Mary Rose Cook's notebook

The public parts of my notebook.

My homepage.

Peak, Anders Ericsson and Robert Pool

Review

Some excellent techniques for deliberate practice. Includes plenty of supporting evidence for skill being born of practice, not talent. A good accompaniment to Mindset.

The brain is plastic (including mindset stuff)

• E.g. Scientific evidence that cabbies' brains changed after acquiring The Knowledge.
• When you exercise, your body will adjust by increasing the supply of oxygen and blood by raising your heart and breathing rate. This allows you to sustain the exercise. If you stay within your limits, you body won't change. Your muscles won't grow. The power of your heart beat won't change. But if you go outside your limits, your physical body will change, allowing you to become more capable. The same is true of the brain.
  • So: get outside your comfort zone.
• The changes in the brain brought about by training:
  • More brain.
  • Redirected areas of the brain.
  • Strengthened connections.
  • Faster signal transmission.
• The brain can change back.
  • Once taxi drivers stopped using their knowledge, it started to disappear.
• Making yourself better at one thing makes you worse at others.
  • E.g. Taxi drivers with the knowledge got worse that lay people at recalling details of a complex 3D figure. The theory is that more of the part of their brains that is devoted to spatial memory was dedicated to The Knowledge.
• The body is extremely adaptable. The brain is the same.
  • E.g. A man trained himself to do 10,000 push-ups non-stop.
• If you're not improving, it's not because you lack talent, it's because you're not practicing the right way.
• People who show an initial aptitude for something don't necessarily continue that way.
• A higher IQ does correlate with skill in intelligent pursuits like chess and science. But not once you only focus on the elite performers. For them, high IQ actually correlates negatively.

Mental representations

• The central idea in the book.
• Experts monitor their performance. When they perform less well, they try to improve their mental representations.
  • A good musician can read a piece of music and know a) what it should sound like and b) how their playing differs from what the piece should sound like. Their mental representation:
    • Encodes the goal.
    • Lets them compare their attempt to the goal to find their mistakes.
    • Lets them devise more effective practice to target their mistakes.
  • A good musician can read a piece of music and see at different levels
    • The overall effect they want to achieve.
    • Specific moments where they need to do things to contribute to that effect.
• A musician's mental representation might include: pitch and length of note, volume, rise and fall, intonation, vibrato, tremolo, harmonic relationship with other notes and with notes played by other musicians.
• An expert might have mental representations in multiple modes: e.g. a spatial one of where their fingers should be and an aural one of what a note should sound like.
• You build a mental representation incrementally. You don't have to see far ahead and find the perfect model in advance. You can build it piece by piece.
• Mental representations need to be built by doing, not just thinking.

Finding good mental representations

• It's hard to know which mental representations lead to the best performance. Often, the best you can do is talk to experts and try to understand how they think about doing their work.
  • Try having an expert engage in their expertise, then stop them and have them describe:
    • The current situation.
    • What has happened.
    • What is about to happen.
• If you can't find the mental reps of experts, try adopting their practice techniques (see below).

Chunks

• Experts have better mental representations: chunks.
  • E.g. Chunks explain why a chess master can recall a board after looking at it for a few seconds. This goes against the understood limits of short term memory: They don't remember the position of every piece. They remember chunks. Which means that if the pieces are randomly placed, their performance degrades to what would be expected if they were remembering each piece individually.
• Chess masters spend the majority of their time reading and analyzing past games. They study the board, predict the next move, then if they get it wrong they think about what they missed. They are building chunks.
  • Time spent on this activity is the best predictor of chess master success - better than time spent playing against others.
• Crucially, chunks come from long-term memory.
• Chunks are arranged into hierarchies.
• Chess players describe positions not by reference to individual pieces but by much more abstract concepts like "lines of force" or "power".
  • (This makes me think of something like cohesion in programming.)
  • An equivalent for doctors is seeing patterns of symptoms. For example, a patient complaining of an earache, but the doctor also noting that one eye is dilated. A less advanced doctor would overlook the dilation.
    • (I've done this plenty of times when debugging. The problem is not overlooking it. The problem is not having the experience to have it ping off something in your head.)
• Mental representations are broad strokes, but masters still zoom in to the lowest level (in chess, the pieces), to analyze the effect of small changes at this low level.
• The main goal of deliberate practice is to develop ever more efficient mental representations.
  • These mental representations play a core role in deliberate practice. (More to come on this.)
• Mental representations don't transfer between domains.
  • E.g. a chess player won't have an advantage over others in general visuospatial abilities.
  • E.g. Experienced surgeons who learnt laparoscopic (keyhole) surgery, which is very different from traditional surgery, picked it up no faster than surgery trainees.
  • (Surely some general stuff transfers. E.g. the idea of navigating graphs.)
• Things encoded by mental representations: facts, images, rules, relationships, and more.
• The goal of mental representations: processing large amounts of information quickly.
• Mental representations let you priority rank information.
• Mental representations let you sift through many possibilities quickly and choose the best.

