Mortal Computation: A Foundation for Biomimetic Intelligence

Published: Nov. 17, 2023

This review motivates and synthesizes research efforts in neuroscience-inspired artificial intelligence biomimetic computing terms of mortal computation. Specifically, we characterize the notion mortality by recasting ideas biophysics, cybernetics, cognitive science a theoretical foundation for sentient behavior. We frame computation thesis through Markov blanket formalism circular causality entailed inference, learning, selection. The ensuing framework -- underwritten free energy principle could prove useful guiding construction unconventional connectionist computational systems, neuromorphic intelligence, chimeric agents, including organoids, which stand to revolutionize long-term future embodied, enactive cognition research.

Language: Английский

---

Persistent representation of a prior schema in the orbitofrontal cortex facilitates learning of a conflicting schema

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: March 1, 2025

Abstract Schemas allow efficient behavior in new situations, but reliance on them can impair flexibility when demands conflict, culminating psychopathology. Evidence implicates the orbitofrontal cortex (OFC) deploying schemas situations congruent with previously acquired knowledge. But how does this role affect learning of a conflicting behavioral schema? Here we addressed question by recording single-unit activity OFC rats odor problems identical external information orthogonal rules governing reward. Consistent schema formation, representations adapted to track underlying rules, and both performance encoding was faster subsequent than initial problems. Surprisingly however, rule reward changed, persistent representation prior correlated acquisition new. Thus, not source interference instead supported accurately independently representing old as acquired.

Language: Английский

---

A Broken Duet: Multistable Dynamics in Dyadic Interactions

Entropy, Journal Year: 2024, Volume and Issue: 26(9), P. 731 - 731

Published: Aug. 28, 2024

Misunderstandings in dyadic interactions often persist despite our best efforts, particularly between native and non-native speakers, resembling a broken duet that refuses to harmonise. This paper delves into the computational mechanisms underpinning these misunderstandings through lens of Lorenz system—a continuous dynamical model. By manipulating specific parameter regime, we induce bistability within equations, thereby confining trajectories distinct attractors based on initial conditions. mirrors persistence divergent interpretations result misunderstandings. Our simulations reveal differing prior beliefs interlocutors misaligned generative models, leading stable yet states understanding when exposed same percept. Specifically, speakers equipped with precise (i.e., overconfident) priors expect inputs align closely their internal thus struggling unexpected variations. Conversely, imprecise less confident) exhibit greater capacity adjust accommodate unforeseen inputs. results underscore important role models facilitating mutual establishing shared narrative) highlight necessity accounting for multistable dynamics interactions.

Language: Английский

---

Cohabitation of Intelligence and Systems: Towards Self-reference in Digital Anatomies

Published: July 15, 2024

Language: Английский

---

Answering, fast and slow: Strategy enhancement of visual understanding guided by causality

Neurocomputing, Journal Year: 2024, Volume and Issue: unknown, P. 128735 - 128735

Published: Oct. 1, 2024

Language: Английский

---