experienced by a robot during an HRI mission. We analyze the key requirements for conducting this research in simulated and real scenarios. The CORTEX architecture is used to integrate the necessary elements, memories and modules, to test different theoretical approaches and a wide set of experiments. Two new memory modules, semantic and episodic memories, have been incorporated alongside the existing working memory and internal simulation module. We describe how these components interact to support both bottom-up reactions to events and top-down responses aimed at improving the internal model and reducing the likelihood of similar unforeseen situations.
Ballesteros-Jerez et al. (2025) studied this question.