Planning

• Rock climbers plan their route. Surgeons plan out an entire surgery before making the first incision.
• The better your mental representations, the better the plan.
  • (This could be great for my programming. I can mostly program by the seat of my pants. But I wonder if I went hard on being able to build a clear mental plan that ended up having few gaps when enacted, if this would force me to build much better mental representations.)
  • (Another example: If I did an outline of an essay, it would help me produce the finished article, but the ideas would change a lot. A writer with better mental representations (a writer who was better) would be able to write an outline (plan) that came closer to their final output.
• Experienced surgeons report that the main way they detect problems is to notice ways in which the performance doesn't match their plan.
  • They then list options and choose the best.

Misc

• The dominant factor in how much a person understands of a set of information is how much they already know about the topic.
• Experts generate multiple possibilities and choose the best.
  • This distinguishes them from less skilled people.
  • (The way the authors write this implies it's a conscious process that generates the possibilities and chooses the best, not purely a mental shotgun.)
  • (I don't like generating multiple approaches. a) I don't like planning and b) once I have an approach I find it hard to focus on any others.)
• Any expert in any field can be seen as highly intelligent.

Practice

• Steve was able to get to remembering 82 digits, whereas Renée only managed to get to 20. The difference was in the quality of their practice.
• Deliberate practice is the most effective method.
• If you do deliberate practice diligently, you will almost certainly become an expert.
• The best types of practice are probably not much fun (the experts aren't the best because they love these effective practice sessions).
• Studies found no violin experts who became experts with fewer than many thousands of hours of practice. There were no prodigies.
• And, in these studies, the people who practiced more tended to be better.

Motivation, not willpower

• Why do experts keep practicing. It's not because they like it.
  • e.g. In a study of spelling bee contestants, the very best spellers reported that they didn't really like practicing.
  • How do you keep going?
    • Willpower isn't transferrable. No evidence that people who have the "willpower" to practice spelling will also have the willpower to practice cooking. In fact, there is evidence showing that willpower is very dependent on activity.
  • Think of willpower like smarts - maybe there is some trait for it, but you can't change that, so focus on what you can control.
• Focus on shaping your motivation, not willpower.
  • e.g. People who are successful at losing weight redesign their lives and form habits that make it possible to keep the weight off.
  • The same is true for deliberate practice: develop habits that keep you doing it.
• Author says that if you're serious about developing a skill, dedicate an hour a day to it. Set aside a fixed time for this.
• Two ways you can keep a habit. Do both:
  • Strengthen the reasons to keep going.
  • Weaken the reasons to stop.

Types of practice

Naive practice

• Just doing the thing without any feedback.
• Improvement will plateau after a period.
• e.g. Playing tennis for 30 years.
• Part of the problem: You stop going out of your comfort zone.

Purposeful practice

• Overall goal: improve your mental representations.
• Well-defined goals.
  • Allow success to be judged.
  • e.g. "Play the piece all the way though without a mistake three times in a row."
  • Goals are small enough to focus on one aspect of the skill to improve on, and to make it clear if improvement is happening.
    • e.g. "Reduce handicap by five strokes" is a fine long-term goal, but not specific enough to generate specific parts of the skill to focus on.
• Giving the task your full attention.
• Quick feedback to self
  • On whether an action resulted in success or failure.
  • On the weaker parts of your process.
• Being out of your comfort zone.
  • Want to be in your zone of proximal development.
  • May need to push through blocks.
    • Use different strategies.
    • Build new techniques.
    • Make one aspect of the task a little easier in order to a) show it's possible to improve and b) give a chance of building a new technique.
      • e.g. Steve remembering strings of random digits. The trainer slowed the speed he read out the digits. Steve remembered many more digits. He realized the problem was that he was encoding the digits too slowly.
    • In Ericsson's research, he has found very few limits on improvement in any task.
• Seeing improvement.
  • Motivation comes from either seeing yourself improve or having others point out your improvement.

