A new arXiv paper proposes a unified theoretical framework for understanding agency by grounding it in temporal organization, relational biology, and process ontology. The framework distinguishes between autonomy, goal-directedness, agency, and open-endedness through formalized timescale analysis, with implications for understanding biological systems, synthetic life, and artificial intelligence.
This academic paper addresses a fundamental problem in philosophy and cognitive science: how agency emerges from material organization. Rather than treating agency as a binary property, the authors develop a graded framework that maps organizational complexity to increasingly sophisticated forms of agency, from chemical systems to semantically closed agents capable of anticipatory behavior.
The theoretical contribution centers on temporal parametrization—the insight that self-referential closure cannot be properly understood without considering distinct timescales at which different processes operate. When these timescales desynchronize, they create what the authors call an out-of-sync dependency structure, mathematically redescribable as an Asynchronous Dynamic Bayesian Network. This temporal lens enables principled distinctions previously conflated in the literature: autonomy emerges from precarious closure under material openness, goal-directedness from viability maintenance, while agency proper requires endogenous anticipatory structures that modulate organism-environment coupling.
The framework bridges several competing theoretical traditions—Rosennean anticipation, organizational closure theory, and computational enactivism—while restricting concepts like Markov blankets and active inference to derived formal redescriptions rather than fundamental principles. This move challenges prevailing approaches in computational cognitive science that treat these concepts as foundational.
For researchers in AI and synthetic biology, this work provides conceptual scaffolding for building systems with genuine anticipatory capabilities rather than mere reactive adaptation. The hierarchical organization from proto-agential chemistry to fully semantically closed agents offers a roadmap for understanding what distinguishes living systems from current artificial systems, potentially informing development of more robust autonomous agents and synthetic lifeforms.
- →Agency is grounded in temporal organization of material systems, requiring distinct timescales for different constitutive processes.
- →The framework distinguishes autonomy, goal-directedness, agency, and open-endedness as discrete organizational levels rather than properties on a single spectrum.
- →Anticipatory structures emerge when organizations acquire endogenous mechanisms to modulate environment coupling based on possible futures.
- →The theory constrains Markov blankets and active inference to formal redescriptions rather than first principles, challenging computational frameworks in cognitive science.
- →The hierarchical model extends from chemical systems to synthetic organisms, providing theoretical guidance for artificial life and AI development.