The Evolutionary Rarity of Reflective Thought: The Moment the Mind Looked at Itself
Daniel de Kooter
Figshare September 12, 2026 preprint DOI: 10.6084/m9.figshare.33176537.v1 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Theoretical or philosophical paper |
|---|---|
| Key points | Argues that reflective cognition—the ability to observe and reason about one's own cognitive operations—is a recent, rare, metabolically expensive evolutionary addition rather than the foundation of intelligence, and that the transition from operational to reflective cognition is a phase transition requiring a persistent self-model, temporal decoupling, and recursive meta-representation. The authors contend no species has been documented with partial reflective capacity, and that efforts to replicate reflective cognition in artificial systems misunderstand what makes autonomous agents effective. |
Abstract
Cognition across biological organisms operates in two layers. The first layer—operational cognition—encompasses all processes by which an organism perceives, evaluates, and responds to its environment. It is universal among autonomous biological agents and does not require consciousness. The second layer—reflective cognition—is the capacity to observe, abstract from, and reason about one’s own cognitive operations. Reflective cognition is not the foundation of intelligence but a recent, rare, and metabolically expensive evolutionary addition that emerged under specific environmental pressures. By examining the dolphin, the great apes, and the archaeological record of the Neanderthal, we trace the boundary between these two layers and specify the minimal architectural conditions—persistent self-model, temporal decoupling, and recursive meta-representation—required for the transition from operational to reflective cognition. The threshold between the two layers is not a gradient but a phase transition: no species has been documented with partial reflective capacity. The current focus on replicating reflective cognition in artificial systems represents a fundamental misunderstanding of what makes autonomous agents effective.