Header 1

Our future, our universe, and other weighty topics


Sunday, December 2, 2018

General Reasons for Doubting Most Scientific Theories

In this blog I have often stated specific reasons for doubting particular scientific theories. In this post I will write more generally, and discuss some reasons why most scientific theories should be held as doubtful, rather than ideas that are very probably true.

Reason #1: Most scientific theories are underdetermined, and a scientific theory usually is no better than rival theories called “empirical equivalents” that could explain the facts or make predictions equally well.

There are in academia quite a few highly intelligent thinkers who have widespread doubt about the reliability of scientific theories. Such thinkers are often philosophers of science. Philosophers of science have often said that most scientific theories are underdetermined. A theory is said to be underdetermined if there is not sufficient evidence to establish its truth. A scientific theory is said to be underdetermined when there are no observations allowing us to tell that such a theory is true rather than other rival theories that can explain observations equally well. Such theories are called “empirical equivalents.” You can find out more about this topic by doing a Google search for “underdetermination of scientific theory.” The article here gives a detailed discussion of the topic.

underdetermination

In many cases, we can readily think of alternate theories to explain the things that a scientific theory was created to explain. For example, we can easily imagine a rival to the theory that man appeared by evolving from lower species : a theory that humanity was specifically introduced to our planet with help from visiting extraterrestrials. In other cases, we may not be able to readily think of a rival theory to a scientific theory. But in such a case there are typically five or ten possible theories that could be imagined by someone willing to expend enough time thinking up alternatives. When we can't think of some rival theory that might be an empirical equivalent to some popular scientific theory, this usually just reflects the weakness of our imagination, not a lack of possible alternatives.

Reason #2: Most scientific theories are not top-notch theories and do not make precise numerical predictions that have been repeatedly verified.

There is a small group of scientific theories (which we may call top-notch theories) that make precise numerical predictions that have been repeatedly verified. Examples include the theory of general relativity, the theory of electromagnetism, and the theory of quantum mechanics. These theories rightfully enjoy high prestige. For example, it's rather impressive if a theory predicts that a particular comet will pass by Earth so close that it can be seen with the naked eye on January 13, 2038, and such a thing does happen on exactly that day.

However, it must be remembered that the vast majority of scientific theories do not make precise numerical predictions that have been repeatedly verified. Most scientific theories make no exact numerical predictions at all. Other scientific theories make exact numerical predictions that have never been verified.

Just as we should not let the accomplishments of some geniuses in New York City make us assume that most New Yorkers are geniuses, we should not let the predictive successive of a small number of scientific theories lead us to the impression that most scientific theories are precisely and successfully predictive. Most scientific theories are not any such thing.

Reason #3: Many scientific theories conflict with other respected scientific theories or conflict with one or more observed facts.

It is not at all true that scientists offer only one scientific theory to explain each type of phenomena. In many cases there are multiple scientific theories that offer rival explanations for the same phenomena. For example, both the Modified Newtonian Dynamics theory (MOND) and the dark matter theory offer rival explanations to account for stellar behavior. In some cases, two widely accepted theories may conflict with each other. For example, it if often said that the theory of general relativity and the theory of quantum mechanics conflict with each other. 

It is also true that a large fraction of scientific theories conflict with one or more observed facts. For example, Neo-Darwinism (which predicts very gradual evolution) conflicts with the fact of the Cambrian Explosion in which most of the animal phyla appeared rather suddenly in the fossil record. The theory that brains create minds conflicts with the fact that when epilepsy patients have a hemispherectomy operation (involving the removal of half of the brain to stop seizures), there is little effect on either intelligence or memory.

The proponents of a scientific theory will always bring to our attention any cases of observations that seem to support the theory, and avoid mentioning any observed fact that conflicts with the theory.

Reason #4: Once a scientific theory becomes popular, its drawbacks and failures are swept under the rug, and its successes are exaggerated.

Once a scientific theory becomes the most popular scientific account to explain something, that theory starts getting treated the way a Republican president gets treated on Fox News (or like a Democratic president gets treated on MSNBC). The failures and drawbacks of the theory are rarely discussed, but any apparent successes of the theory are trumpeted and exaggerated. An example is the Big Bang theory. We hear frequently about how the theory successfully explains the expansion of the universe and the cosmic background radiation. But we very rarely hear about how the Big Bang fails to predict a universe consisting of very little antimatter. In fact, the Big Bang theory predicts a universe with equal amounts of matter and antimatter, and we live in no such thing. 

After reading a balanced, objective account of the pros and cons of the most popular scientific theories, one that realistically discussed all the problems with such theories, your overall confidence in scientific theories would be substantially weakened.

Reason #5: It is easy to use “tricks of the trade” to get what may seem like some empirical support for a scientific theory, even if the theory is wrong.

A typical scientific theory will have some proponents who present this or that reason for accepting the theory. But we should not believe in the likely truth of some theory because some proponent has gathered a few facts to support it, or presented some experimental data that may seem to point in its favor. We should always remember that there are more than 15 different ways in which a scientist or a science writer may lead you to believe something is true even though the evidence for it is poor. Some of these ways are discussed in my long post “The Building Blocks of Bad Science Literature.”

Among the shady tactics used by over-eager scientists and science writers are the following:
  • Cherry-picking data to support a conclusion (discussing only items in support of the conclusion, and ignoring things that conflict with the conclusion).
  • Click-bait headlines that announce things different from what a scientific paper actually showed.
  • Scientific papers done using too-small sample sizes, so that there is a very high chance that the announced result is a false alarm.
  • Misleading visuals such as brain scan visuals that suggest some brain part is dramatically more involved during some activity, even though the actual variation is only 1% or less.
  • Dubious Monte Carlo simulations in which some computer experiment reveals a result that largely just reflects the assumptions of the person who programmed it.
  • Improper exclusion of alternative explanations, in which some researcher leads you to believe that some particular thing must be the cause of some experimental or observational result, failing to discuss alternate reasonable explanations that might plausibly explain the result in a different way.
  • Data dredging in which some experimental data is put through dubious statistical transformations until some trace of “statistical significance” is found.
  • Cases in which statisticians were requested to do dubious things by scientific researchers (surprisingly common according to one recent investigation).

A researcher armed with many thousands of dollars of grant money and a wide variety of such “tricks of the trade” will have little trouble dredging up some superficial evidence that seems to support his pet theory, no matter how far-fetched it may be. Similarly, if you give an astrologer $100,000 in research money, a wealth of data sets, and some good computer equipment and analytic software, there is a good chance he will be able to produce a few results that may suggest that there is some truth behind astrology.

Reason #6: Scientists grind out scientific theories almost as commonly as bloggers grind out blog posts.

Occasionally you will hear someone say that while it is easy for a scientist to present a hypothesis, it is very, very hard to produce a theory, because “a scientific theory is something that has been well-established by evidence.” This claim does not use a correct definition of the term “scientific theory.” Scientists use the word “theory” the same way as ordinary people do, to mean a hypothetical explanation of something. The claim that scientists only call something a theory when it is well-established by evidence is disproved by the fact that one of the main theories of physics is universally called “string theory,” and it is admitted even by string theorists that there is currently no empirical evidence for string theory.

A modern theoretical physicist may churn out five or ten theories a year. Since there are thousands of theoretical physicists, and thousands of biological and psychological theorists, you have a situation in which scientific theories are rather "a dime a dozen." 

No comments:

Post a Comment