There are domains where it’s easy to perform experiments (physics, chemistry) and others where it’s unfeasible (biology, economy) or impossible (psychology). The quality of scientific understanding in these different domains is necessarily different. Has there been any thoughts or study devoted to the subject of doing statistics or Bayesian learning where you can suffer from lack of feedback or hysteresis? Is there a mathematics for doing science in low feedback domains?
Surely you can do some tests or experiments, but the staple of physics-like experiments, having identically prepared systems, is partially or totally lacking.
It’s only unfortunate that the poorly qualified first paragraph has totally hidden the real question, which was at the end of the second paragraph. I guess I’ll have to wait for the next open thread...
Well you can’t do things that would be really nice to do in biology like “rerun the tape of life”, not to mention the tremendously interdependent system that is any living organism. And the artificial laboratory conditions of psychological experiments, along with variability of subjects, form a huge impediment to study.
It’s not that you can’t do experiments, but it’s much more difficult to isolate parts of biological, economic and psychological systems and experiment usefully on them.
Sure, but I think it’s reasonable to say that humans are ill-behaved as experimental subjects compared to other biological organisms, which are ill-behaved compared to particles, stars and galaxies.
I mean ill-behaved in the sense that their behaviors cannot be reliably modeled by compact mathematical models.
The whole point of Bayesianism is that you get as much information as possible from a small amount of data. It works perfectly well in noisy domains.
The recent post on Knightian Uncertainty may or may not be relevant to your interests—it’s not the same thing but it seems like it might be related to what you were really getting at.
There are domains where it’s easy to perform experiments (physics, chemistry) and others where it’s unfeasible (biology, economy) or impossible (psychology).
The quality of scientific understanding in these different domains is necessarily different. Has there been any thoughts or study devoted to the subject of doing statistics or Bayesian learning where you can suffer from lack of feedback or hysteresis? Is there a mathematics for doing science in low feedback domains?
In what universe can’t you do experiments about biology or psychology?
Guess I should’ve stayed home today.
Surely you can do some tests or experiments, but the staple of physics-like experiments, having identically prepared systems, is partially or totally lacking.
It’s only unfortunate that the poorly qualified first paragraph has totally hidden the real question, which was at the end of the second paragraph. I guess I’ll have to wait for the next open thread...
I just realized that MrMind probably meant evolutionary biology and evolutionary psychology, which makes waayy more sense.
Well you can’t do things that would be really nice to do in biology like “rerun the tape of life”, not to mention the tremendously interdependent system that is any living organism. And the artificial laboratory conditions of psychological experiments, along with variability of subjects, form a huge impediment to study.
It’s not that you can’t do experiments, but it’s much more difficult to isolate parts of biological, economic and psychological systems and experiment usefully on them.
You can’t do things in cosmology like “rerun the tape of the universe” either.
Sure, but I think it’s reasonable to say that humans are ill-behaved as experimental subjects compared to other biological organisms, which are ill-behaved compared to particles, stars and galaxies.
I mean ill-behaved in the sense that their behaviors cannot be reliably modeled by compact mathematical models.
The whole point of Bayesianism is that you get as much information as possible from a small amount of data. It works perfectly well in noisy domains.
The recent post on Knightian Uncertainty may or may not be relevant to your interests—it’s not the same thing but it seems like it might be related to what you were really getting at.