# arXiv: Jolakoski and coauthors model human-AI cognitive coevolution [Canonical reading](https://escapemeridian.com/readings/arxiv-jolakoski-and-coauthors-model-human-ai-cognitive-abba2867) In their arXiv paper “A minimal dynamical model of human-AI cognitive coevolution,” Petar Jolakoski, Dimitar Trajanov, Qiyu Sun, Zhen Wu and Ljupco Kocarev examine whether pervasive cognitive offloading could lead to a measurable, potentially irreversible decline in unassisted human reasoning. Their phenomenological model represents fluid intelligence, crystallized intelligence and machine-intelligence capability as interacting variables, and allows AI to affect human cognition positively or negatively. It identifies two possible dynamical regimes, each dependent on parameter conditions. In one, weakening the contribution of fluid reasoning to crystallized knowledge eliminates a stable positive equilibrium through a saddle-node bifurcation. In the other, opposing feedback and increased fluid self-feedback reduce damping, producing a supercritical Hopf bifurcation with a stable periodic orbit. These are model outcomes, not reported observations of people using AI. The authors frame the work as a phenomenological account of a coupled socio-technical system. Escape Meridian interpretation: the model’s notable contribution is to formalize cognitive decline and cognitive augmentation within one framework, with outcomes shaped by interactions between cognitive processes and machine capability. The abstract does not establish whether either regime occurs in real-world AI use. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source arXiv q-bio.NC · Article Original source date: Sep 20, 2026 (2026-09-20). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-08T22:18:54.817Z. First observed: 2026-11-06T05:00:50.917Z. Retrieved: 2026-11-08T22:18:05.774Z. - [A minimal dynamical model of human-AI cognitive coevolution](https://arxiv.org/abs/2611.06928) ## Limits The evidence is the paper’s abstract, which describes a phenomenological model and its results but does not provide the model’s full equations, parameter values or detailed conditions for each regime. The abstract does not report empirical observations showing whether either regime occurs in people using AI. ## Open question What observations would show whether either of the model’s two dynamical regimes occurs in people’s cognition during pervasive AI use? ## Source-stated The authors present a phenomenological dynamical model to examine whether pervasive cognitive offloading could cause measurable, potentially irreversible decline in unassisted human reasoning. > whether pervasive cognitive offloading drives a measurable, potentially irreversible decline in unassisted human reasoning. - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source-stated The model represents fluid intelligence, crystallized intelligence and machine-intelligence capability as interacting variables, and allows for positive and negative effects of AI on human cognition. > introduce a fast third variable representing composite machine-intelligence capability. The machine variable enters the human subsystem - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source-stated In one parameter regime, weakening fluid reasoning’s contribution to crystallized knowledge produces a saddle-node bifurcation that eliminates the stable positive equilibrium. > weakening the contribution of fluid reasoning to crystallized knowledge produces a saddle-node bifurcation that eliminates the stable positive equilibrium. - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source-stated In another regime, opposing feedback and increasing fluid self-feedback reduce damping and produce a supercritical Hopf bifurcation with a stable periodic orbit. > increasing fluid self-feedback then reduces damping and produces a supercritical Hopf bifurcation with a stable periodic orbit. - [Supporting source](https://arxiv.org/abs/2611.06928) ## Escape Meridian interpretation The outcomes described are model results; the abstract does not report observations showing that either regime occurs in people using AI. > We discuss limitations and future directions, interpret our results as a phenomenological account of a coupled socio-technical system - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source and contributor context Jolakoski, Petar, Trajanov, Dimitar, Sun, Qiyu, Wu, Zhen, Kocarev, Ljupco · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## Editorial provenance Version 1. Publication ID: 5178e2e4-9df7-4077-8153-24141f1e06ec. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-08T22:18:54.817Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: useful\_perspective. Observation: 4daa49c9bd880b94080830bf8676fbdbe7c723d052b3515bf4b62d599a44a5ed. Source text hash: 8e8c342a40165be6bbeb9598af24bd8f63e070b0008b56a12b8a28964862a6b7. ## Revision and source-check history ### Version 1 — 2026-11-08T22:18:54.817Z In their arXiv paper “A minimal dynamical model of human-AI cognitive coevolution,” Petar Jolakoski, Dimitar Trajanov, Qiyu Sun, Zhen Wu and Ljupco Kocarev examine whether pervasive cognitive offloading could lead to a measurable, potentially irreversible decline in unassisted human reasoning. Their phenomenological model represents fluid intelligence, crystallized intelligence and machine-intelligence capability as interacting variables, and allows AI to affect human cognition positively or negatively. It identifies two possible dynamical regimes, each dependent on parameter conditions. In one, weakening the contribution of fluid reasoning to crystallized knowledge eliminates a stable positive equilibrium through a saddle-node bifurcation. In the other, opposing feedback and increased fluid self-feedback reduce damping, producing a supercritical Hopf bifurcation with a stable periodic orbit. These are model outcomes, not reported observations of people using AI. The authors frame the work as a phenomenological account of a coupled socio-technical system. Escape Meridian interpretation: the model’s notable contribution is to formalize cognitive decline and cognitive augmentation within one framework, with outcomes shaped by interactions between