Heuristics & Biases

Ruby (+108/-91)
Ruby
brook (+1852/-1228)
Ruby (+75/-16)
Ruby (+15)
Ruby (+69/-75)
brook (+3139/-4)
Ruby (+29/-3)
jimrandomh (+149/-210)
Vladimir_Nesov a bit of tidying up

Heuristics and Biases are the ways human reasoning differs from a theoretical ideal agent, due to reasoning shortcuts that don'don't always work (heuristics) and systematic errors (biases).

See also: Affect Heuristic, Confirmation Bias, Fallacies, Predictably Wrong, Rationality, Your Intuitions Are Not Magic, Bias, Heuristic

Cognitive biases”biases” are those obstacles to truth which are produced, not by the cost of information, nor by limited computing power, but by the shape of our own mental machinery. For example, our mental processes might be evolutionarily adapted to specifically believe some things that arent true, so that we could win political arguments in a tribal context. Or the mental machinery might be adapted not to particularly care whether something is true, such as when we feel the urge to believe what others believe to get along socially. Or the bias may be a side-effect of a useful reasoning heuristic. The availability heuristic is not itself a bias, but it gives rise to them; the machinery uses an algorithm (give things more evidential weight if they come to mind more readily) that does some good cognitive work but also produces systematic errors.

Our brains are doing something wrong, and after a lot of experimentation and/or heavy thinking, someone identifies the problem verbally and concretely; then we call it a “(“(cognitive) bias. Not to be confused with the colloquial “that“that person is biased, which just means “that“that person has a skewed or prejudiced attitude toward something.

We are here to pursue the great human quest for truth: for we have desperate need of the knowledge, and besides, we'we're curious. To this end let us strive to overcome whatever obstacles lie in our way, whether we call them “biases”“biases” or not.

It'It's also useful to know the kinds of faults human brains are prone to, in the same way it'it's useful to know that your car'car's brakes are a little gummy (so you don'don't sail through a red light and into an 18-wheeler).

Wait a minute... fallacies, biases, heuristics... what'what's the difference??

While a bias is always wrong, a heuristic is just a shortcut which may or may not give you an accurate answer. Just because you know complex mathematical methods for precisely calculating the flight of objects through space doesn'doesn't mean you should be using them to play volleyball. Which is to say, heuristics are necessary for actually getting anything done. But because they are just approximations they frequently produce biases, which is where the problem lies. "Fallacy""Fallacy" is often used to mean a very similar thing as bias on LessWrong. [Needs better clarification]

Well, no. If it were that easy we wouldn'wouldn't need a community initially dedicated to overcoming bias (the name of the blog which this website grew out of). Unfortunately, learning about a bias alone doesn'doesn't seem to improve your ability to avoid it in real life. There'There's also the (major) issue that knowing about biases can hurt people. Instead of being purely focused on removing negative habits, there is now a major focus at LessWrong to implementing positive habits. These are skills such as how to update (change your mind) the correct amount in response to evidence, how to resolve disagreements with others, how to introspect, and many more.

See also: Affect Heuristic, Confirmation Bias, Fallacies, Predictably Wrong, Rationality, Your Intuitions Are Not Magic

The field“Cognitive biases” are those obstacles to truth which are produced, not by the cost of heuristics and biases was essentially createdinformation, nor by Kahneman and Tverskylimited computing power, but by the shape of our own mental machinery. For example, our mental processes might be evolutionarily adapted to specifically believe some things that arent true, so that we could win political arguments in a seriestribal context. Or the mental machinery might be adapted not to particularly care whether something is true, such as when we feel the urge to believe what others believe to get along socially. Or the bias may be a side-effect of experiments provinga useful reasoning heuristic. The availability heuristic is not itself a bias, but it gives rise to them; the machinery uses an algorithm (give things more evidential weight if they come to mind more readily) that people consistently makedoes some good cognitive work but also produces systematic errors.

Our brains are doing something wrong, and after a setlot of errors when judging problems experimentation and/or heavy thinking, someone identifies the problem verbally and concretely; then we call it a “(cognitive) bias.” Not to be confused with the colloquial “that person is biased,” which just means “that person has a skewed or prejudiced attitude toward something.”

A bias is an obstacle to our goal of obtaining truth, and thus in our way.

We are here to pursue the great human quest for truth: for we have exact statistical answers. This isndesperate need of the knowledge, and besides, we'tre curious. To this end let us strive to say that humans only make errors on this typeovercome whatever obstacles lie in our way, whether we call them “biases” or not.

It's also useful to know the kinds of problem;faults human brains are prone to, in the same way it's justuseful to know that your car's brakes are a lot easier to track the errors they are making whenlittle gummy (so you knowdon't sail through a red light and into an answer precisely.

This has a number of connotations; firstly, more people guessing the same thing doesn't necessarily improve the quality of your answer, contrary to what your naïve expectation may be. Secondly, more empirical information can worsen your ability to predict future outcomes as you become more certain of a biased conclusion. Thirdly, it seems likely that if we can systematically improve cognitive biases then it would benefit us in achieving many goals. 18-wheeler).

