Grounded World Models for Higher Layers of Meaning

Stefano de Giorgis

Post-doc researcher at Institute for Cognitive Sciences and Technologies – National Research Council (ISTC-CNR), Italy

The project involves knowledge representation techniques, neuro-simbolic AI, and cognitive semantics. According to the embodied cognition principles, individuals construct a cognitive representation and conceptualization of the external world based on their perceptual capabilities, according to their affordances, goals, and ultimately their sensori-motor perception. Consequently, decisions and behaviors stem from this internalized model. The notion of “Grounded Wold Models” elaborates on Jay Wright Forrester “mental model” definition, and is here defined as Cognitive Representations of (a certain portion of) the world, grounded in (i) embodied experience of a subject cognizer, and (ii) external representable and measurable multimodal data. The first purpose of this research is therefore to further extend the current state of the art on embodied cognition pillars, and in particular on the formalization of “image schemas”. The second pivotal pillar of this research is the topic of moral values. Moral and cultural values shape human’s daily behavior, influencing social norms, grounding legal systems, and fostering collective behavioral nudges in everyday life. The aim of this project strongly relates to the Value Alignment problem, adopting a “theory-heavy” approach, namely considering existing theoretical frameworks such as Moral Foundations Theory, Basic Human Values, and Moral Molecules theory, formalized and re-engineered as knowledge bases, injecting this formal knowledge into neuro-symbolic AI architectures to develop morally an embodied aware grounded world models. The final result aims at grounding moral values theories into embodied cognition, represented in a formal way as ontologies and knowledge bases, guiding decision systems using knowledge-informed learning architectures.

Keywords: Knowledge Representation, Moral Values, Embodied Cognition, Neuro-symbolic AI

Scientific area: Knowledge Representation, Artificial Intelligence, Cognitive Semantics

Bio: I received a PhD in Digital Humanities and Knowledge Representation at University of Bologna. Currently I’m a post-doc researchet at the Institute for Cognitive Sciences and Technologies at National Research Council (ISTC-CNR), Italy. My main research area includes AI, Knowledge Representation and Cognitive Semantics. I’m trying to survive through the XXI catastrophic century doing research in AI, and to live thanks to travels, books, movies, many doubts and black humor.

Visiting period: May 2024 at Jönköping Artificial Intelligence Laboratory, Sweden