Wednesday, July 11, 2007

Extreme and special cases and causes

Static vs. dynamic causes of phenomenons
Latent possible effects may always be there
Inference form extreme/artificial cases is problematic

Looking for relationships between a resulting phenomenon and various causes is very common nowadays. Part of the following is very confusing. So, turn off your logical powers, you will need them here.

Static = the "default" case. What causes things in the "usual" case: No special treatment was applied AND not for a very special group/case you have selected.

Dynamic = special cases. Extreme behaviors of one of the parameters. duefully applied intervention. Some effect was introduced that is never there in the "normal" case.

Potential = a latent effect that can be used to cause a huge (or small) effect. But when not applied it has zero effect.

An example for a latent "potential": Heart disease has various causes and surgeon knife is not one of them. But miracousely, a bypass operation can have near 100% effect on the heart state once applied. Hence, bypass surgery is a "potential" cause, that is totally absent from the causes of the static case, and strictly there in the dynamic/interevention case.

A big difference exists between what causes things in the static case, and the dynamic case.things that have zero or negligible effect in the static case may have a main effect when done on purpose to have an effect. This (1) gives hope to find - for example - effective intervention to improve happiness, but it also implies that (2) the inference form extreme cases to the average is limited. I will start with (2)

(2) The inference problem.
You cannot always infer from extreme cases to the normal case. Extreme cases may practically be different things.

Example 1. Are reshus monkeys violent (relatively) because of their genes, or because of culture? Frans de Waals has two pieces of evidence that culture plays a role. young reshus monkeys that grow up for six months mixed up with other non-violent monkeys, got used to live much less violent. Similarly, a group whose big aggresive males died, started to live peacefully for a decade, even after all males in the group were totally new (males change troops as adolescenes). This can be explained in two ways. The favorite way is that culture palys a big role in the violent behavior. I feel the more correct definition is that extreme cultural effects may change the behjavior, but I would guess that it is mainly genetic. It is only in very special cases

Example 2. The great psychologist Roy Baumeister in his book "meanings of life", says that people need their life to have value. One of the evidence is mothers that killed their babies. They show strong signs of distress for the anti-value acts they did etc.
Does it show that peolle need value in life? not at all. It only shows that extremely unjustified action carry a strong negative feeling. But it is quite possible that in the average case value does have to be there.
("need for value" may have other evidence. It is hardly the point here)

This problem is rampant with experimental and other kinds of scientific inquiry. Elegant and logicaly strong conclusions are many times exactly the kind I am condemning here. An extreme case can show minute details on how things actually work. You can iron out alternative hypotheses, and be clear of various biases and protocol problems etc.

That we have reason to choose a line of inquiry, does not eliminate its build-in problems. Truth does not work in our service. Reality never apologizes for being sonfusing and hard to decipher. Saying "that is the way to know" when you do not really know is stupid and pathetic.

(1) There may always be strong unknown potential cause.

The analisys of the static/default case does not tell us what potential causes are not being applied.

Example: Happiness research may give a gloomy feeling that we cannot improve happiness much, as half is genetic, 40% unknows, and the rest 10% are correlates we have little control of.
This is the static picture. Are there effective interventions? maybe. these shold be things not seen in the static case, but very powerful once discovered and applied in practice.

To sum up:
Static vs. dynamic causes. in default cases have one set of causes, while intervention may actually change part of the very structure of waht causes things.
Inference form extreme cases is problematic, as extreme/artificial cases may be very different in waht causes what.
Looking at static causes alone, one cannot conclude that there are no potential other causes.

No comments: