Far Transfer: How Adults Actually Learn

Most adult learning fails. Not because people are lazy or incapable, but because the learning is designed wrong. It optimizes for information delivery when it should optimize for transfer.

Transfer, specifically far transfer, is the ability to take knowledge from one domain and apply it in a completely different context. It is the difference between memorizing a formula and knowing when to use it in a situation you have never seen before.

I wrote an entire book on this subject. Far Transfer went through five editing passes and scored 9.9 out of 10 on the Forge evaluation rubric. It draws on 475 real coaching transcripts as source material. This post covers the core ideas.

Why Most Training Does Not Work

Corporate training, online courses, even university programs. They overwhelmingly focus on near transfer. Learn this specific thing, apply it in this specific way. That works for procedural tasks. It fails for everything else.

The real world does not present problems in the same format as the training. A business owner who takes an AI course learns how to write prompts in the course environment. Then they sit down at their desk on Monday and cannot figure out how to apply any of it to their actual workflow. The knowledge was acquired. The transfer did not happen.

This is not a motivation problem. It is an architecture problem.

I see this in coaching sessions constantly. As I put it while developing the book's voice guide: "What is the reader doing right now that isn't working? They're probably rereading stuff, highlighting. Throw those away. They don't work at all." The research is clear on this. Passive review strategies like rereading and highlighting create what learning scientists call the Fluency Illusion. Smooth processing feels like mastery, but it is just recognition. If it feels easy, you are not learning.

The 6-Domain Architecture

Far transfer is not a single skill. It is the output of six cognitive domains working as an integrated system. I call this the 6-Domain Architecture:

1. Attention (The Gatekeeper). Control what gets in. Without attention, nothing downstream works. The protocol: 15-minute focus sprints with a single target and no switching. Before consuming any information, ask one question. Does this serve my current objective? If the answer is no, discard it. In the AI era, this matters more than ever because the volume of information available is effectively infinite. The gatekeeper has to be strong.

2. Knowledge (The Library). Store what matters in retrievable form. Recognition is not recall. The practical tool is the Compression Card. One sentence plus one diagram on a 3x5 card. If it does not fit, you have not understood it yet. Pair this with a spaced-repetition system where the system decides when you review, not you. This is counterintuitive for most adults, who are used to deciding their own study schedule. But the research shows that self-directed scheduling leads to too much review of easy material and not enough review of hard material.

3. Skill (The Builder). Deliberate practice with feedback. Repetitions without correction reinforce errors. The key rule: three consecutive perfect reps, then stop. More reps after mastery equals wasted time. And interleave three domains per practice session. Blocked practice, where you drill one thing at a time, feels productive but transfers poorly. Interleaved practice feels harder. That is the point.

4. Reasoning (The Strategist). See through surface features to the skeleton underneath. This is where far transfer actually lives. The core exercise is what I call the Remix Drill: strip the surface features off a concept, extract the structural skeleton, then re-apply that skeleton in a completely new domain. For example, the concept of feedback loops appears in thermostats, coaching check-ins, market corrections, and AI training. The surface looks different every time. The skeleton is identical. Practicing this extraction explicitly is what builds transfer capacity.

5. Metacognition (The Control Tower). Thinking about your thinking. The only domain that monitors all others. The protocol is Confidence Calibration: rate your confidence before checking the answer. Track calibration over time. High-confidence wrong answers are the biggest learning opportunity, because they reveal blind spots you did not know you had. The After-Action Review structure supports this: What did I expect? What actually happened? What is the gap, and what is the one fix?

6. Execution (The Engine). Knowing what to do is worthless without doing it. Reduce friction to zero. The tool is the If-Then Algorithm: implementation intentions. When X happens, I will do Y. Pre-decide, do not deliberate in the moment. Stage the next action so activation energy equals zero.