The Sequence, Predictably Wrong, offers a goodan excellent introduction to the topic for those who are not familiar.

While a bias is always wrong, and gives you the wrong answer, a heuristic is just a shortcut which may or may not give you an accurate answer. Just because you know complex mathematical methods for precisely calculating the flight of objects through space doesn't mean you should be using them to play volleyball. HeuristicsWhich is to say, heuristics are necessary for actually getting anything done, butdone. But because they are just approximations they frequently produce biases, which is where the problem lies. A further confusion"Fallacy" is with fallacies, which are errors in reasoning, rather than flaws in the reasoning algorithm [unclear, rewrite].often used to mean a very similar thing as bias on LessWrong. [Needs better clarification]

A good example of a heuristic is the affect heuristic-- people tend to guess unknown traits about people or things based on the perceived goodness of badness of known traits, whether or not they would explicitly expect these traits to correlate with one another.traits. In some circumstances this is a useful shortcut-- you may like to assume, for instance, that people who are good singers are more likely to be good dancers, too. However, it also frequently produces (unconscious) biases-- a bias towards believing that people who are tall and good looking have better moral character, for instance. An example of a fallacy, on the other hand, would be planning fallacy-- people's explicit models of how long things will take to get done are consistently very optimistic.

While a bias is always wrong, and gives you the wrong answer, a heuristic is just a shortcut which may or may not give you an accurate answer. Just because you know complex mathematical methods for precisely calculating the flight of objects through space doesn't mean you should be using them to play volleyball. Heuristics are necessary for actually getting anything done, but because they are just approximations they frequently produce biases, which is where the problem lies. A further confusion is with fallacies, which are errors in explicit logic. reasoning, rather than flaws in the reasoning algorithm [unclear, rewrite].

The Sequence, Predictably Wrong, offers a good introduction to the topic for those who are not familiar.

Heuristics and Biases are the ways human reasoning differs from a theoretical ideal agent, due to reasoning shortcuts that don'don't always work (heuristics) and systematic errors (biases).


Basics

The field of heuristics and biases was essentially created by Kahneman and Tversky in a series of experiments proving that people consistently make a set of errors when judging problems that have exact statistical answers. This isn't to say that humans only make errors on this type of problem; it's just a lot easier to track the errors they are making when you know an answer precisely.

This has a number of connotations; firstly, more people guessing the same thing doesn't necessarily improve the quality of your answer, contrary to what your naïve expectation may be. Secondly, more empirical information can worsen your ability to predict future outcomes as you become more certain of a biased conclusion. Thirdly, it seems likely that if we can systematically improve cognitive biases then it would benefit us in achieving many goals.

Predictably Wrong offers a good introduction to the topic for those who are not familiar.

Wait a minute... fallacies, biases, heuristics... what's the difference??

While a bias is always wrong, and gives you the wrong answer, a heuristic is just a shortcut which may or may not give you an accurate answer. Just because you know complex mathematical methods for precisely calculating the flight of objects through space doesn't mean you should be using them to play volleyball. Heuristics are necessary for actually getting anything done, but because they are just approximations they frequently produce biases, which is where the problem lies. A further confusion is with fallacies, which are errors in explicit logic.

A good example of a heuristic is the affect heuristic-- people tend to guess unknown traits about people or things based on the perceived goodness of badness of known traits, whether or not they would explicitly expect these traits to correlate with one another. In some circumstances this is a useful shortcut-- you may like to assume, for instance, that people who are good singers are more likely to be good dancers, too. However, it also frequently produces (unconscious) biases-- a bias towards believing that people who are tall and good looking have better moral character, for instance. An example of a fallacy, on the other hand, would be planning fallacy-- people's explicit models of how long things will take to get done are consistently very optimistic.

So if I learn all the biases, I can conquer the world with my superior intellect?

Well, no. If it were that easy we wouldn't need a community initially dedicated to overcoming bias (the name of the blog which this website grew out of). Unfortunately, learning about a bias alone doesn't seem to improve your ability to avoid it in real life. There's also the (major) issue that knowing about biases can hurt people. Instead of being purely focused on removing negative habits, there is now a major focus at LessWrong to implementing positive habits. These are skills such as how to update (change your mind) the correct amount in response to evidence, how to resolve disagreements with others, how to introspect, and many more.


Related Tags

Affect Heuristic

Confirmation Bias

Fallacies

Rationality

Techniques

TheHeuristics and Biases are the ways human reasoning differs from a theoretical ideal agent, due to reasoning shortcuts that don't always work (heuristics) and systematic errors (biases).

The heuristicsways human reasoning differs from a theoretical ideal agent, due to reasoning shortcuts that don't always work (heuristics) and biases program in cognitive psychology tries to work backward from biases (experimentally reproducible human errors) to heuristics (the underlying mechanisms at work in the brain)systematic errors (biases).

See also

Load More (10/14)