Summary: Reflections and practice notes on a metacognitive technique aimed at refining the process of thinking, rather than the thoughts themselves.
Epistemic Status: Experimental and observational, based on personal practice and reflections over a brief period.
Introduction
While doing a simple math problem, I realized that my faculty of thinking was often confused and inefficient. We were drawing polygons which corresponded to multi-holed toruses (or formally genus g surfaces), and in trying to generalize the square → one-holed torus pattern to an octagon, I completely forgot to use the previous pattern of ‘opposite sides gets glued together’ in any reasonable sense.
genus g surfaces
Zooming out to my broader life, I had been practicing Vipassana meditation for some time, and I was starting to notice incrementally finer thoughts, emotions, and thought patterns in my daily life.
Given the above experience, I hypothesized that if I focused my attention on how to think better rather than the specific thoughts, I can probably learn to learn, reason, and discover which much more efficiency. Importantly, I had a strong faith that our ability to learn is not a constant determined at birth.
Awareness of mental processes rather than objects of thoughts, and
Reward Mechanism for the quality of the thought process.
The two-step structure is rather similar to what I practice in Vipassana meditation, which teaches:
Fine-grained awareness of every sensation in the body, since every thought and emotion corresponds to a sensation, and
Equanimity with those sensations.
This tuning strategy basically imposes a reward modeling subprocess onto your mode of thinking. Instead of focusing on specific patterns of thinking, for instance a conscious steer towards more rigorous reasoning when intuition tries to jump in, the technique teaches you to assign conscious reward to your way of thinking which then informs the deeper parts of your consciousness what processes should be encouraged next time.
Even for the short interval that I’ve practiced, I’ve found that this technique is extremely powerful in accelerating efficiency of thought. Since it interferes with the subconscious mind, I advise caution.
There’s not a lot of case studies online for metacognitive tuning strategies, so I thought I’ll post my notes along the way. I’ll try to organize them as coherently as I can, but most of them will be swiftly edited notes.
Notes
The technique suggested to tackle some quick and interesting problem to record your thoughts, so I chose some Fermi questions and jotted down my thoughts.
Example 1:
How many minutes of sunlight does Stockholm, Sweden, experience on Midsummer’s Day? (Assuming clear skies.) [From the archives of Estimation Game].
Thoughts: Where is Sweden? I remember the day in London when the sun was setting at 10. Probably rising at 4? Sweden is north of London. So lets say 6 minus around 2 hours of night time as upper bound for sunlight and 6 hours of night time as lower bound. So let’s say between 1200 minutes and 1000 minutes of sunlight.
Analysis: judgement of the relevance of personal experience in this question; quick geographical intuition about the fact that sunlight time deviates more the more north you go.
Example 2:
How many words long is the Shakespeare play, A Midsummer Night’s Dream? [From the archives of Estimation Game]
Thoughts: 2 acts and 7 scenes for each act? I’m not sure how many words are in a scene. How long is midsummer night’s dream compared with Romeo and Juliet, which I read? Maybe 200 pages, 20 lines a page, and 10 words per line? So like 40000?
Analysis: failure to evaluate the reliability of my past experience—if I thought about it I don’t actually know the length of Romeo and Juliet; failure to make a common sense check at each progression—about act and scene; good use of reliable evidence—Shakespearean plays were mostly in iambic pentameter so 10 words per line is a reasonable estimate
I mainly chose to write down these observations to practice, but as the skill of metacognitive awareness gets built up, the process should be automated.
What’s going on?
I feel like you can sort of understand your mind as chains since from observing my own thoughts, no two thoughts ever come at the same time—they merely appear and disappear very fast. Hence every unit of thought can be broken down into loops in a chain l1,...,ln which may itself be a chain.
This technique can be understood as an important connective in an attempt to systemize your cognitive strategies. An orchestra of sporadically playing instruments now welcomes a conductor. The conductor must know what she wants, or at least have a taste for what’s good. They pay attention to the action on the stage, note its effectiveness, and introduce new players on the way. That is to say, this technique tells you how you can steer your mind—through self-awareness at a recursively self-referential level, provided you have some idea about where to steer it to.
