AI-Mediated Cognitive Extension
Engineering Solutions to Substrate Constraints
Engineering Solutions to Substrate Constraints
Human cognitive capacity is fundamentally constrained by working memory: about four chunks of information at a time. This program develops AI-human hybrid systems that extend cognitive capacity by optimizing information compression rather than attempting to expand biological processing power.
Human working memory holds roughly four chunks (Cowan, 2001; the older estimate was seven, plus or minus two). Capacity varies a little between people and tracks fluid intelligence, but no one holds forty. Expertise does not add slots; it packs more into each chunk. The limit holds across cognitive activities and is an architectural feature of biological neural substrates.
The human retina samples approximately 1 billion bits per second, yet only about 10 bits per second reach conscious awareness, a compression ratio of about a hundred million to one. This massive information loss reflects evolutionary optimization for survival rather than comprehensive environmental modeling.
As professionals accumulate expertise, they transition from fluid intelligence (adaptive reasoning) to crystallized intelligence (pattern matching). While initially enhancing performance, this transition progressively fills working memory with stored patterns, leaving insufficient capacity for adaptive thinking.
The program's hypothesis is that most professionals peak in their 20s and 30s as accumulated expertise crowds working memory, and that the few who sustain performance use strategies that keep it free.
The program's working proposal is that consciousness is information processing through a biological neural substrate with severe bandwidth limits. The compression ratio of about a hundred million to one reflects optimization for rapid decision-making rather than faithful environmental rendering. Modern information-dense environments create cognitive demands orders of magnitude beyond evolutionary contexts, making augmentation not just beneficial but necessary for optimal human performance.
Creating composite cognitive substrates where AI handles bandwidth-intensive operations while biological cognition maintains adaptive reasoning and consciousness.
The augmentation does not bypass working memory constraints; it optimizes how that memory is used. Rather than filling working memory with pattern recognition tasks, AI performs compression externally, presenting optimized information that preserves working memory for adaptive reasoning. Working memory capacity stays the same; performance is predicted to improve substantially.
Reducing risk by testing each step before major hardware investment
Years 1 to 2 from funding | $2-5M
Objective: Test the program's principles using existing display technology
Key Components:
Success Criteria:
Years 3 to 4, if Phase 1 meets its criteria | $15-25M
Objective: Deploy AR-based sensory augmentation if Phase 1 meets its criteria
Key Components:
Expected Outcomes:
Each can come out for or against the proposal on its own
The first large studies of everyday AI use have arrived since this program was written. They bear directly on its predictions.
Together these sharpen the program's central distinction: extension that frees working memory for adaptive reasoning, versus substitution that lets the reasoning lapse. The protocols measure both.
Phase 1 (years 1 to 2 from funding): $2-5M
Phase 2 (years 3 to 4): $15-25M
Ongoing (year 5 on): $15-30M annually
Core Research Team:
Extended Collaborators:
Baines, M. K. (2026). AI-Mediated Cognitive Extension: Engineering Solutions to Substrate Constraints - The Physics of Human Sensory Augmentation. Ic² Research Institute. Zenodo.
This comprehensive 70+ page research paper provides the complete theoretical framework, detailed experimental protocols, test methods, and implementation specifications for the AI-Mediated Cognitive Extension program. The paper is deposited on Zenodo with a DOI, which makes it permanently archived and citable.