Deliberate practice

• All of purposeful practice, plus "informed practice".
  • Informed practice
    • Knows where it's going and how to get there
    • Two parts
      • A well-developed field
        • Objective measurement of skill
          • The skill level of the best performers is clearly set apart from the skill level of novices.
          • There is direct competition.
          • If you can't measure improvement, you can't know what the best training techniques for improving performance are.
        • Consensus on effective teaching methods.
        • Enough competition to provide incentives to practice and improve.
        • Building higher-level skills on top of solid lower level skills.
          • e.g. Finger placement in the piano.
      • A teacher. (This is crucial.)
        • Provide practice activities.
        • Provide feedback.
        • Every teacher knows that no student can work things out (e.g. proper finger placement) on their own.
        • Provide better mental models.
• Fields where deliberate practice is possible: music, dance, most sports done by individuals, chess.
• Even if deliberate practice is not strictly possible for your chosen field, you can still use the tenets of deliberate practice to guide you.

Effective practice techniques

• Adopt the practice techniques of experts.
  • e.g. Nurmi, an expert runner, dominated any length of distance running he chose to train in. Eventually, other runners learnt to match him by adopting his training techniques: interval training, using stopwatch, training year round.
• When practicing, if something works, keep doing it, if it doesn't, stop.
• Feedback loop
  • E.g. fighter pilots
    • Dogfight.
    • Afterwards, instructors ask:
      • What actions did you take? Why?
      • What were your mistakes?
      • What could you have done differently?
    • Instructors would give suggestions for things to try next time.
    • After a while, the trainees would start do do this process by themselves to avoid the pain of having the instructors do it.
• Focus practice on one small, specific skill at a time.
• Tweak everyday work activities into sessions where you practice a specific sub-skill.
• Go through a library of exercises with clear answers (e.g. a set of mammograms where it's known if they show cancerous lesions or not) and get feedback on whether you were right or wrong.
• Design exercises that target common mistakes or broken mental models.
  • e.g. A common mistake in a bile duct operation is to mistake a cystic duct for the bile duct. Laparoscopy students were shown videos of surgeries where the surgeon approached a cystic duct and asked to say what they'd do next. Note how this targets not just a mistake but a broken mental model (what a bile duct looks like).
• Focus on improving skills, not knowledge.
• Be careful not to drift off during your practice sessions. Great practice means paying attention to every detail of performance.
  • E.g. The swimmer trying to make each stroke as close to perfect as possible.
• You shouldn't be having too much fun.
• Tear apart high quality performances.
  • E.g. Chess players studying the games of other players. They study the board, predict the next move, then if they get it wrong they think about what they missed.
• If you hit a plateau.
  • Isolate the part of the skill you're unable to do. You're probably convinced you've reached some natural limit on it. Find a way to do this thing, by doing it in a new way.
    • How do you isolate the weaker part of the skill?
      • When you're practicing, it will probably be only one or two parts of the skill that are holding you back. Find them.
        • One way to do this is to put more pressure on yourself. This will make your weak areas more obvious.
          • e.g. Playing a better opponent at tennis.
          • e.g. A manager managing when work is chaotic and intense.
    • e.g. The digit memorizer hit a limit of the number of digits he could remember. His trainer slowed down the pace of reading out the digits. The memorizer could remember more than before. Now the way to improve was clear: get faster at encoding the digits.
    • New ways
      • Relax a constraint like the memorizer did.
      • Change the practice exercise
        • e.g. Bodybuilders changing the weights they're lifting.
• If you're improving, keep doing a practice exercise. If you're not, find a new one.
  • This dovetails well with the plateau advice above: when you plateau, change the exercise.
• Schedule practice. That way, it happens. Make sure nothing can conflict with it.
  • (I think this is why it's much more effective for me to practice algorithms at Trade rather than trying to do it at work. My practice is not going very well at
• Break practice into hour long chunks with breaks between. You can't concentrate intensely for more than an hour.
• It's really motivating to detect yourself improving. To aid this, break the skill into small pieces.
• Reading material to gain familiarly before a class. (E.g. When I read the Airtable onboarding documents before the sessions.)

Finding a coach/expert

• Find someone who is exposed to many other practitioners. This breadth of experience means their recommendation of an expert may be better because they can compare people and interact with lots of people.
• Look at the characteristics you think good performers have. Then list people who have those characteristics.
• If you look at testimonials, focus on how much the student improved, not how fun the lessons were.
• Ask about what the practice exercises will be - make sure they will guide you on what to practice, what to pay attention to, what errors you're making and how to recognize a good performance.

10,000 hours

• Wrong in several ways:
  • The best violin students average about this much practice by the time they're about 20. That doesn't really make much sense as a "special" figure because 20 is an arbitrary age, and, at this point, these students are very good, but not yet masters. Gladwell chose 10,000 hours because it's a nice round number.
  • 10,000 hours at age 20 is an average. Some very good students will have practiced less, some more by 20.
  • Didn't distinguish between any old practice and deliberate practice. The former would not be effective.
• Right in one:
  • Becoming an expert in an established field requires huge effort over many years.

#notebook #medianotes