Cognitive Drift Papers is a collection of concept papers, guides, working papers, and essays by A. Jacobs examining how human thought can remain coherent and functional while becoming progressively less grounded in the conditions it is meant to understand and navigate. It forms the cognitive branch of the broader Reality Drift framework.
Cognitive Drift describes the progressive misalignment of human models, attention, interpretation, and judgment with external reality. Thought can continue to develop, update, and produce plausible explanations even as its corrective contact with the world weakens. The problem is not necessarily irrationality or incorrect information. A person can reason consistently from mediated inputs, inherited assumptions, and increasingly recursive representations without recognizing that the underlying frame has become less answerable to evidence or consequence.
The collection also examines Co-Cognition, the process through which thought is distributed across recursive exchanges among people and external systems. AI can help structure partial ideas, expand reasoning, and increase clarity, but a productive exchange is not automatically a grounded one. Co-Cognition names how thought is produced. Cognitive Drift names a possible trajectory of misalignment. They are related concepts, not synonyms.
Begin with 00_Start_Here for an introduction and reading path. The overview materials explain the distinction between cognitive coherence and grounding, while the canonical papers establish the definitions and boundaries of the central concepts.
00_Start_Here/
01_Overview_and_Guides/
02_Core_Concept_Papers/
03_Working_Papers/
04_Substack_Essays/
- 00_Start_Here contains introductory material and a suggested route through the collection.
- 01_Overview_and_Guides contains framework overviews and practical guides to AI-mediated cognition, recursive thought, and grounding.
- 02_Core_Concept_Papers contains canonical definitions and concept papers, including Cognitive Drift and Co-Cognition.
- 03_Working_Papers contains longer explorations and developing applications of the framework.
- 04_Substack_Essays collects essays that make the ideas accessible through contemporary examples and wider reflection.
The papers investigate what happens when cognition increasingly depends on compressed, mediated, and recursively transformed representations. A summary may replace a source, an interpretation may become the input for another interpretation, and fluent language may become mistaken for independent validation. More information and more coherent explanation do not necessarily produce better contact with the underlying reality.
The recurring question is whether thought remains corrigible by something outside its own representational loop. Can a person reconstruct where an idea came from, distinguish external evidence from generated interpretation, recognize when a framing has shifted, and allow observation or consequence to overturn an attractive explanation?
Cognitive Drift describes human-scale misalignment under continued cognitive function. Co-Cognition describes distributed thought in which external systems help shape the next stage of reasoning. Recursive Compression explains how previously transformed representations become inputs to later cycles. Semantic Fidelity concerns whether intent, context, distinctions, and meaning survive those transformations. Constraint Collapse identifies the loss of corrective authority when feedback remains available but no longer redirects the process.
Related areas of inquiry include Filter Fatigue, cognitive outsourcing, cognitive debt, attention fragmentation, memory and externalization, judgment under mediation, provenance, and the boundary between human intent and system-generated structure. These are not all interchangeable concepts or formally equivalent mechanisms. They describe different pressures, consequences, and questions within the wider study of mediated cognition.
AI-mediated thought illustrates both the capability and the risk. A person can externalize an unfinished idea, receive a structured response, evaluate it, and reuse it in another round of reasoning. This can improve synthesis, memory, creative range, and conceptual clarity. It can also allow the system's framing to become progressively more influential than the purpose or evidence that began the exchange.
A bounded Co-Cognitive loop preserves human intent, source distinctions, independent evaluation, external testing, and a point at which refinement ends. In an unbounded loop, prior outputs may become the main evidence for later outputs, while coherence substitutes for contact. The concern is not AI use itself but whether the exchange remains answerable to conditions beyond the exchange.
Reality Drift describes how systems can remain operational and internally coherent while losing corrective contact with the realities they represent. Cognitive Drift examines the corresponding condition at the level of human attention, models, meaning, and judgment. The same general problem appears when representations become easier to reuse and refine than they are to test against their referents.
The collection is descriptive and analytical rather than clinical. Cognitive Drift is not a diagnosis, a synonym for cognitive bias, or a claim that external tools necessarily degrade thought. Its concern is the relationship between continued cognitive function and the conditions capable of correcting it.
Jacobs, A. (2026). Cognitive Drift Papers. Reality Drift Framework.
This repository is distributed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license. See LICENSE.
Reality Drift Framework, 2023–2026
A. Jacobs
Coherent thinking is not the same as grounded thinking.