Mahault Albarracin Ph.D.

A research post discussing that changing beliefs has a thermodynamic cost, and this cost can be predicted, linking to a pre-print paper.

Co-inventor on a patent related to specifying active inference-based agents using natural language.

Co-authored a paper on predicting trajectories of objects, particularly under occlusion, a crucial task in autonomous driving.

Proposes an active inference account of self-esteem, conceptualizing it as an inferential capacity to interpret one’s standing within a social group, linking individual well-being to social interactions.

Provides key steps towards a computational phenomenology of collective action under a shared goal, which may help in understanding and designing multi-agent systems and their collaboration.

Explores the connection among sustainability, resilience, and well-being within the framework of active inference, defining sustainability as the enduring capacity to adapt to environmental changes.

Presents a framework for developing human-interpretable, explainable AI systems based on active inference and the free energy principle, focusing on transparent introspection and decision-making.

Advances computational phenomenology by integrating Husserlian phenomenology with active inference to understand social action based on shared goals and cognitive processes.

Explores the formation of epistemic communities and echo chambers by modeling confirmation bias through the active inference framework, showing how agents' self-confirming beliefs reinforce ingrained ideas.

Conceptualizes and formalizes resilience using active inference, investigating it as inertia, elasticity, and plasticity in adaptive systems.

Examines gender as a dynamic negotiation shaped by cultural scripts and affordances, integrating enactive-ecological predictivism to understand gender fluidity.