You are on page 1of 1

Inference to the Best Explanation

A Historical Example

1) We have observed n orbits of Uranus, and each time its orbit is elongated.
2) Therefore, it is probably always elongated.

The inference from 1 to 2 is a justified by induction. But what explains why is elongated.
For that we need abduction.

3) If there an 8th planet of such-and such mass with so-and so position, then the orbit
of Uranus would always be elongated.
4) Therefore, there is an 8th planet.

This argument offers an explanation of why the orbit is elongated. That means it offers a
hypothesis. Hypotheses cannot be deduced from observations, but observations can be
deduced from hypotheses. So we must think creatively to come up with a hypothsis and
then test it against observations. We test it by using the surprise principle.

The Surprise Principle

Start with observation O1. Now come up with two competing hypotheses that both
explain O1. Now you need to make further observations to decide which theory is the
best explanation.

An observation O2 favors H1 over H2 just in case that


-If H1 were true , you would expect O2, and
-If Hw were true, you would not expect O2.

The surprise principle tells us one hypothesis is better than another. But it does not follow
from that alone that the hypothesis is true. There could be some other hypothesis not yet
thought of that is superior to it. In principle, inference to the best explanation is always a
comparative enterprise and we need to keep making observations to see if our hypothesis
is confirmed or surprised. However, in many contexts it is easy to settle on what is the
best explanation.

You might also like