= 27d567bc2d48d9f40e9ccc20ea370b15)
[CORE] Concepts for Understanding the World
Background:
I'm recently doing a big project to increase my scholarship and modeling power for both rationality and traditional "serious" topics. One thing I found very useful is taking notes with a clear structure.
The structure I'm using currently is as follows:
- write down useful concepts,
- write down (as a separate category) useful heuristics & things to do in various situations,
- do not write facts, opinions or anything else (I rely on unaided memory to get more filtering).
Heuristic: learn concepts before facts!
Note that you can be mistaken about facts, but you can't harm your epistemology by learning concepts. Even if a concept turns out to be useless or misleading, you are better off knowing about it, understanding how it's misleading, and being able to avoid the trap when you see it.
Let's share concepts!
Please give (at a minimum) a name and a reference (link). A short description in plain language is also welcome.
I started to write these "concept" notes electronically only 2 days ago, so the list below is VERY random and sparse. In any case, let's make this a collective effort :)
[ Added: up-to-date version of the list below is here: http://squirrelinhell.github.io/concepts/ ]
adaptation-executer
➡ organism acts only to execute evolved adaptations
➡ it does not depend on awareness of the goal (maximizing inclusive fitness)
adaptive capacity
➡ ability of a system to adapt to changing environment
akrasia
➡ acting against your own better judgement
Aristotelian epistemology
➡ statements are true or false
➡ truth can be shown by deductive reasoning
base rate neglect
➡ using specific evidence as if it replaced priors
Bayesian epistemology
➡ beliefs have degrees of certainty from 0 to 1
➡ updated upon seeing evidence
behaviorism
➡ an approach that explains behavior by conditioning (rather than thoughts or emotions)
brotherhood
➡ all-male social group with rules that prevent girl-related drama within the group
CDT Stage 1
➡ subject to reflexes
➡ early childhood
CDT Stage 2
➡ reflexes are objects
➡ subject to personal interests (desires, needs, wishes)
➡ childhood to early adolescence
CDT Stage 3
➡ personal interests are objects
➡ subject to relationships/social pressure
CDT Stage 4
➡ relationships are objects
➡ subject to systems (principles, rules, structure, commitments)
CDT Stage 5
➡ systems are objects
➡ can juggle around multiple systems
➡ can handle errors or incompleteness of a system
Chesterton Fence
➡ rule not to do reforms without understanding reasoning behind current state
cognitive ease
➡ how hard it is to think something gets used as a proxy for more complicated judgements
Conservation of Expected Evidence
➡ expectation of posterior probability is equal to prior probability
➡ unlikely strong evidence is balanced by likely weak evidence in the opposite direction
➡ absence of evidence is (weak) evidence of absence
Counterfactual Mugging
➡ Omega says it had tossed a fair coin to determine if you win $10000 or lose $100
➡ you only get $10000 if you would accept loss of $100
➡ it is already known that you lost
➡ do you pay $100?
countersignaling
➡ not signaling X to show that you are above people who signal X
➡ doesn't work when confused with lower level
Curse of Development
➡ when making progress, effectiveness will often decrease before it becomes higher than ever
denominator neglect
➡ in "X out of Y", absolute value of X influences intuitive judgement
dimimishing returns
➡ decreasing marginal output, as input is incrementally increased
Dunnig-Kruger effect
➡ bias in which relatively unskilled persons overestimate their skill a lot
ending on a high note
➡ the last part leaves the strongest impression
➡ either (1) do something positive at the end
➡ or (2) end the interaction after something positive happens
exposure therapy
➡ treatment for anxiety or phobias by forced prolonged exposure while safety is guaranteed
forced legibility
➡ reorganising a complex system which seems irrational, but it was only a failure to understand it
halo effect
➡ seeing something positive or negative influences judgement of all other aspects
homeostasis
➡ property of a system that regulates a variable to keep it at a constant level
honing mode
➡ conversation converges on an idea
➡ focused on corrections and critique
hyperbolic discounting
➡ value(t) = 1/(1 + C * t)
➡ intuition discounts approximately like this
➡ any discounting except exponential is inconsistent under passage of time
inclusive fitness
➡ ability to pass on genes (including genes passed on by relatives)
iterated hurt
➡ someone is hurt by the knowledge that you were willing to hurt them
➡ this works even if there is no actual hurt
jamming mode
➡ conversation is a divergent exploration
➡ based on remixing and building on ideas
Laws of Authority
➡ enforced by a selected powerful entity
Laws of Reality
➡ enforced automatically by the environment
Laws of Society
➡ enforced by group consensus
➡ large majority finds the terms beneficial or acceptable
➡ defectors are punished by volunteers
learned blank
➡ taking for granted lack of skill/knowledge in some area
level reversal
➡ level N+1 is sometimes superficially similar to level N-1
➡ that the next level is counterintuitive is what makes levels recognizable in the first place
loss aversion
➡ intuitive judgement is based on losses and gains relative to a "reference point"
➡ losses weigh around 2 times more than gains
lost purpose
➡ pursuing an instrumental goal that no longer has value
marginal cost
➡ change of cost per one additional unit of produced resource
marginal utility
➡ change of utility per one additional unit of consumed resource
marginalism
➡ a theory that explains prices of products in terms of their marginal utility
meta-contrarianism
➡ being contrarian to a contrarian position
➡ might be more concerned with signaling than accuracy
motivation system
➡ outputs "wanting" and "not wanting"
➡ implemented entirely in S1
mutual knowledge
➡ everyone knows, and they know that everyone knows, and they know that others know that everyone knows, etc.
