Solve Any Problem with Iterative Learning

“Ibn Alhazen was the first person ever to set down the rules of science. He created an error-correcting mechanism, a systematic and relentless way to sift out misconception in our thinking. ‘Finding truth,’ he said, ‘is difficult and the road to it is rough. As seekers after truth, you’ll be wise to withhold judgment and not simply put your trust in the writings of the ancients. You must question and critically examine those writings from every side. You must submit only to argument and experiment and not to the sayings of any person. For every human being is vulnerable to all kinds of imperfection. As seekers after truth, we must also suspect and question our own ideas as we perform our own investigations, to avoid falling into prejudice or careless thinking. Take this course, and truth will be revealed to you.” Excerpt from Cosmos: A Spacetime Odyssey, Episode 5: Hiding in the Light 🎞️

The ever-surprising 21st century teaches us to learn fast, all the time. We’ve covered the benefits of the growth mindset in Self > Principles > Growth Mindset. But how specifically do we grow and learn? Here we introduce a method, which has been called the scientific method, build-measure-learn, critical thinking ▶️, problem-solving, even the baloney detection kit… In relations to our personal growth, let’s call it the iterative learning method. Iterative learning helps us explore the world, improves our decision-making over time, and prevents us from falling victim of misjudgments or changing contexts (see Self > Mind Hacking > Cognitive Biases and Self > Principles > Context Matters and Things Change).

Iterative learning calls to mind a knowledge-acquiring method that was taught by one of my history teachers: learn → challenge → befriend. When you know nothing about a topic, you need to learn as much about it as possible to grasp its foundations and details. When you know enough, you can start poking holes at the topic, challenging it to see if it’s logical, useful, consistent, ethical, etc… Finally, when you’ve fully comprehended a topic’s principles, merits and pitfalls, you can befriend it, trusting it as a part of your knowledge base.

Iterative learning, however, goes beyond learn → challenge → befriend. It also has an error-correcting mechanism, as Ibn Alhazen had proposed, built into it. When examining and challenging a topic, iterative learning encourages you to constantly test and validate your knowledge. Children actually do this intuitively — when playing with sand, for example, they would touch it, move it, throw it, taste it… do any number of things to learn and challenge how sand operates. Sometimes their experiments may fail (e.g., ruining a sand castle after pouring over too much water), but a child would just continue on with the next experiment. A lot of adults, unfortunately, seem to have lost this ability to experiment, fail and do over.

To regain our childhood ability of play and experiment, let’s use a common decision, what to eat for dinner, to explain the iterative learning method — which can be summarized in seven steps: be aware > understand > question > hypothesize > act > validate > iterate and persist.

Step 1. Be aware

“There is no shame in not knowing; the shame lies in not finding out.” Russian Proverb

You cannot learn something if you are not even aware of it. One of the most crucial life skills we can develop, therefore, is to be aware of the forces affecting our lives.

How do you become aware of something? Take where to go for dinner for example, how are your thoughts about dinner triggered? Maybe it’s customary for you to have dinner at say 7pm. Or maybe you are meeting with friends for dinner. Maybe you are reminded of dinner looking at suggestive ads on your way back from work. Or from experience and logistics you know if you don’t get the ingredients right now, there will be no dinner. Or maybe you are simply hungry — your internal organs alert you to get food… Triggers of awareness can be customary, social, economic, experiential, physiological, and many more.

Awareness itself, though, can only fall into one of four categories:

1. Known known (I know I want dinner): this is something that you are pretty certain to be true. It is the most desirable kind of awareness since you know what outcome to expect. However, known knowns, as in any knowledge, can change in context or time (e.g., I’ve just eaten) — this is what trips us up sometimes — what we’ve known to be true has changed to be unknown or untrue. In this case, we’d need to be aware that things have changed. Iterative learning provides you a way to verify known knowns and detect change.
2. Known unknown (I know I don’t know what to eat for dinner): this is something we know that we don’t know right now, but thanks to being aware of it, we can find out what it is. Known unknowns can be knowable (let me go online to look for a restaurant for dinner), or they can be unknowable or risky (there is a chance for rain tonight, will my outdoor dinner be ruined?) Iterative learning helps you turn known unknowns to known knowns and manage their risks.
3. Unknown known (my friend has a good idea for dinner but won’t tell me due to politeness): this is knowledge that you don’t know exist. Unknown knowns can lead to anything from a missed opportunity (my friend and I would have both enjoyed that restaurant they had in mind) to an exploitation (you are having fast food for dinner because the advertisers know if they expose you to their ad on your way home, you would subconsciously want to have their food). Iterative learning turns the subconscious into conscious, and unknown knowns into known knowns through questioning and testing (ask your friend where they wish to go or become aware the subliminal effect ads have on you).
4. Unknown unknown (the restaurant you’ve decided to go to gets caught on fire): these are things you couldn’t have known until it happens. By understanding, questioning and hypothesizing about a problem, sometimes unknown unknowns can become known unknowns (e.g., the restaurant could have noticed signs of fire risks). While iterative learning cannot detect unknown unknowns, it can help you navigate surprises or accidents when unknown unknowns become known unknowns or unknown knowns.