I’m still thinking about this but it seems useful to develop a language for describing the types of thought processes that I’d want at different moments: common sense checkers, perspective flips, scaling a problem down and back up, etc. It quickly appears necessary to introduce some hierarchy or structure to these components.
On Risks, Costs, and Over-Regulation
This practice may be too cognitively intense if I push it too far. Today I was diving one or two layer deeper on almost every thought I had and intensively evaluating them gave noticeable albeit unsustainable results. I should try to 1) do the reward part more subtly and 2) add an encourager for thinking about nothing when I’m tired. How do I think about nothing? Relaxing all muscles may be a way. [...]
I spoke with my mother today who warned me of the pitfall of over-regulating my life and thoughts and the energy cost of turning this feature on. What might this turn into? They say you lose joy when you try to control everything. And how can I truly judge every thought chain I have well? Perhaps mere awareness is sufficient to tune your meta-cognition and a reward signal would be unconsciously processed. At the very least, a short analysis of the thought chain should give the inaccessible parts of your brain enough information to reward itself. I am averse to turning myself into an RL agent and this direction of strengthening awareness seems more appropriate. At the very least, I should narrow the scope of this technique at its strongest to learning environments i.e. math, which has a pretty clear metric of success and benefits considerably from this optimization. [...]
I’ve decided that any thought associated with any emotion should not be part of this exercise. If an emotion arise and I reward or punish the pathway that led to it based on how good I think it is, I am effectively pushing these things back into the depths of my mind. In this case what I want is long-term peace and clarity. My past practice has informed me that the only way to eradicate residue emotions, especially negative ones, is to welcome it and let it bubble up, and then trace it to its very origin. To put it back into the language of the technique, that is to say the only thought process I reward here is the “stay with it, let it be, dig deeper” chain.
Now that leaves a concentrated pool of cognitive processes to be analytically tuned i.e. thought processes under the umbrella of problem solving, research, learning, and general intellectual activity.
Conclusion
So far, I’ve jotted down my immediate thoughts in the 72 hours after coming in contact with this technique. I expect to post more notes as I integrate this tuning strategy into my life. Past experiences with meditation have taught me to be extremely cautious when it comes to techniques that directly modify the mind, but systematizing cognitive strategies seems to be a strictly positive endeavor when limited to the scope of rational thinking. So far, I’ve noticed considerable improvement in focus, creativity, and structure when approaching a problem. I expect that this technique will be most useful to people looking to bridge a gap between a known goal of good thought processes and their current state.
Notes on Tuning Metacognition
Summary: Reflections and practice notes on a metacognitive technique aimed at refining the process of thinking, rather than the thoughts themselves.
Epistemic Status: Experimental and observational, based on personal practice and reflections over a brief period.
Introduction
While doing a simple math problem, I realized that my faculty of thinking was often confused and inefficient. We were drawing polygons which corresponded to multi-holed toruses (or formally genus g surfaces), and in trying to generalize the square → one-holed torus pattern to an octagon, I completely forgot to use the previous pattern of ‘opposite sides gets glued together’ in any reasonable sense.
Zooming out to my broader life, I had been practicing Vipassana meditation for some time, and I was starting to notice incrementally finer thoughts, emotions, and thought patterns in my daily life.
Given the above experience, I hypothesized that if I focused my attention on how to think better rather than the specific thoughts, I can probably learn to learn, reason, and discover which much more efficiency. Importantly, I had a strong faith that our ability to learn is not a constant determined at birth.
The Technique
I stumbled upon an enlightening post Tuning your Cognitive Strategies which taught, in my language, the following technique:
Awareness of mental processes rather than objects of thoughts, and
Reward Mechanism for the quality of the thought process.
The two-step structure is rather similar to what I practice in Vipassana meditation, which teaches:
Fine-grained awareness of every sensation in the body, since every thought and emotion corresponds to a sensation, and
Equanimity with those sensations.