mystic epistemology
➡ everything is possible
➡ let's believe nothing
Newcomb's problem
➡ Omega prepared a trasparent box and an opaque box
➡ transparent box contains $1000
➡ opaque box contains $1,000,000 iff Omega predicted you will take only the opaque box
➡ do you take both boxes, or only the opaque box?
Nyquist frequency
➡ half of the sampling rate
➡ for every sinusoid with frequency above, there is a sinusoid (alias) that has the same samples and frequency below
Parfit's Hitchhiker
➡ you will be saved from death on a desert by a car driver iff you promise to pay $100 later
➡ the driver can detect lying perfectly
➡ does your decision theory allow you to commit to paying with certainty you won't change it later?
Peter principle
➡ if promotion is based on performance in the current role, then it stabilizes on reaching incompetence
planning system
➡ consciously building chains of actions
➡ can imagine intermediate states
regression to the mean
➡ prediction from a correlated variable is attenuated by amount of correlation
➡ more noise or smaller sample requires bigger adjustment
reinforcement learning
➡ agent interacts with the environment
➡ correct actions are learned from when it gets rewards
Schelling fence
➡ "arbitrary boundary used to prevent a ""slippery slope"" situation"
Schelling point
➡ prediction of what others expect you to think you are expected to do
self-reinforcement
➡ using conditioning to influence your own patterns of behaviour
➡ doesn't work (learning is based on surprise)
➡ some techniques that work might look very similar (they actually manipulate emotions)
Smoking Lesion
➡ alternative world in which smoking is correlated with cancer but does not cause it
➡ there is a genetic lesion that increases chances of both
➡ you don't know if you have it
➡ you want to smoke, but you dislike cancer much more
➡ do you smoke?
societal collapse
➡ disintegration of a human society
➡ often together with most civilization advances
status illegibility
➡ social groups are held together by unclear relative statuses in the middle section
subject-object model
➡ a view of development as a shift from being subject to X, to manipulating X as an object in context
➡ intermediate stages: becoming aware of broader view, having 2 conflicting views, adopting the broader view with occasional lapses
sunk cost fallacy
➡ including resources that have already been spent in a decision about the future
S1 storytelling
➡ System 1 interprets data by constructing plausible stories
➡ it does it more easily with *less* data
System 1
➡ unconscious, effortless thinking
➡ fast, parallel and runs all the time
System 2
➡ conscious, effortful thinking
➡ slow, serial and runs on demand
➡ it can install new patterns of behaviour, but sucks at controlling anything directly
➡ uses working memory
ugh field
➡ mental flinch from thinking about something (or even admitting there is a problem)
vivid probability
➡ unlikely outcomes/events are overweighted when they evoke vivid mental imagery, and neglected otherwise
➡ e.g. explicitly mentioning, adding details, presenting probability as frequency all contribute to overweighing
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
= f037147d6e6c911a85753b9abdedda8d){kind=logo}
Comments (17)
A very useful concept I've found to be Yvain's noncentral fallacy.
Once I learned it I couldn't not unsee it, it appears in almost all discussions around me, both live and on the net.
I have also started a pursuit of learning useful concepts/models explicitly.
Some useful resources:
Model Thinking course on Coursera
Mental models by Gabriel Weinberg
Creating a Latticework of Mental Models: An Introduction
And, of course, the Useful Concepts Repository @ LessWrong
Cool, thanks! I didn't know about the third and the fourth.
After checking everything from comments here, I have enough material that it'll take me months to work through it all.
