Interoceptive Machine Framework: Bio-Inspired AI for Adaptive Autonomy
A recent study published on arXiv (2604.24527) introduces the interoceptive machine framework, which adapts concepts of biological interoception into computational designs for artificial intelligence. This framework categorizes interoceptive elements into three key principles: homeostatic (regulating internal viability), allostatic (reassessing based on anticipatory uncertainty), and enactive (generating data through active interaction). These principles serve as theoretical guidelines aimed at enhancing the development of artificial agents, enabling better self-regulation and behavior that is sensitive to context, rather than providing direct mappings to neurophysiological processes.
Key facts
- Paper titled 'Interoceptive machine framework: Toward interoception-inspired regulatory architectures in artificial intelligence'
- Published on arXiv with ID 2604.24527
- Proposes an integrative framework grounded on interoception and embodied AI
- Framework organizes interoceptive contributions into three functional principles: homeostatic, allostatic, and enactive
- Homeostatic principle focuses on internal viability regulation
- Allostatic principle involves anticipatory uncertainty-based re-evaluation
- Enactive principle involves active data generation through interaction
- Principles are abstractions, not direct neurophysiological mappings
Entities
Institutions
- arXiv