This tuning strategy basically imposes a reward modeling subprocess onto your mode of thinking. Instead of focusing on specific patterns of thinking, for instance a conscious steer towards more rigorous reasoning when intuition tries to jump in, the technique teaches you to assign conscious reward to your way of thinking which then informs the deeper parts of your consciousness what processes should be encouraged next time.
Even for the short interval that I’ve practiced, I’ve found that this technique is extremely powerful in accelerating efficiency of thought. Since it interferes with the subconscious mind, I advise caution.
There’s not a lot of case studies online for metacognitive tuning strategies, so I thought I’ll post my notes along the way. I’ll try to organize them as coherently as I can, but most of them will be swiftly edited notes.
Notes
The technique suggested to tackle some quick and interesting problem to record your thoughts, so I chose some Fermi questions and jotted down my thoughts.
Example 1:
Example 2:
I mainly chose to write down these observations to practice, but as the skill of metacognitive awareness gets built up, the process should be automated.
What’s going on?
I feel like you can sort of understand your mind as chains since from observing my own thoughts, no two thoughts ever come at the same time—they merely appear and disappear very fast. Hence every unit of thought can be broken down into loops in a chain l1,...,ln which may itself be a chain.
This technique can be understood as an important connective in an attempt to systemize your cognitive strategies. An orchestra of sporadically playing instruments now welcomes a conductor. The conductor must know what she wants, or at least have a taste for what’s good. They pay attention to the action on the stage, note its effectiveness, and introduce new players on the way. That is to say, this technique tells you how you can steer your mind—through self-awareness at a recursively self-referential level, provided you have some idea about where to steer it to.
I’m still thinking about this but it seems useful to develop a language for describing the types of thought processes that I’d want at different moments: common sense checkers, perspective flips, scaling a problem down and back up, etc. It quickly appears necessary to introduce some hierarchy or structure to these components.
On Risks, Costs, and Over-Regulation
This practice may be too cognitively intense if I push it too far. Today I was diving one or two layer deeper on almost every thought I had and intensively evaluating them gave noticeable albeit unsustainable results. I should try to 1) do the reward part more subtly and 2) add an encourager for thinking about nothing when I’m tired. How do I think about nothing? Relaxing all muscles may be a way. [...]
I spoke with my mother today who warned me of the pitfall of over-regulating my life and thoughts and the energy cost of turning this feature on. What might this turn into? They say you lose joy when you try to control everything. And how can I truly judge every thought chain I have well? Perhaps mere awareness is sufficient to tune your meta-cognition and a reward signal would be unconsciously processed. At the very least, a short analysis of the thought chain should give the inaccessible parts of your brain enough information to reward itself. I am averse to turning myself into an RL agent and this direction of strengthening awareness seems more appropriate. At the very least, I should narrow the scope of this technique at its strongest to learning environments i.e. math, which has a pretty clear metric of success and benefits considerably from this optimization. [...]
I’ve decided that any thought associated with any emotion should not be part of this exercise. If an emotion arise and I reward or punish the pathway that led to it based on how good I think it is, I am effectively pushing these things back into the depths of my mind. In this case what I want is long-term peace and clarity. My past practice has informed me that the only way to eradicate residue emotions, especially negative ones, is to welcome it and let it bubble up, and then trace it to its very origin. To put it back into the language of the technique, that is to say the only thought process I reward here is the “stay with it, let it be, dig deeper” chain.
Now that leaves a concentrated pool of cognitive processes to be analytically tuned i.e. thought processes under the umbrella of problem solving, research, learning, and general intellectual activity.
Conclusion
So far, I’ve jotted down my immediate thoughts in the 72 hours after coming in contact with this technique. I expect to post more notes as I integrate this tuning strategy into my life. Past experiences with meditation have taught me to be extremely cautious when it comes to techniques that directly modify the mind, but systematizing cognitive strategies seems to be a strictly positive endeavor when limited to the scope of rational thinking. So far, I’ve noticed considerable improvement in focus, creativity, and structure when approaching a problem. I expect that this technique will be most useful to people looking to bridge a gap between a known goal of good thought processes and their current state.