Also if you don't know it, Meaningness has some interesting remarks about formulating concepts and problem descriptions: http://meaningness.com/metablog/how-to-think#feynman-objects
Whoa, awesome! Aligns well with my current interests. A lot of great insights there...
http://lesswrong.com/r/discussion/lw/nqz/open_thread_jul_04_jul_10_2016/dcyy
This comment has a great link to a pretty big list of concepts.
Thanks, that's an awesome list. I'll work through it carefully.
Minor typo: hallo effect (should be halo effect). Also a nitpick: the negative version of the halo effect is usually called horn effect.
I'm also curious about why you have included "brotherhood" as a useful concept...
Ah, in that case I noticed that the modern definition of "brotherhood" tends to just be "all-male group", and that we have lost the cultural context in which it made sense originally. In the times when you had honor duels etc., having the second part in the definition of brotherhood (preventing girl drama) was crucial, and probably implicitly obvious to everyone (so it wasn't even mentioned). So after redefining it back, it makes sense that the modern world needs "brotherhoods" (and "sisterhoods") less than ever.
... more than ever?
If we assume that we are not more able to cope with those problems, we only fool ourselves that we are - then more. So it depends on how optimistic you are about the current society.
In any case this is not the first time when I realize that to make a concept more useful, I can adopt a definition that is similar, and yet crucially different, from the "common wisdom" one. One other example of this is my definition of mnemonic technique.
For rationality related concepts, see this page
Thanks, I've seen that before but it didn't come to mind now.
It covers a lot of the "rationality" part, though not so much of "understanding the world" in a broader sense (esp. economy, sociology, politics etc.)
I'm not sure how do you define concept. According to what I understood, I think you might be missing these:
Feed back https://en.wikipedia.org/wiki/Feedback the impact of something halts its cause.
feed forward https://en.wikipedia.org/wiki/Feed_forward_(control) the impact of something reinforce its cause
self fulfilling prophecy https://en.wikipedia.org/wiki/Self-fulfilling_prophecy a prophecy is being fulfilled because the prophecy was made, usually because active agents tried to prevent the prediction from happening
emergence https://en.wikipedia.org/wiki/Emergence a collection of simpler elements and behavior generating a more complex pattern when multiple elements are connected to each other.
I like the difference between single points and frontlines of influence. It comes up in vegetation science, where there is the Eternal Question of the Continuity of Communities; and some events and environments in my life felt like approaching lines, like the marriage, and some like points, like the wedding.
And, also from veg.sc., the notion of (scaled) phytosociological relevés. When you do that, you have to "edit" the picture of "grass and shrubbery" you see so that you can estimate the % of soil covered by different species, and not be distracted by flowers & dead plants, and probably recognize levels within the "grass", etc., and look for patchiness, etc., and find a way to express it all. I have known people compiling relevés in ten minutes flat; a colleague wave a hand at a spot of Erythronium caucasicum, half-screened by the rain, and remark "Huh, no clones here"... It just becomes a way of seeing stuff, with practice. Wiki
And lastly, the Image of the Species, which is the image of something you have encountered many times and recognize "directly", which might differ from such shaped by a different set of observations.
All this sounds interesting, but without more resources (or biology background) I'm not sure I'm getting this.
Is the frontline (as you mean it) only considered in time (not e.g. physical space)? I.e. it's just a different way of saying "something exerts influence for a period of time" vs "something changes suddenly"?
I think I get what the process looks like, but does it mean as a concept? E.g. what else would you use it to describe?
Do you mean the observation that human brains represent categories by remembering "typical examples" of items in that category?
No, it is usually used for space. Something like internal design of workplaces, or being distracted by a coughing fit at an opera, or placing the cherry on top of the cake, or skirting puddles, all of that:) But you can say, for exaple, that learning about human hormone system by reading about separate hormones gives you points of "illumination", and then imagining the profile of, for example, pregnancy, is more of a line. (Maybe?.. I seldom have to articulate that. For me, the "line" is more like the front of a cloudbank, where you know there is a whole bag of "weather" contained, but don't yet know what that weather would be.)
I guess I meant a certain skill, which allows to output a strictly formalized answer, has to be useful across a really wide set of circumstances, and when internalized feels like a rush of data and corrections.
No, it is rather a blended memory of all such organisms one sees. Like, "well, it is rather too oblong for yeast, but I still think it is yeast", you know? Typical examples are things people admire, and they should be the basis for the Images of the Species, but in practice I think it never happens and this is likely for the best.
Please find better names for these things, if you think they are useful. I simply remembered what I found applicable outside of botany, but, well:)
How about the cobra effect? It's a classic example of incentives yelding the exact opposite of what they were put in place for.