Source

Step 2. Understand

“Insufficient facts always invite danger.” Mr. Spock, “Space Seed” 🎞️

Now that you are aware of a problem or topic, and you know how much you do or don’t know about it, you’d need to learn more to make good judgements. Let’s say you want to go to the best Italian restaurant in town for dinner, you can ask:

1. What are the fundamental principles or key themes? What makes a good Italian restaurant? Is it taste, variety in menu, originality, location, ambience, clientele, pricing, or all of the above?
2. What are the relevant information pertaining to this topic? Should you consider the reputation of the restaurant, the awards the chef has received, or opinions of established food critics? When the information become too much to handle, try breaking them down and grouping them into chunks.
3. How does the information relate to one another (what is the taxonomy)? For example, if all of the above factors are important, how would you rate their relative importance? And would, say, ambience affect your evaluation of a restaurant’s taste and vice versa?
4. How have the historic opinions about the topic changed? Maybe a few years ago high-end Italian restaurants with Michelin-starred chefs were really popular, but in recent years people had become more interested in homemade-style Italian food. This change might affect the recommendations you come across.
5. What are the criticisms? Even the best restaurants have their critics. Could the critics’ considerations, for example, disagreeing to how some restaurants trade traditional recipes for new inventions, inform your decision?

FLICC Taxonomy of Logical Fallacies

To research most topics, Wikipedia is usually a good place to start. However, if we really want to know how something is affecting us, we should keep our eyes and ears open, try to learn as much about it from trusted sources and from all angles, and not shying away from controversial ideas 🎧. That’s how we become less susceptible to manipulations (see related discussion in World > 21st Century Challenges > Cognitive Warfare).

To test your knowledge about a topic, try the Feynman Technique: explain the topic in simple language, observe the gaps in your knowledge, learn more, then repeat.

For more learning techniques, check out 14 Techniques to Accelerate Your Learning.

Step 3. Question

“Deep doubts, deep wisdom; small doubts, small wisdom.” Chinese Proverb

Philosopher Bertrand Russell once said ▶️: “I think one of the troubles of the world has been the habit of dogmatically believing something or another. And I think all of these matters are full of doubt and the rational man will not be too sure that he’s right…” [26:48].

Journalist Warren Berger put it more practically ▶️: “Questions are… survival skills for all of us. And that becomes even more true in a time of dynamic change. We’ve got so much that we have to adopt to. We have to solve problems. We have to deal with change and uncertainty. And question… is one of the primary tools that lets you do that” [4:54].

The more important a topic or problem is, the better it is for us to ask the questions early, so we can manage its effects on us sooner. Going back to the Italian restaurant example: your chances of finding the best Italian restaurant would be higher if you had 1 week instead of 1 hour to look for it.

Here are a few tips on asking a good question ▶️:

