Skip to content

Learning

This page contains notes from the MOOC Learning How to Learn on Coursera. I expect there to be much overlap in scope and content between this and Memory.

Week 1

Diffuse vs Focused Thinking

  • Focused thought is analogous to conscious, task-related thought. Could also be analogous to Csikszentmihalyi's flow or a telic/goal-oriented state.
  • Diffuse thought is analogous to the default mode network.
  • Focused thought constricts our neural activity to very small, specific networks in the brain. Diffuse thought allows activity to branch more readily from one network to a possibly unrelated one, such as synesthesia.
  • To learn a new skill or discipline, you need to balance diffuse and focused thought (similar to balancing exercise and resting). Focused learning like a workout.

Procrastination

  • Thoughts about a task can activate the insular cortex, leading to feelings of pain. Do cope with this, people may switch to a more pleasurable activity. However, if you just start the undesirable task the feelings of pain quickly taper off. This is reminiscent of some aspects of avoidant behavior that might be treated by CBT.
  • Recommendation of time-boxing technique (Pomodoro).

Practice

  • Math might be more difficult to master because it involves more abstraction.
  • You want to strengthen neural connections through repetition- what fires together...
  • Importance of depth

Memory

  • Working memory based in prefrontal cortex.
  • Working memory doesn't hold 7 chunks, it holds 4 chunks. No magic number 7.
  • Working memory is like a CPU cache- very small, only holds a little when closer to the core.
  • LTM like a warehouse, distributed over an area. LTM is a NoSQL database? Repetition of facts / procedures when learning is like a manual replication procedure sort of?
  • Spaced repetition is recommended. Cramming is bad.

Sleep

  • Wakefulness state generates toxic byproducts in brain. Brain cells (neurons? glia?) shrink during sleep, allowing toxins to be washed away. Sleep as self-cleaning (Xie et al, 2013).
  • Coffee/caffeine then is not inherently bad in the short-term, but can prevent a necessary cleaning.
  • Sleep prunes unwanted info, strengthens wanted info (sort of like deduplication).
  • Sleep is diffuse mode / dmn on steroids; allows connections to be made that might not otherwise be made under the watchful eye of the prefrontal cortex.

Sejnowski Interview

  • Mental multi-tasking as context switching with a single core CPU (since most people aren't actually able to attend to multiple things at once).
  • An enriching environment (undergrad, companies with good culture- i.e., Bell Labs) leads to stronger connections in the hippocampus; impoverished environments (cubical farms, isolation) less so.
  • In the absence of an enriching environment, exercise can increase neurons (i.e., increase BDNF)

Week 2

Chunking

  • Chunk: A set of relationships tying declarative facts together (like a graph). A network of neurons that are strongly interconnected through repeated activation. The cohesiveness of a chunk makes the individual facts it contains easier to remember. Helps create a "big picture" rather than getting tunnel vision about individual facts. A chunk is analogous to a simplified mental map (see Mappers vs Packers; the folks on that Wiki aren't rigorous personality psychologists though so I take it with a grain of salt).
  • Stress diminishes our ability to form chunks via our working memory.
  • Chunking is a bit like compression.
  • Chunks as schemata?

How to chunk

  • Get a sense of a pattern for what you want to learn (i.e., listen or watch someone playing to learn a song; mentorship / learning from someone else who has more mastery). Learn specific mini-chunks (i.e., musical phrases for a song). Synthesize these mini-chunks into a larger chunk.
  • Goal is to make a chunk second nature / not hard to activate (i.e., language).
  • Learning at the start involves a heavy cognitive load, so it makes sense to use example / toy problems and try to imitate them. Like working up gradually from 1 mile runs to 5k to 15k. Worked out solutions are a resource- you still need to use your own judgement to construct chunks.
  • A step-by-step overview of a general chunking process:
    1. Focus on the information that you would like to chunk/synthesize (i.e., turn off your phone, minimize interrupts).
    2. Comprehend concept (alternate between focused and diffuse networks if you do not have a good gist of concept). Direct experience is often necessary for this (since being able to do something purely in theory doesn't really confer mastery).
    3. Practice. Gain context for the chunk (i.e., where it should be connected, and where it should not be).
      • Chunking is bottom-up (via practice), and top-down (taking currently existent networks and connecting them to new discrete chunks). Context as a bridge between practice and higher-level schema.
      • Context can be gained by skimming chapter section titles and figures first.

Illusions of Competence

  • Re-reading a chapter in a textbook can be ineffective at crystallizing a memory. More effective is to read a chapter once, look away, and see how much you can recall about its contents (an active approach).
  • Retrieval of knowledge itself can enhance connections.
  • Making a semantic map for studying / organizing before you actually have a mental map in your head is not as effective.
  • Re-readings are effective when spaced.
  • Common illusions of competence:
    • Reading a worked out problem and feeling that you totally understand it conceptually, but you haven't put any mental effort into it yourself / haven't actively learned it so you don't actually know it. Exposure / superficial understanding alone is not enough.
    • Highlighting and underlining is not useful when overdone. Be minimal and look for main ideas beforehand. Margin notes, however are good if used to make connections.
    • Mind-mapping is not as effective as simple recall in learning. It's more important to have the nodes of knowledge set in memory rather than the connections- the connections will result later as long as the knowledge is there.
    • Thinking material is internalized just because it's available (via Google, textbook).
  • The best way to know if you're actually learning is to test yourself. Recall can be seen as a mini-test. Any mistakes you make in your self-test can be used to re-educate yourself (for systems administration, this is what labs / testing environments are for).
  • Switch up your environment when learning (environment-specific learning).

