Department of Philosophy, Universitat Autònoma de Barcelona

"Janus Cognition: Bio ↔ Xeno Round-Trip (“Xeno” = non-human cognitive systems such as LLM agents, neuromorphic controllers, and robot swarms.)"

ABSTRACT

This keynote presents Janus Cognition: a round-trip design strategy that pairs each bio-inspired cognitive model with a deliberately xeno (non-human) counterpart and requires both to satisfy the same cognitive contract (assumptions → guarantees). For core faculties—attention, memory, inference, social repair—we shuttle ideas from biology to xeno systems and back, using contracts, invariants, and counterfactual stress-tests that travel across embodiments (humans, LLM agents, neuromorphics, swarms). The aim is to replace anthropomorphic assumptions with verifiable properties—including xenoprobes (tests unfriendly to human priors)—so robustness is demonstrated rather than presumed. Attendees leave with a small library of dual-design patterns, cycle-consistency checks for cognitive claims, and a reviewer’s checklist to evaluate cognition across biological and synthetic agents..

SPEAKER'S BIOGRAPHY

Jordi Vallverdú (ICREA / Universitat Autònoma de Barcelona) is a philosopher of computer science and cognition whose work bridges cognitive science, artificial intelligence, and formal methods. He studies how plural “ways of knowing” can be operationalized and tested in real systems, from human cognition to LLM agents, neuromorphic hardware, and biohybrids. He has authored and edited numerous books and articles on cognition, AI, and epistemology, and advises interdisciplinary projects in computing, health, and robotics. His current agenda develops Janus Cognition—bidirectional models that carry guarantees from biological systems to synthetic agents, and back.