• Ask 5Ws and 1H: Who? What? When? Where? Why? And how? Figuring out the “why”, in particular, will help us find out “how” to solve a problem.
• Ask the naïve/dumb questions: these are questions that you may think the answers are obvious or straight-forward, but when you dig into them, you may be surprised how much you don’t know. Again, use the Feynman Technique to identify gaps in your knowledge. For example, what qualifies as an Italian restaurant? Does a New York Pizza place or a gelato shop count as an Italian restaurant?
• Ask 5 times: The five-whys technique helps us get to the fundamentals of a topic, uncovering root causes and issues, simply by asking why to your answers 5 times — Why do you want to find the best Italian restaurant? Because I like Italian food. Why best but not the most tasty? Because I care about the ambience. Why do you care about ambience? Because I am bringing my long-time crush to the restaurant for our first date. Why do you want to bring your date to the best Italian restaurant? Because I want the date to succeed. Why look for the best Italian restaurant; why not look for one that you feel most confident and comfortable in? … There you go.
• Ask deep questions: While the five-whys technique prompts us to get to the root cause of something, Systems thinking used by journalists goes even further in reflecting not only What’s happening? (Events), but What’s happening over Time? (Trends & Patterns), How are the interconnected policies, structures and power dynamics fueling the patterns? (System’s Structure), and Why is the system structured this way? What assumptions, beliefs, experiences, and worldviews are driving the system? (Mental Models). See more discussions in World > 21st Century Challenges > Method: Long-term Systems Thinking.
• Question your assumptions: If one of your assumptions is that freshly made pasta is better than industrially manufactured ones — ask why? Is it because fresh ingredients taste better, or because additives make industrial pastas taste worse? Are there other factors? Do the factors always hold true?
• Tackle counter-arguments by using Jonathan Baron’s Active Open Mindedness:

1. Explain your question and why it is important.
2. Present the most obvious answer or answers.
3. Consider less obvious alternatives, or objections to the obvious answers.
4. Rebut the criticisms, or explain how the original answers can be modified to deal with them.

• A similar method to Active Open Mindedness is the older Socratic method, which in its simplest form, involves peers questioning each other’s assumptions.
• Explore something new/absurd: Are there relevant questions that no one has asked before? For example, can a Mediterranean restaurant that serves Italian food possibly be the best Italian restaurant in town? By being curious and asking new, even absurd, questions, you’d see your problem space in a new light and stretch your imagination in terms of the answers.
• Ponder “what if”: What if I don’t look for the best Italian restaurant but one that I enjoy the most? What if I travel outside of town; would I get better results? Hypothetical question is a good tool to reframe our thoughts and to explore new possible answers or futures.
• Ask questions, then follow up. Some people don’t ask enough questions and just accept their circumstances as-is. Others ask questions to no end without finding answers or taking meaningful actions. These two extremes are what philosopher Andy Norman calls mentally “under-immune” and mentally “over-immune”. When one is mentally under-immune, they can be easily misled or manipulated (e.g., tricked by fake news); when one is mentally over-immune, they enter an analysis paralysis (e.g., become a political pessimist). Neither is desirable. Instead, hypothesize, act and validate your questions.

System thinking. See more discussions in World > Systemic Bullying

Step 4. Hypothesize

After asking good questions about your topic or problem, it’s time to find some answers. When a topic or problem is big, it helps to break it down so we can find specific answers to specific questions.

Let’s say you’ve become aware that your energy level goes significantly down after each meal. After some research, you’ve come to understand that having too much carbohydrate in one’s diet may cause sluggishness and low energy level. As a result, you now question whether your meals are too carb-heavy.

To answer this question, you need to formulate a hypothesis, or a theory that is testable, quantifiable, refutable and falsifiable ▶️. Your hypothesis is: “assuming a carb-heavy diet is making me tired all day, if I eat low-carb foods instead of high-carb foods, I should have more energy throughout the day”. You know this hypothesis works because it is:

• Testable: you can conduct tests, replacing high-carb foods in your diet with low-carb ones, to verify whether your hypothesis is true. By contrast, a non-testable theory is, for example, you think a ghost is haunting you and zapping your energy — you can’t ask the ghost to stop to see if that makes a difference.
• Quantifiable: you can measure your calorie intake with the assumption that given the calories you consume remain the same, low-carb foods give you more energy. A non-quantifiable theory is, again, you cannot ask a ghost to turn up or down its haunting level (an 8 instead of a 5 please), and see if you have subsequently more or less energy.
• Refutable: you can refute or prove your hypothesis wrong if you notice that your change of diet makes no difference on your tiredness throughout the day. A theory is non-refutable if it is not testable.
• Falsifiable: you can find a counter-example to disapprove the hypothesis, such as feeling groggy after having just a salad and no carb. A non-falsifiable theory is always true, e.g., I feel like a haunting ghost is making me tired all day. Since your feeling is always true to you, the theory is impossible to falsify.