Sejnowski Lectures

  • Alpha in monkey troops has highest levels of serotonin. Serotonin associated with lower risk-taking (?). Inmates in prisons have lowest levels of serotonon.
  • Acetycholine involved in focus (i.e., Alpha GPC or eggs).
  • Dopamine can be implicated in long-term rewards as well

The Value of a Library of Chunks

  • Successful CEOS (Bill Gates, etc) set aside week long reading periods... therefore you should too.
  • Gradually build up chunks (e.g., chessmasters have library of patterns that they acquire as they gain expertise). Design Patterns in programming, related fields.
  • Transfer: Concepts and problem-solving methods from one field can help you / be re-appropriated for use in another field.
  • Manners of tackling problems: Sequentially (focused mode) or holistically (diffuse mode).
  • The diffuse mode can make connections between tightly-coupled chunks that were created in the focused mode.
  • Diffuse mode insights should be verified with focused mode (since intuition is not always correct; we are in fact very irrational animals).
  • Law of serendipity: Once one concept is added to mental library, more concepts will come more easily.

Overlearning

  • Overlearning is generally good if you can establish some level of automaticity for a skill / knowledge.
  • Continuing to practice after you've mastered as much as you can in a study session can be a waste of time. Like learning how to do carpentry with only a hammer (i.e., you get a very constricted view of knowledge). It's important to walk away.
  • Re-learning something you've already learned can make you feel better at something than you are, since you've actually only mastered the easy stuff and not moved on to harder topics (i.e., there's always something more difficult to learn / fields are nuanced).
  • Deliberate practice: Deliberately focus on more difficult topics rather than what you've already mastered.
  • Einstellung: Your initial idea might prevent a better idea from being found (i.e., functional fixedness).
  • If you reinforce your initial idea, you need to unlearn it (i.e., if you are taught that ego depletion or any of the numerous non-replicable effects in psychology are true/valid, it may be harder to pivot your paradigm).
  • Interleaving: Switching between using different tools/operators in different situations. Try using different concepts, procedures, etc. Skip between different chapters.
  • Just knowing how to use a tool isn't enough, you need to know when.
  • Developing expertise in multiple fields means you can have cross-pollination of ideas. But your expertise may be uneven compared to a specialist. One discipline specialists might be too entrenched in their field and not be able to come up with novel approaches.

Week 3

Procrastination

  • As we repeat processes, they become automatic on some level (habits). There is a loop of stimulus, response and reward (conditioning) that reinforces this automaticity. Most habits allow us to conserve cognitive resources, by taking common tasks and having them automated at a low-level of the brain. Bad habits (like watching too much Andromeda or Gordon Ramsay's Kitchen Nightmares or some other mix of trashy television), however, can also be automated, which is not such a good thing.
  • These bad habits can be likened to addictions.
  • Techniques
    • Focus on process rather than product. It's more important to kick off the steps to get to the answer (Pareto principle) than the end result (which if you contemplate beforehand can cause you internal pain). Automaticity is more process-friendly than product-friendly.
    • What Marcus Aurelius said
  • Replacing bad habits with good ones (CBT)
    • Determine stimulus (environment, time) that causes bad habit to kick in.
    • Establish a new habit for studying, possibly with a different stimulus (i.e., the learning room rather than the common room in the apartment).
    • Find a new reward for the good habit.
    • Replace your belief system (stoicism looks good).
  • Write a weekly todo list and then write daily lists each day in the evening. Tip: Mix mentally intense tasks with autopilot physical tasks. Tip 2: Plan a quitting time.
  • Leisure is important; people with hobbies outperform people whose work life is their entire life.

More Memory

  • Multimodal learning is good. The instructors champion co-opting visuospatial systems (since they are more evolutionary well-honed) in particular (citing this)- related to method of Loci.
  • A discussion of HM.
  • A discussion of mnemonics- including making up phrases / acronyms and the method of Loci. Mnemonics don't incur a penalty to get to semi-skilled expertise (?)
  • Spacing works because it invokes reconsolidating memories / updating them from an inactive state, thus making them stronger (i.e., some amount of elaboration must occur when the memories are being re-consolidated; I don't think this is the traditional view of this theory however)

Week 4

  • Periods where there is difficulty in understanding new knowledge / skills might be reflective of a period of mental re-structuring.
  • Creating metaphors / relations to concepts can help learning (also can help in creating new knowledge, e.g., fluidic view of electricity in 18th century).
  • Metaphors can help you get out of functional fixedness (local minima)
  • Personifying a concept and pretending to be it could be helpful in some cases.
  • Natural intelligence is important, but even just mediocre intelligence with lots of practice can do great things.
    • “Perseverance is a virtue of the less brilliant.” --Ramon y Cajal
  • Imposter syndrome is not as uncommon as you'd think (reassuring).
  • What you learn from one source is only one part of the big picture (i.e., it's a good idea to seek out many different texts / learning materials / lectures and synthesize them yourself).
  • The right hemisphere is biased towards big picture thinking (though left/right dichotomy is not strict). Right hemisphere = "devil's advocate" that questions status quo, left hemisphere is more rigid.
  • Social study sessions can help you realize knowledge gaps (as long as everyone's prepared and on-topic).
  • A checklist for tests
  • Tests as a way of learning (periodically quizzing yourself- 1 hr of testing more useful than 1 hr of studying in some ways).
  • Start with hard problems on a test, switch to easy problems after a minute or two (if you're not done). Hard problems will still be primed when you return to them later, and diffuse thinking may lead you to conclusions more easily (hard start - jump to easy).