Inference to the Best Explanation

1. Inference to the Best Explanation

You may recall that some deductive arguments provide explanations to help us understand why something happened, not convince us that something happened. These arguments still have premises and a conclusion, and follow the same logical rules, but it helps when reading arguments in natural language to know that they are not always trying to persuade us that the conclusion is true. There is also a common kind of inductive argument that is explanatory. Like all inductive arguments, explanations are not absolutely true or false. Just as other inductive arguments might predict the most probable outcome, an explanatory inductive argument tells us the most likely cause for something.

Explanatory reasoning overlaps with causal reasoning. After all, both are about things happening that cause other things to happen. However, explanatory reasoning is generally concerned with solving a single problem or understanding how and why something happened once; we aren’t assuming that the same conditions will cause the same event to happen again as we do with causal reasoning.

Imagine you find your car window broken and your phone missing. In trying to explain the event, you speculate that someone broke your window to steal your phone. This explanation explains all the relevant facts (broken window, missing phone) and furthermore seems likely. But there are other possibilities. Perhaps a stray baseball broke the window, and then a dog jumped through the window and nabbed the phone. This probably doesn’t feel as satisfying as an explanation, but why not?

Inference to the best explanation is a form of inductive argument where we consider a set of observed facts and come up with the most likely explanation. That somebody broke your window to steal your phone is an inference to the best explanation. Here is its form:

1. Observed facts: Your car window is broken and your phone is gone.
2. Explanation: The hypothesis that a thief broke the window and stole your phone provides a reasonable explanation of the observed facts.
3. Comparison: No other hypothesis provides as reasonable an explanation.
4. Conclusion: Therefore, a thief broke your car window and stole your phone.

Notice that this is an inductive argument because the premises could all be true and yet the conclusion is false. Just because something is reasonable doesn’t make it true. After all, sometimes things happen that defy reason. So, perhaps the baseball-dog hypothesis was actually true. In that case, the premises of the argument would still be true (after all, the thief hypothesis is still more reasonable than the baseball-dog hypothesis) and yet the conclusion would be false. But inference to the best explanation arguments are not intended to be deductive arguments, but inductive arguments, and as we learned earlier, inductive arguments don’t always lead to true conclusions. When they are wrong, however, it should be because something extraordinary happened.

However, what makes an explanation reasonable? What if your car was parked in front of a baseball field and next to a dog park? What if the door was left unlocked and the phone was too old to be very attractive to thieves? The notion of a “most reasonable explanation” is by nature subjective.

term to know

Inference to Best Explanation

A form of inductive reasoning where we work from a set of observed facts to come up with the most likely explanation for an event.

---

2. Conditions of Good Explanations

There are certain conditions that any good explanation must meet. The more of these conditions that are met, the better the explanation.

2a. Explanatoriness

To meet the condition of explanatoriness, explanations must explain all the observed facts.

EXAMPLE

The baseball explanation (by itself) does not explain the missing phone, so it is less attractive as an explanation.

term to know

Explanatoriness

Explanations must explain all the observed facts.

2b. Depth

To meet the condition of depth, explanations should not raise more questions than they answer.

EXAMPLE

If the car is parked in an underground parking garage, the baseball explanation would lead to additional questions like, who would be tossing a baseball around in an underground parking garage and why?

term to know

Depth

Explanations should not raise more questions than they answer.

2c. Explanatory Power

To meet the condition of explanatory power, explanations should apply in a range of similar contexts, not just the current situation in which the explanation is being offered.

EXAMPLE

If there were a rash of broken car windows and missing valuables, the best explanation would not be that many baseballs were flying around and many unleashed dogs were leaping through windows.

term to know

Explanatory Power

Explanations should apply in a range of similar contexts, not just the current situation in which the explanation is being offered.

2d. Falsifiability

To meet the condition of falsifiability, explanations should be falsifiable, meaning there must be a way to test the explanation for truth or falsehood. This does not mean that the theory can be disproven, just that it can be truly tested. In this case, perhaps a theory is that the baseball came to life in the extreme heat of the car and burst free of the windshield. However, imagine a further part of the theory is that such spontaneous genesis never occurs when the baseball is being observed. So, if we attempt to replicate the event (to test the theory), the test conditions themselves make it impossible. Non-falsifiable theories are not necessarily false, but they give us nothing to do as critical thinkers. In the real world, many popular beliefs (such as psychic powers or ghosts) are non-falsifiable, but good critical thinkers should be skeptical of any claim that cannot be tested.

big idea

Recall the example from the previous tutorials where we described rigorous testing to find the causes for our dog’s reactions. A key habit in critical thinking that applies to any form of inductive reasoning is being able to come up with a way to test your theory.

term to know

Falsifiability

The best explanations can be tested for truth or falsehood.

2e. Modesty

To meet the condition of modesty, explanations should not attempt to explain anything more than the observed facts.

EXAMPLE

It isn’t necessary to explain a rash of cat burglaries in the neighborhood where the phone was (ostensibly) taken, only the single incident of the broken window and missing phone.

term to know

Modesty

Explanations should not attempt to explain anything more than the observed facts.

2f. Simplicity

To meet the condition of simplicity, when all other things are equal, the simplest explanation is the best—the one with the fewest entities and events. This is often referred to as Occam’s Razor (in philosophy, a razor is a rule of thumb that allows a thinker to eliminate, or shave off, unlikely explanations for a phenomenon). For example, the baseball and dog explanation involves two entities (one baseball, one dog) and two events (the window breaking, the dog getting into the car), where the thief hypothesis requires only one entity and one event.

terms to know

Simplicity

The better explanation is the one with the fewest entities and events.

Occam’s Razor

All other things being equal, the simplest explanation is best. The simplest explanation is the one requiring the fewest entities and events.

2g. Conservativeness

To meet the condition of conservativeness, explanations that force us to give up fewer well-established beliefs are better than explanations that force us to give up more of them. For example, say the person with the stolen phone is visiting a friend, and somebody suggests the friend actually stole the phone. This requires the victim of the crime to question a valued friendship, and what he knows about the person.

It’s crucial to see how each of these are used not to come to the only possible true conclusion, but to the most likely one. For example, there are many cases, especially in science, where the right explanation is not so simple. Indeed, science must often posit new, sometimes purely theoretical, entities.

EXAMPLE

As early as the 1600s, some scientists speculated that there were “little animals” or “worms” that caused illness and spread from person to person, even across great distances. This idea introduced a new entity, a microorganism, that would not be directly observed for another couple of centuries and at the time could not be rigorously tested. However, these scientists were seeking ways to explain things that could not be adequately explained with simpler premises. Moreover, they actively pursued ways to find out more and test germ theory with the materials they had.

term to know

Conservativeness

The better explanation is the one that forces us to give up fewer established beliefs.