|
Pages: [1] 2 3 ... 5
|
 |
|
Author
|
Topic: Cognitive Biases (Read 11176 times)
|
Zanthius
Enlightened
 Offline
Posts: 941
|
I am starting to write a new article about cognitive biases. Not because I feel like I have so much original to say (most have been discovered by psychologists already), but because I feel like this is so central to all my other theories.
|
|
|
|
« Last Edit: July 16, 2017, 12:54:10 am by Zanthius »
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Okay. Here is an early draft of the article:
http://www.archania.org/the_never-ending_battle_against_our_cognitive_biases.html
I didn't find any bias for this, so I invented one myself and called it "The illusion of homogeneous perfection".
I made this script that simulated the average value of coin tosses all up to a 100 000, and you can see, it is quite a bit of difference between 100, 1 000, 10 000, and 100 000. It doesn't get really close to an average value of 0.5 every time, before around 100 000 tosses. So imagine how much better a study based upon 100 000 individuals is, compared to a study based upon 100 individuals. At 100 coin tosses, it looks like you can get an average value anywhere between 0.65 and 0.35, which is quite a lot of uncertainty.
|
|
|
|
« Last Edit: July 19, 2017, 09:20:12 pm by Zanthius »
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
When I think about how harmful different drugs are, I often think about what condition people seem to be in that are using the drugs. Now I am thinking that maybe we can judge how harmful ideologies are according to what condition people that believe in the ideologies seem to be in. For example, many religions seem to limit how people develop their understanding. But even agnostic or atheistic ideologies can be intellectually limiting unless they focus a lot on overcoming cognitive biases. One of the most healthy types of ideologies a person can have, seems to be something like what Eliezer Yudkowsky has (the author of lesswrong.com), since it focuses so much on overcoming cognitive biases.
|
|
|
|
« Last Edit: July 16, 2017, 07:21:06 pm by Zanthius »
|
Logged
|
|
|
|
Death 999
Global Moderator
Enlightened
Offline
Gender: 
Posts: 3874
We did. You did. Yes we can. No.
|
EY is't exactly the author of LW. He was a founding member and wrote the original 'sequences' which form its core, but it's really basically a Reddit+Wiki for rationalists.
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
|
|
|
|
« Last Edit: July 19, 2017, 08:14:47 pm by Zanthius »
|
Logged
|
|
|
|
|
|
Death 999
Global Moderator
Enlightened
Offline
Gender:
Posts: 3874
We did. You did. Yes we can. No.
|
Did you mean to say something in response?
Also, science can be and indeed usually is done without Bayes rule.
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Also, science can be and indeed usually is done without Bayes rule.
Bayesianism is a much more accurate description of ideal Science, than Karl Popper's falsificationism. Popper’s central idea that scientific theories should make bold falsifiable predictions, seems to be nicely captured by Bayes’ theorem, since it gives more weight to a hypothesis the more unexpected an occurrence is without the hypothesis. Bayesianism is also not a binary approach to Science, like Karl Popper's falsificationism.
That most scientists don't even know about Bayes' theorem is scandalous. Scientists probably make many, many , stupid mistakes just because they don't understand Bayes' theorem. I am not interested in what can be and usually is done by scientists. Most scientists aren't exactly flawless. I am interested in what gives the best description of ideal Science, and Bayes' theorem certainly gives a much better description than Karl Popper's falsificationism, or any other theory of Science.
So we find that many phenomena in the cognitive sciences, plus the statistical methods used by scientists, plus the scientific method itself, are all turning out to be special cases of Bayes’ Theorem. Hence the Bayesian revolution.
http://www.soest.hawaii.edu/oceanography/researchers/francois/RESEARCH/RESEARCH_NOTES/SCIENTIFIC_NOTES/Popper-as-an-exception-to-Bayes.html
A number of scientific virtues are explained intuitively by Bayes' rule, including:- Falsifiability: A good scientist should say what they do not expect to see if a theory is true.
- Boldness: A good theory makes bold experimental predictions (that we wouldn't otherwise expect)
- Precision: A good theory makes precise experimental predictions (that turn out correct)
- Falsificationism: Acceptance of a scientific theory is always provisional; rejection of a scientific theory is pretty permanent.
- Experimentation: You find better theories by making observations, and then updating your beliefs.
In a Bayesian sense, we can see a hypothesis's falsifiability as a requirement for obtaining strong likelihood ratios in favor of the hypothesis, compared to, e.g., the alternative hypothesis "I don't know."
https://arbital.com/p/bayes_science_virtues/
|
|
|
|
« Last Edit: July 21, 2017, 01:18:39 am by Zanthius »
|
Logged
|
|
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
I have added a section about the fallacy of rosy retrospection to my page about cognitive biases:
I think this cognitive bias heavily influenced the last presidential election in the United States. It probably also heavily influenced the decision in the United Kingdom to leave EU.
|
|
|
|
« Last Edit: August 19, 2017, 04:32:27 pm by Zanthius »
|
Logged
|
|
|
|
Death 999
Global Moderator
Enlightened
Offline
Gender:
Posts: 3874
We did. You did. Yes we can. No.
|
I think the more important problem in the last US election was more along the lines of 'don't trust people who don't criticize themselves'.
|
|
|
|
|
Logged
|
|
|
|
|
|
Death 999
Global Moderator
Enlightened
Offline
Gender:
Posts: 3874
We did. You did. Yes we can. No.
|
You don't hear her criticize herself out loud, no, but there are visible symptoms of the process going on. Like, say, not running her mouth all the time.
|
|
|
|
|
Logged
|
|
|
|
|
|
|
|
|
Pages: [1] 2 3 ... 5
|
|
|
|
|