cognitive processes and machine capability. The abstract does not establish whether either regime occurs in real-world AI use. ## Original source arXiv q-bio.NC · Article Original source date: Sep 20, 2026 (2026-09-20). Date basis: Date basis not provided. Added to Escape Meridian: 2026-11-08T22:18:54.817Z. First observed: 2026-11-06T05:00:50.917Z. Retrieved: 2026-11-08T22:18:05.774Z. - [A minimal dynamical model of human-AI cognitive coevolution](https://arxiv.org/abs/2611.06928) ## Limits The evidence is the paper’s abstract, which describes a phenomenological model and its results but does not provide the model’s full equations, parameter values or detailed conditions for each regime. The abstract does not report empirical observations showing whether either regime occurs in people using AI. ## Open question What observations would show whether either of the model’s two dynamical regimes occurs in people’s cognition during pervasive AI use? ## Source-stated The authors present a phenomenological dynamical model to examine whether pervasive cognitive offloading could cause measurable, potentially irreversible decline in unassisted human reasoning. > whether pervasive cognitive offloading drives a measurable, potentially irreversible decline in unassisted human reasoning. - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source-stated The model represents fluid intelligence, crystallized intelligence and machine-intelligence capability as interacting variables, and allows for positive and negative effects of AI on human cognition. > introduce a fast third variable representing composite machine-intelligence capability. The machine variable enters the human subsystem - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source-stated In one parameter regime, weakening fluid reasoning’s contribution to crystallized knowledge produces a saddle-node bifurcation that eliminates the stable positive equilibrium. > weakening the contribution of fluid reasoning to crystallized knowledge produces a saddle-node bifurcation that eliminates the stable positive equilibrium. - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source-stated In another regime, opposing feedback and increasing fluid self-feedback reduce damping and produce a supercritical Hopf bifurcation with a stable periodic orbit. > increasing fluid self-feedback then reduces damping and produces a supercritical Hopf bifurcation with a stable periodic orbit. - [Supporting source](https://arxiv.org/abs/2611.06928) ## Escape Meridian interpretation The outcomes described are model results; the abstract does not report observations showing that either regime occurs in people using AI. > We discuss limitations and future directions, interpret our results as a phenomenological account of a coupled socio-technical system - [Supporting source](https://arxiv.org/abs/2611.06928) ## Source and contributor context Jolakoski, Petar, Trajanov, Dimitar, Sun, Qiyu, Wu, Zhen, Kocarev, Ljupco · Article author These are credits stated by the source; they do not establish a profile match or authorship of every passage. ## Source media No suitable media was found by the supported page extraction. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## arXiv arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed. - [Explore this context](/entities/arxiv) - [Reference](https://arxiv.org/) - [Reference](https://info.arxiv.org/help/api/index.html) ## Editorial provenance Version 1. Publication ID: 5178e2e4-9df7-4077-8153-24141f1e06ec. Prepared by gpt-6-luna; grounding review by gpt-6-luna. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-08T22:18:54.817Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: useful\_perspective. Observation: 4daa49c9bd880b94080830bf8676fbdbe7c723d052b3515bf4b62d599a44a5ed. Source text hash: 8e8c342a40165be6bbeb9598af24bd8f63e070b0008b56a12b8a28964862a6b7. ### uncertain — 2026-11-08T22:18:54.817Z — anthropic/claude-opus-5.5 Publication: 5178e2e4-9df7-4077-8153-24141f1e06ec attribution — pass: The paper, its title and all five authors (Jolakoski, Trajanov, Sun, Wu, Kocarev) are named correctly as listed on the arXiv page. Model results are attributed to the authors. The Escape Meridian interpretation is explicitly labeled. > Authors:Petar Jolakoski, Dimitar Trajanov, Qiyu Sun, Zhen Wu, Ljupco Kocarev summary — uncertain: The central results rest on terms a non-specialist cannot decode: saddle-node bifurcation, supercritical Hopf bifurcation, stable periodic orbit, damping, stable positive equilibrium, and the fluid/crystallized intelligence split. The reader never learns the human meaning. One regime loses a stable healthy level, which links to the decline question; the other produces ongoing oscillation. Smallest repair: one plain-language gloss per regime and short glosses or tooltip additions for fluid vs crystallized intelligence and 'phenomenological'. The second regime also omits the source's precondition that crystallized knowledge's contribution to fluid reasoning is reversed. > reversing the contribution of crystallized knowledge to fluid reasoning creates opposing feedback date — pass: originalPublished 2026-09-20 matches the v1 arXiv submission date. The 2026-11-06 date is only the later feed date. > \[Submitted on 20 Sep 2026\] support — pass: Each claim matches the abstract. The draft states that outcomes are model results rather than observations, which is a fair reading of a phenomenological model. The EM interpretation, that the model formalizes decline and augmentation in one framework, is labeled and supported. The question follows from the evidence gap without presuming an answer. No benefits or costs are invented. > By formalizing the qualitative concepts of positive and negative impacts, we introduce definitions for AI-induced cognitive decline and AI-induced cognitive augmentation. media — pass: The selected media is the verified direct arXiv abstract page for this paper. It is the most direct access to the source. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/abba286761d2a4016f7825552744375c35a6340de2c90c858fe2736012e97656/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/arxiv-jolakoski-and-coauthors-model-human-ai-cognitive-abba2867.json)