08 October 2005

Untestable Hypothesis and Falsifiability

The Scientific Method has some five key steps to it, as mentioned previously, and reminded here.

  1. Observation/Question

  2. Hypothesis

  3. Prediction

  4. Experiment

  5. Conclusion

One of the toughest parts of the Scientific Method is simply determining whether it's possible to design an experiment to test your hypothesis. If it IS possible to test it, and there are clear conditions for what counts as refuting your hypothesis, the hypothesis is called falsifiable, and this is a good thing in science. "Falsifiable" means the same thing as "testable," it doesn't mean "proven false." Yeah, it's confusing. So some examples may help.

Hypothesis: There are NO black swans.
Test: Look for a black swan.
Falsifiable? Yes
Potential Falsification: Finding a black swan.
Truth Status: False (there are black swans).

Hypothesis: There are pink elephants.
Test: Look for a pink elephant.
Falsifiable? No
Potential Falsification: None. If you looked around the whole world, maybe it was hiding in Japan while you were in New Zealand. If you saw the whole world simultaneously, maybe it's on Mars. Or another solar system. It's impossible to actually carry out the test.
Truth Status: So far it appears false, but we're not sure. The statment "there is no such thing as a pink elephant" is a good example of something that a non-scientist would call a fact but scientists would say is unproven.

Hypothesis: Throwing a virgin in an active volcano appeases the gods and prevents the volcano from erupting.
Test: Don't throw a virgin in volcanos.
Falsifiable? Partially.
Potential Falsification: The problem is that if you do throw in the virgins and it doesn't erupt, you're not sure if it's that or something else which prevented the volcano from erupting. If you don't throw in virgins and it does erupt, you're not sure if it's that or something else which did make it erupt. The only condition which definitely would falsify it would be if we threw in a virgin and the volcano erupted anyway. Another way to think about it is that we can't test supernatural forces.
Truth Status: Essentially false, as we've got other explanations for volcano eruption that do not evoke supernatural forces.

Hypothesis: Waging war in Iraq keeps America safe.
Test: Don't wage war in Iraq and see if we get "less safe."
Falsifiable? Perhaps, but are we willing to try the test? And how would we quantify it?

Responding to the president's [Oct 6 (?), 2005] address , the Senate Minority Whip, Dick Durbin, D- Illinois, said the speech left too many questions about the Iraq war unanswered.

"He owes it to the American people -- and the Democrats are calling on him to tell the American people -- how will this end? How can we measure success? How can we get beyond the generalities of the speech that we heard today?" Durbin said.



doris said...

Ah, but maybe the virgin wasn't really a virgin?

I love your examples .... I hope you use them in your classes?

kaka said...
This comment has been removed by a blog administrator.
Holly said...

This post was very helpful!

Mark said...

Ok, now some people would say your first example about pink elephants is wrong, because unfalsifiabily means even in theory (not just practice) there is absolutely no way to falsify the proposition. In theory we could look just everywhere for pink elephants. A true unfalsifiable statement would be like say "Some humans live forever," because it is not imaginable to keep watching someone live forever. No matter how long the person is observed, he could still die.

Anonymous said...


How is it not possible to keep watching someone live forever (to look for an infinite amount of time), but it is possible to look everywhere at once (infinite perceptual field)? Both propositions involve mastering the infinite. I think you missed the point about elephants being able to move. We cannot, in theory, see everywhere at once. You also seem to be assuming that the universe has finite dimensions... where is your evidence for that?