The 12 Skeleton Models

Across every domain I have studied, the same structural patterns keep appearing. I documented 12 essential skeleton models that recur everywhere:

Feedback Loops. Network Effects. Exponential Growth and Decay. The Prisoner's Dilemma. Power Laws (the 80/20 pattern). Supply and Demand. Selection Filters. Diminishing Returns. Activation Energy. Signal-to-Noise Ratio. Phase Transitions. Interference Patterns.

Each of these appears in business, biology, physics, psychology, and technology. The person who can recognize a feedback loop in a thermostat, a coaching relationship, and a market correction has far transfer capacity. The person who only sees it in the context where they first learned it does not.

The AI Paradox

This is critical for anyone using AI tools. There are two paths.

Path A: Cognitive Offloading. AI replaces thinking. The user becomes dependent. Skills atrophy over time. This is the default path for most people who adopt AI without a learning framework. They get faster but they do not get smarter.

Path B: Hyper-Learning. AI amplifies thinking. The user forms a hypothesis first, then uses AI to test and refine it. Skills compound over time. The AI becomes a practice partner, not a replacement.

The rule I teach in every coaching session: never ask AI for the answer until you have formulated a hypothesis. This single discipline is the difference between AI making you smarter and AI making you dependent.

One client described the shift after applying this approach: "So one of the things that it did that I was really pleased with. I was headed down this path where ChatGPT was telling me, yeah, that's great, do it. And Claude said, no, you're not even close." That is Path B in action. He brought a hypothesis. The AI pressure-tested it. He learned from the correction. His next hypothesis was better.

Another client, after just one session applying these principles, said: "Claude actually started using Claude over the weekend, and I used your example for getting critical feedback. And it was amazing." The tool did not change between Friday and Monday. His approach to using it changed. That is far transfer.

What Far Transfer Looks Like in Practice

In coaching sessions, far transfer shows up as the moment when a client takes a skill learned in one context and spontaneously applies it in another.

One client learned AI-powered document analysis for insurance reviews. Within weeks, she was using the same pattern to create custom GPTs for photo lineups in her business, and teaching colleagues how to do it. She described the reaction: "Dr. Jean was amazed, asking how I did it. I told her it was Dr. Jeff who taught me." The specific skill was not "how to build a custom GPT." The transferred skill was "how to identify a repetitive task and design an AI workflow to handle it."

A financial services client learned to use Claude for customer report analysis. He took that same structural approach and applied it to financial statement comparison, competitor analysis, and client prospecting. He described the progression: "It's getting really good. Both of these save us not only a lot of time, but the thing I find even more interesting is it just makes us better." That word, "better," is the marker of far transfer. He was not just faster. His analytical output improved because the AI was catching patterns he missed.

The Honed Companion App

The Far Transfer book is designed as an analog framework. Cards, timers, physical practice. But the digital companion, Honed (at gethoned.app), translates those protocols into an interactive experience. It includes 21 features covering spaced repetition, confidence calibration, compression practice, and the After-Action Review structure. The app is currently in beta.

The Intelligence Loop Connection

Far Transfer connects directly to the Intelligence Loop methodology, which I use to measure how effectively any system converts input into useful output. The formula is I = M x C x U x T. Model Quality times Courageous Action times Update Efficiency times Transfer Power. The multiplication structure means a zero anywhere in the chain produces zero intelligence. A person with excellent attention (M) and deep knowledge (M) but no execution (C = 0) produces nothing. Far Transfer exists to eliminate the hidden zeros.

Practical Implications

If you are designing training for your team, your clients, or yourself, stop optimizing for content delivery. Start optimizing for transfer conditions. Teach principles, not procedures. Vary the practice contexts. Build in reflection checkpoints where people examine their own thinking.

The businesses that learn fastest in the AI era will not be the ones that consume the most courses. They will be the ones that build the deepest transfer capacity. That is a fundamentally different design problem, and most people are not solving it yet.

The book Far Transfer lays out the complete system, including all six domains, the 12 skeleton models, the protocols for each domain, and real examples drawn from hundreds of coaching sessions. It is comparable in scope to Atomic Habits, Make It Stick, and Deep Work, but focused specifically on how adults can build transfer capacity in the age of AI.