Overcoming bias

[ciphergoth]

http://www.overcomingbias.com/

This site is totally fascinating, and like TV Tropes and Wikipedia, it has that hyperlink-means-staying-forever power.

ETA: I'm going to start adding some especially cool entries here as I find them:

• Belief in belief
• Making beliefs pay rent
• Hindsight bias
• The proper use of doubt

ETA: I can't help but notice that all the essays that make me go "eee!" are by Eliezer Yudkowsky, who also describes the Twelve Virtues of Rationality. I think I have a new hero.

ETA: I'll probably link back to here in another post once I've added a few more links.

• Outside the laboratory
• Human evil and muddled thinking
• How to seem (and be) deep

Comments

[juliet]

She makes the very good point early on that "angry ancestors have more kids" is not saying "being angry makes you want to reproduce", or anything of that sort. But then goes on to explicitly make the screwup mentioned above. And reports the "people having sex because of X Y & Z" story at the end without any kind of criticism or "this is a possibility" caveat. She does say that evolutionary psychology demands a great deal of separating of assorted not-necessarily-connected facts, but doesn't seem to actually be *applying* this.

Also, in that 12 virtues of rationality piece:

"It is especially important to eat math and science which impinges upon rationality: Evolutionary psychology, heuristics and biases, social psychology, probability theory, decision theory. But these cannot be the only fields you study. "

Now, even if psychology is a "science" at all (which I would at least contest), evolutionary psychology sure as hell isn't. And certainly not in the same league as heuristics/biases, probability theory, and decision theory.

Some of that - biases in particular - are a psychological issue, and *can* be studied in classic scientific manner, but you run into a lot of problems studying anything psychological in that fashion - hell of a lot of confounding variables, and the more you take out, the further you get from an actual-real-life situation. Observer/experimenter effect is a very real problem! (Not that this is always acknowledged, either.) But evolutionary psychology you really can't study like that *at all*.

If you're looking at psychological issues, there are a lot of problems with using quantitative scientific methods. Which isn't to say that they don't have their place; but one needs to be very cautious about drawing conclusions from that alone, and very aware of one's own biases and the biases implicit in the questions being asked. There's a *lot* of people doing research who aren't.

[ciphergoth.livejournal.com]

This is very interesting indeed - I must ask you more about this next time we have a chance to have a natter, which should be soon.

What do you think about the study of evolutionary psychology of other animals? One big problem with trying to do it with humans is that you can't tell what's innate and what's cultural, but animals don't have nearly so much in the way of culture.

[juliet]

I am more than happy to talk about it at length, possibly at too much length ;-)

Animals: well, there may still be learnt vs innate issues, especially with social animals. (And it should be noted that I work on the assumption that animals do have some form of mental life and awareness.) The general point about how behaviours may not be "deliberately" selected for also still stands. I've read less animal evolutionary psych but the "observe behaviour X, make up evolutionary rationale Y, state Y as fact" approach still seems to be prevalent.

And of course animal experimentation suffers from very much the same problems as human experimentation, in that the more variables you control, the less normal the environment is and therefore the less typical the behaviour seen. For example, there's been experiments showing that rats have more neural connections and heavier brains when kept in cages with more stuff to do and with other rats to interact with. That's a normal situation for non-lab rats, but an abnormal one for lab rats. So the lab rats that have been being experimented on for the last howeverlong have all, basically, had abnormal rat-brains (& probably lacked in social knowledge). What does that imply for their behaviour, learning, & so on? (Also, I get very angry on behalf of the rats, but I am trying to ignore that bit for the purposes of the argument...)