README
The doorway. Sets one micro-goal: "understand section 2 of this article" or "learn how this technique works."
Not a long-form summary. A single, clear target. This is what focuses you before diving in.
Turn passive reading into active engagement through dialogue and structured retrieval.
R2R2: Reading Through Dialogue
A protocol where reading is not consumption. It's a loop: README to READMORE to MEMORY, then back again. Built for how brains actually work—especially neuro-divergent brains.
The Core Problem
Most reading is passive. You open a document, scan text, close it, and forget most of what you read. The problem isn't your attention—it's the format.
Reading works when it's active: when you retrieve information from memory, explain ideas in your own words, and apply them. But active reading requires dialogue. It requires someone or something asking you questions and responding to your thinking.
R2R2 is that dialogue made practical.
The Protocol: README → READMORE → MEMORY
R2R2 is built on one immutable loop. It has three parts, and they work only when connected:
READMORE
The active engine. You read a small chunk (3–5 pages, one sitting). Then you answer three sequential questions:
- Recall: "What was the main point?"
- Meaning: "Why does that matter?"
- Transfer: "How could you use this?"
One question at a time. No batch Q&A. Retrieve from memory, don't skim. Explain your thinking.
MEMORY
The continuity layer. After you answer, you document your session:
- What was read
- What was understood
- What was unclear
- Evidence of progress
This becomes visible. You see your own trajectory. That visibility creates return pressure—you want to keep going.
Then the loop returns to READMORE. You read the next chunk. Your MEMORY shapes the next questions. Repeat.
The Agent Layer: Rubber-Ducking That Responds
An agent is not a chatbot dispensing answers. In R2R2, the agent is your reading partner—someone who asks questions and responds to your thinking.
This does four critical things:
Creates Accountability
You read a chunk. The agent asks. You must answer. That act of answering forces retrieval from memory—the actual mechanism of learning.
Prevents Overload
One question at a time. Not a flood of prompts. Sequential dialogue lets your brain process before moving forward.
Generates Feedback
Your answer is shallow? The agent doesn't judge—it scaffolds. It offers support and asks you to re-read with guidance. Learning, not gatekeeping.
Enables Adaptation
The agent uses SOUL modes—different engagement approaches depending on your state. Overwhelmed? Triage mode (read 3 pages, two questions, stop). Want to go deeper? Research mode (map topic interests, recommend next source).
Why This Works for Neuro-Divergent Readers
R2R2 is explicitly designed for how ADHD and other neuro-divergent brains work:
Passive text feels boring
Dialogue + agent questions = novelty + engagement
Long reading blocks overwhelm
Small chunks (3–5 pages) with clear next step
No feedback = no motivation
Immediate agent response + visible progress in MEMORY
Missed days = shame spiral
No punishment. Easy restart with reduced friction
Too many options = paralysis
One question at a time. Micro-goals. Clear path forward
Reward consistency is hard
Streaks (sessions completed) not speed. Celebrate showing up
Backed by Science
R2R2 isn't theory. It's built on proven cognitive mechanisms:
- Retrieval Practice (Karpicke & Roediger, 2008): The mechanism of learning is retrieval effort, not re-reading. R2R2's three-question format forces retrieval—recall, meaning, transfer—at increasing difficulty.
- Spacing and Interleaving (Learning Scientists): Retrieving information repeatedly over time works better than massed practice. MEMORY + loop returns creates spacing naturally.
- ADHD-Informed Design (CDC Classroom Guidance): Structure + immediate feedback + completion over perfection. That's R2R2.
- Gamification Tied to Effort (Park & Kim, 2019): Badges and streaks work when tied to real behavior, not cosmetic trophies. Session completion is real effort.
A Real Example: One 10-Minute Session
You want to understand an article about attention regulation. Here's what one session looks like:
1. README
Your goal: "Understand how working memory limits affect reading comprehension"
2. Read
You read: Pages 2–4 of the article (about 5 pages)
3. Recall
Agent asks: "What was the main point about working memory?"
You answer: "It says people can only hold about 3–4 things at once while reading complex text."
Agent response: "Good. That's accurate. Notice you cited a specific number—where did that come from in the text?"
4. Meaning
Agent asks: "Why does that limit matter for reading?"
You answer: "Because if a sentence has too many ideas, readers can't hold them all at once and lose comprehension."
Agent response: "Exactly. You've connected the mechanism to the outcome. Next: can you explain why short sentences help?"
5. Transfer
Agent asks: "How could you apply this idea—about working memory limits—to your own writing?"
You answer: "I could break long sentences into shorter ones. Give readers fewer ideas per sentence."
Agent response: "That's one way. Stick with that. Next session, we'll explore how to structure paragraphs around this principle."
6. Document
MEMORY entry: "Session: Working memory in text comprehension. Understood: working memory limits + chunking. Unclear: how this applies to long-form vs. short-form text. Next: sentence structure patterns."
You see: Session logged. Progress visible. You've read and understood 5 pages in 10 minutes through dialogue, not skimming.
A Multi-Domain AHA Scenario (GAMMA -> BETA -> THETA)
This is the higher-order mechanism many users report after repeated R2R2 loops: an exponential jump in understanding, followed by critical re-checking, followed by durable memory encoding.
1. User Input Across Domains
User: "I read a neuroscience book chapter on working memory, a research article on retrieval practice, and a blog post on AI reading companions."
2. Agent Summarizes Achievement
Agent: "You completed a cross-domain synthesis. You connected brain limits (neuroscience), learning mechanism (retrieval), and implementation pattern (agent dialogue). That is not surface reading. That is systems-level comprehension."
3. GAMMA AHA Expansion
User: "Wait. Does that mean working-memory limits + retrieval loops + dialogue scaffolding = all of that plus a compounding learning effect?"
Agent: "Yes, exactly. You are seeing the second-order effect: each loop improves both content understanding and your future ability to learn new content. That extra term is the compounding factor."
4. BETA Critical Questioning
Agent: "Important check: this is not the final answer yet. Are we sure the compounding effect comes from understanding quality, not only from motivation or novelty? What would count as evidence either way?"
User: "We would need repeated session data: recall depth, transfer quality, and consistency over time, not just excitement from one good session."
Agent: "Correct. Good science move. You turned insight into a testable claim."
5. Grounding to a Cognitively Close Analogy
Agent: "Think of this like strength training with a coach. One workout builds some strength (content learned). But the bigger gain is that your coordination improves, so future workouts become more effective. R2R2 does the same for reading: each cycle trains both knowledge and learning capacity."
6. THETA Memory Solidification
Agent: "What happened here: you moved from synthesis to hypothesis, from hypothesis to critique, and from critique to grounded model. This should now enter MEMORY as a stable pattern."
MEMORY entry: "Pattern observed: multi-domain reading + retrieval dialogue can produce compounding understanding. Status: plausible mechanism. Next step: validate with longitudinal evidence."
In R2R2 terms: GAMMA gives the AHA leap, BETA pressure prevents over-claiming, and THETA turns the validated insight into reusable memory.
Why This Matters for B1C3
R2R2 is cognitive offloading in practice. It shows what happens when you move information retrieval and dialogue off-device (into an agent, into MEMORY, into the protocol structure) and onto a partner—one that's patient, accessible, and designed for how brains actually work.
For neuro-divergent people especially, this changes reading from "I should be able to focus" into "here's a structure that works."
It's not a tool. It's a pattern. Institutions (libraries, schools, universities) can deploy this as infrastructure.
Get Involved
R2R2 is actively practiced. The protocol is open. The code is in development.
Interested in deploying R2R2 as a reading program? Running a pilot at an institution? Contributing to the protocol?