As you get better at iterative learning, you may be able to do something more advanced when hypothesizing: predicting the outcome. Predictions help us better plan and complete our everyday tasks by estimating the chances of a desirable outcome as well as minimizing mistakes or surprises. For instance, after a series of experiments, you may be able to predict that eating cake will put you right to sleep, and decide to skip it to stay awake. Study done by U.C.L.A. shows that even if you get the answers wrong, trying to guess the answer to a question or predict the outcome of a hypothesis aids learning.

Step 5. Act

This is where all your thinking and learning meet reality. Eat, drink, love, talk, build, try, create, dance, sing, play, share… do whatever you need to do (within moral and legal bounds) to carry out your hypothesis and find results.

Step 6. Validate

Believing or questioning anything without validating your ideas can lead us to the wrong conclusions, yet validation is probably the least practiced step in the entire iterative learning process.

No one likes to be wrong (see ample examples in Self > Mind Hacking > Cognitive Biases). Validating our hypothesis exposes us to the possibility of being wrong (i.e., refutable and falsifiable). And when proven wrong, we have to be willing to accept the evidence, let go of our beliefs, and update your understanding of the problem. This is not easy to do, but it’s crucial, since validation gives us valuable information on how to improve our assumptions and decisions.

Back to our sleepy problem: it would be a happy outcome if you find out a low-carb diet does raise your energy level throughout the day. With this information, you can then go on to conduct subsequent experiments regarding what types of low-carb foods will give you the most energy boost. However, let’s say after a period of testing, the amount of carbohydrate in your food has no effect on your tiredness. In this case, you must accept that your hypothesis is wrong and go back to the drawing board — could your energy level be affected by sleep, or even an illness? Being proven wrong might be frustrating, but it informs us about the topic in hand, including what is not a cause, helping us plan the next experiment.

ClearerThinking proposes Three Tenets, Six Questions to help you validate your hypothesis:

Three Tenets

1. Remember that “Most problems are potential improvements screaming at you.”
2. Be honest with yourself about your flaws.
3. Identify both the immediate cause and the root cause of your problem.

Six Questions

1. What went wrong?
2. Have you made a mistake like this before?
3. What was the immediate cause of the problem?
4. What was the root cause of the problem?
5. What can you do to correct the problem in the short term?
6. What can you do to prevent problems like these in the long term?

Since admitting one could be wrong is difficult, it might be a good idea to do a “peer review” of important hypothesis or decisions. Tell your friends or trusted acquaintances about your experiments; ask if they agree with your conclusions.

In validating our assumptions, we will also improve faster if we embrace uncertainty, ambiguity, complexity, and things that are outside our control. See more in Groups > Method: Collaboration > Agile Methodology.

To quote author Maria Konnikova in The Biggest Bluff 📖:”… can you control yourself and play well when you are losing… trying to still be objective as to what your chances are… then you’ve conquered the game” [p.60]. See more Life > Principles > Life Actualization through Moral Struggles.

Step 7. Iterate & Persist

“If you can’t fly then run, if you can’t run then walk, if you can’t walk then crawl, but whatever you do you have to keep moving forward.” Activist Martin Luther King Jr.

One of the most quoted motto of the tech startup industry is: “fail fast, fail early” — not only it is essential to fail and try again, to accelerate learning it’s important to fail quickly and soon — so you can get to success as fast as possible. How do you speed up failure? Break down your problem into the smallest testable units (in tech this is called MVP, minimum viable product).

To effectively “fail fast, fail early”, we need to have a growth mindset and build resilience. To quote The Biggest Bluff again: “Learn to lose constructively and productively, lose and come back, lose and not see it as a personal failure” [p.60].

Q&A

Q: Are there any drawbacks to iterative learning?

• A: Iterative learning takes effort and time. Take the most powerful example of iterative learning: Evolution; it takes literally billions of years (see Self > Human Basics > Evolution). To afford enough energy and time to learn, focus on the most important things that are closest to your goals and values (see Life > Methods > Achieving Goals, and Life > Life Perspectives > Motivations, Values & Purpose).

Iterative learning can be universally applied to any discipline or situation for our selves, lives, groups, or the world. For this reason, we will use the iterative learning method to investigate concepts throughout 21st Century Personhood.

Do you have any suggestions, doubts, hypothesis or experience for this topic? Please comment below 👇!