I. Popper’s falsificationism

Recall Popper’s falsificationism: A hypothesis is scientific if and only if it has the potential to be refuted by some possible observation.

This sounds like a promising approach to defining science and Popper has been a very popular and inspiring figure amongst actual scientists. However, not everyone fully appreciates how radical Popper’s position is.

According to Popper’s picture, a theory is scientific if it is falsifiable. Here is the process:

1.   Take any theory and deduce an observational prediction from it.

2.    Check to see if the prediction comes out as the theory predicts.

3.   If it doesn’t, the theory is refuted.

4.   If it does, Popper thinks this doesn’t tell us anything. The theory is not confirmed. It is not even more likely than before. The most success we can hope for according to Popper’s falsificationism is that a theory has not been falsified (yet)!

According to Popper, it is never possible to confirm a theory by showing its agreement with observations. Popper thinks that the truth of a scientific theory can never be supported by observational evidence, even if the theory     makes a huge number of predictions that come out as expected.

What does it mean to confirm a theory? Roughly, when people say that a theory is confirmed, they mean that there is enough evidence to support it. This evidence is usually partial rather than decisive. A confirmed theory can turn   out to be false. Popper denies that a scientific theory can be confirmed. This is a radical position!

II. Popper and induction

Part of Popper’s motivation is his skepticism about inductive reasoning. Inductive reasoning is used a lot in science, but some philosophers have worried that this kind of reasoning involves unanswered questions.

The following is an example of an inductive argument:

P1. Swan 1 observed at time t1 was white P2. Swan 2 observed at time t2 was white

…

C. Therefore, all swans are white

We can rephrase this as:

P1. All the swans I have observed so far have been white C. Therefore, all swans are white

The basic problem of induction is to answer these questions:

-     What reason do we have for thinking that patterns observed in past experience will hold in the future?

-    What justification do we have for using past observations as a basis for generalizations about things we have not yet observed?

According to Popper, worries about induction are entirely warranted. Popper thinks that inductive reasoning can’t be justified. He rejects the notion that observation of particular cases (or successful predictions) is in any way relevant to confirming the truth of general scientific theories.

Popper claims that while a successful prediction is irrelevant to confirming a law, a failed prediction can immediately falsify it. Popper’s “solution” to the problem of induction is to reject induction. Since science, according to Popper,  only ever seeks tofalsify theories, and never seeks to verify or confirm them, science need only utilize deduction. It  need not use induction at all.

III. Theory choice and corroboration

If theories can’t be confirmed, how does Popper think we should choose between competing scientific theories? Popper doesn’t think a scientific theory can be confirmed, but he does think a scientific theory can be corroborated.  According to Popper, a theory is corroborated if it has survived many tests to falsify it.

Let’s consider an example. Suppose we want to build a bridge. We need to use physical theories to tell us which designs are stable and will withstand the weight we require it to.  How do we choose which theories to rely on  according to Popper?

If we have a choice between competing theories, one of which has been falsified, then our choice is easy. We should choose the theory that hasn’t been falsified. If we have to choose between two theories that haven’t been falsified,   the situation is more difficult.

Suppose we have two competing theories A and B. Theory A has been tested and not falsified many times. Theory B is a new theory that has not yet been tested. It might seem sensible to use theory A since it has been tested successfully many times. However, Popper refuses to say that when a theory passes many tests that we have more  reason to believe that the theory is true. According to Popper, theory A and theory B are both just conjectures that have not yet been falsified. Neither theory is more confirmed than the other.

Popper’s answer to this problem is to use the notion of corroboration. He thinks that we should choose the theory that is corroborated over the one that isn’t. But here we have a problem:

1.   Either “corroboration” really just means “confirmation”

2.   Or it means something completely different and gives us no information about the truth of the theory.

In the first case, Popper’s theory is the same as previous theories concerning confirmation. In the second case, it’s   hard to see why we should take the corroboration of theories as a recommendation to use them if it doesn’t provide any reason to think that the theory is true.

In Theory and Reality, Godfrey-Smith provides a helpful analogy in understanding the difference between corroboration and confirmation. He says that it is like the difference between an academic transcript and a letter of recommendation. An academic transcript is just a record of what classes you have taken in the past and the grades you have received. It measures past performances, but doesn’t explicitly predict how you will do in the future. A    letter of recommendation, in contrast, makes claims about what you have done in the past and makes predictions   about how you are likely to do in the future. The corroboration of a theory tells us about past successes of the theory, whereas the confirmation of a theory (if it were possible) would be a kind of recommendation for using the theory in the future.

According to Popper, corroboration provides no evidence at all that the theory in question is true. It does not even provide evidence that the theory is preferable to an untested (and unfalsified) competing theory. Instead, corroboration shows merely that a theory is “scientific.”

IV. Salmon’s criticism

Salmon criticizes Popper’s account of corroboration and its role in theory choice. Popper claims that it is rational to  choose well-corroborated theories, though he rejects that we have inductive grounds for thinking that these theories are likely to be true.

Salmon argues that one of the main goals of a scientific theory is to make predictions. We want to make predictions to satisfy our curiosity, to test a theory and to help us find the best (or at least a reliable) way to solve some practical problem, such as building a bridge that doesn’t fall down.