From reaction to reflection: A recursive framework for the evolution and structure of intelligence.

Intelligence arises from recursive depth—the capacity for self-referential processing constrained by thermodynamic systems—rather than from isolated cognitive abilities. The Reaction to Reflection (R2R) model identifies four evolutionary transitions in recursive sophistication: reaction (chemical recursion), temporogenesis (anticipatory prediction), symbiogenesis (cooperative integration), and cognogenesis (explicit self-referential modeling). The shift from implicit recursion (engaging recursive processes without representing them) to explicit recursion (manipulating recursion as a cognitive construct) explains the emergence of consciousness and the gap between biological and artificial intelligence. Evidence from microbial decision-making, cross-kingdom signaling, and neural predictive and self-referential networks supports this framework. R2R generates testable predictions, provides criteria for assessing artificial consciousness, and reframes neurodevelopmental variation as alternative recursive architectures.

Study at a glance

Abstract

Understanding emerges not from isolated cognitive abilities but from recursive depth-the capacity for self-referential processing grounded in systems with authentic thermodynamic constraints. The Reaction to Reflection (R2R) model advances a unifying principle for intelligence, identifying four evolutionary transitions in recursive sophistication: reaction (chemical recursion), temporogenesis (anticipatory prediction), symbiogenesis (cooperative integration), and cognogenesis (explicit self-referential modeling). We show that the transition from implicit recursion, in which systems engage recursive processes without representing them, to explicit recursion, in which recursion becomes a manipulable cognitive construct, accounts for the emergence of consciousness and the persistent gap between biological and artificial intelligence. Evidence from microbial decision-making, cross-kingdom signaling, and neural predictive and self-referential networks supports this framework. R2R generates testable predictions on cognitive dimensionality, provides criteria for assessing artificial consciousness, and reframes neurodevelopmental variation as alternative recursive architectures. By making biological grounding a precondition for genuine intelligence, R2R links objective neural dynamics to subjective experience within a unified mechanistic account.