This project is a single-person research venture led by me, Ryan Koch. I say led because it is very much a collaboration with AI.

My work investigates how deliberately constructed signs can organize memory and develop new capacities for differentiation, association, and communicating acquired knowledge. I design learning materials, use them to encode knowledge in my own memory, and revise them through that experience. I am testing whether the competent possession of 26 concepts about a subject domain enables sufficient understanding to critically evaluate unfamiliar materials on the subject.

At another level, the project is an experimental epistemology; it attempts to examine assumptions about knowledge, understanding and thinking by drawing from ideas in psychology, neuroscience and information theory. The knowledge structures I have designed help to distinguish among perception, memory, storage, and access, as well as between internal and external evidence of knowledge. To the extent that these components can be distinguished and measured, there is the potential to modulate their relative contributions in the learning process, and perhaps improve it.

Rather than seeking faithful visual representations of factual knowledge, I use AI to produce non-representational mnemonic imagery whose value lies in its effects on learning. AI enables me to design structures based on established findings in learning research, and iterate improved materials based on the performance of those structures. I also use AI to compose glosses within tightly defined constraints. These glosses must serve the knowledge structure while remaining accurate, attentive to context, and open to revision.

This process places me inside a feedback loop: my experience of learning informs the production and revision of the materials I learn from. Because part of that process involves encoding knowledge in my own memory, I can assess some aspects of the materials’ usefulness almost as they are produced. Ultimately, this is a loop that I would like to refine and systematize to enable others to insert themselves into loops of their own design, making their own materials.

The loop is modeled after history as a recursively-revised picture of the world, and from human memory as a worldview continually reshaped by experience. It suggests an approach to education in which learning materials function as provisional memories: structures that help us interpret new information while remaining subject to revision themselves.

Writing and subsequent technologies have greatly expanded access to recorded knowledge. Yet material availability does not guarantee cognitive access. However much information exists, we still face the task of making it available to perception, memory, judgment, and reasoning. I want to intervene to help the process of cognitive access keep pace with the extraordinary pace of material access and production.

This project is an invitation to approach learning as an ongoing process of construction and revision. The materials cost $26, and I want to be clear about where the value lies. Supporting efficient, durable memory is the most empirically-grounded aspect of the project, and the one I have evaluated most directly through my own learning. The more open question is whether learning 26 concepts within this structure provides a foundation for broader understanding and critical judgment. Your involvement can help investigate that question and shape the materials as they develop. The aim is to create stable artifacts we can use to communicate what we know and how we know it, then revise as our understanding develops.

As of fall 2026, this is the project’s provisional account of itself. I will document discoveries and revisions on the research and design page as the work develops.