5.4

The University of Auckland

Skip to 0 minutes and 14 secondsRemember the distinction between non-deductive and deductive arguments. Patrick gave it to you back in week three. A deductive argument is an argument for which the premises are offered to provide a logically conclusive support for its conclusions. A non-deductive argument is an argument for which the premises are offered to provide probable but not conclusive support for its conclusions. And remember, now back to the second week when Patrick talked about the difference between arguments and non-arguments, and made the point that typically explanations are not arguments. An explanation is a statement or collection of statements asserting why or how something is the case. Explanations can, however, be used in arguments.

Skip to 1 minute and 4 secondsAnd one of the most common forms of non-deductive arguments that use explanations are inferences to the best explanation. And that's what we're going to talk about in this clip. If you've read a detective story, you've probably come across an inference to the best explanation. Suppose I go to the pantry looking for an oatmeal cookie, what some of us call a biscuit. You can probably see I'm a man who likes the occasional cookie. I open the tin and it's empty. I observe the absence of cookies, oatmeal or otherwise. I might want an explanation for that observation. You can bet I'll want an explanation. Suppose four people live in my house, me, my wife, and two of our children.

Skip to 1 minute and 48 secondsI'm confident that I didn't eat the last of the cookies. Justine has been at work since early this morning, and I saw there were cookies in the jar after she left. Kelly comes out in spectacular hives if she eats oatmeal cookies, and she looks unblemished. Only Raf remains. It must've been him. We offer an inference to the best explanation. When we have an observation that seems to need explaining, the absence of cookies in this case, and we conclude that the best available explanation of the observation is probably true because it's the best available explanation of the observation. A little more formally, S is a state of affairs, a collection of data, facts, observations, givens. H would, if true, explain S.

Skip to 2 minutes and 36 secondsNo other hypothesis can explain S as well as H does. Therefore probably H is true. Inferences to the best explanation, these arguments which include explanation, are common in scientific reasoning. We prefer the asteroid theory because it explains the observations better than current alternatives. And here's another example from science. Smallpox was rife in 18th century Europe claiming around 400,000 lives a year. And it was pretty indiscriminate. People got it regardless of their backgrounds. It was widely noticed though that milkmaids tended not to get it. That's our S. Smallpox is rife, but milkmaids don't get it. We can imagine a range of possible explanations. Maybe milkmaids drank more milk than most people, and milk contained something that protects people from smallpox.

Skip to 3 minutes and 38 secondsMilkmaids worked in the countryside, and perhaps smallpox was an urban disease. Maybe milkmaids often caught another relatively innocuous disease, cowpox. And perhaps if you'd had cowpox, you didn't get smallpox. And now we might go through these potential explanations in much the way we did when we were looking for the cookie eater. Do other milk drinkers have immunity from smallpox? Did other people who lived in the countryside get smallpox? What happens if you give someone other than a rural-dwelling, milk-drinking dairymaid cowpox? Do they gain immunity to smallpox? It turned out, of course, that the cowpox explanation was best. We owe modern vaccination to that discovery. We can put that in our inference to the best explanation schema like this.

Skip to 4 minutes and 35 secondsS, smallpox is rife but milkmaids don't get it. H, cowpox gives immunity to smallpox. No other hypothesis can explain S as well as H does, therefore probably H is true. By virtue of their form, inferences to the best explanation are always strong. And that's all we require for non-deductive arguments to succeed logically. An inference to the best explanation is a strong argument because if it is indeed the case that the offered explanation is the best available, then we are justified in believing the conclusion. We are justified in believing it because the explanation is the best hypothesis available.

Inference to the best explanation

Inferences to the best explanation are common in scientific reasoning.

We offer an inference to the best explanation when we conclude that the best available explanation of an observation or state of affairs is probably true because it’s the best available explanation of the observation. A little more formally:

• S is a state of affairs, a collection of data, facts, observations, givens.
• A hypothesis, H, would, if true, explain S.
• No other hypothesis can explain S as well as H does.
• Therefore, probably,
• H is true.