A Third Form: Abductive Reasoning

· philosophy

The previous outline of deductive and inductive reasoning is correct, but incomplete; it leaves out a third, less mentioned, form of reasoning: abductive.

Despite not being well known by this name, it is a form employed very often and very naturally. Here's an example:

1. I observe that the grass is wet in the morning.

2. If it were to rain overnight, the grass would be wet.

3. Therefore, it is likely that it has rained overnight.

More abstractly, (1) is an observed effect, perhaps a surprising one; (2) is a hypothesized cause, that, were it to be the case, the observed effect would follow very naturally; and (3) is the conclusion that the cause is likely to have happened.

In a similar manner to the “inductive” form of reasoning discussed previously, the “abductive” form is also provisional and prone to error.

Say, for the sake of argument, that someone went around the neighborhood watering the grass with a hose. This is unlikely, but not impossible. The “rain” hypothesis is still the better explanation because it is more likely and, altogether, simpler than the alternative.

Maybe, then, we notice that the grass is wet every morning. Additionally, we do not observe other effects that would follow from rain being the case, such as water puddles on the street. We can then suspect that our original hypothesis was wrong, and maybe morning dew is actually the simpler explanation; and thus we revise our conclusion (3).

Abductive reasoning is a central form of reasoning for the sciences. Investigations very often start not from first principles, but from observations that seem surprising or unexpected at first sight.

When the current tools of knowledge don't satisfactorily explain something which is indisputably observed, a need is called for a more robust toolbox.

Another example is:

  1. I observe the Sun and the Moon rising and setting on the horizon about every day.
  2. If the Sun and the Moon were to rotate circling the Earth, that would happen.
  3. Therefore, the Sun and the Moon, more likely than not, move along circles around the Earth.

The Ptolemaic model was taught for hundreds of years, but as the body of astronomical observations grew, the observed orbit and behavior of nearby planets required a number of ad-hoc additions.

The simpler model that naturally agreed with observation was, after all, taking a whole perspective shift, as is well known today. The Copernican model (with Newtonian Gravity) followed:

  1. I observe the Sun and the Moon rising and setting on the horizon about every day. Likewise, I observe such and such orbit for Mercury, Venus, Mars, Jupiter, Uranus, ...
  2. If Newtonian Gravity were the case and the Sun -- instead of Earth -- were to be (approximately) the center of the Solar system, all of those observations would follow naturally.
  3. Therefore, Newtonian Gravity and the Copernican model are most likely correct. (Or, at the very least, are the most useful framework available for explaining what is being observed.)