Six varieties of inductivism…

…and why they are all wrong headed. Inductivism is the theory that there is a process called induction that takes evidence and uses it to produce knowledge (useful or explanatory information). I have recently found out that inductivists like to say there are lots of different problems of induction and say Popper’s criticism of inductivism is relevant only to some of the flavours of inductivism. Here are six varieties of inductivism.

Popper argued, rightly, that there is no such thing as induction. The idea doesn’t make sense because evidence does not imply theories. Instead we propose conjectures about the way the world works. We propose conjectures about how those theories might fail to work and then come up with conjectures about experiments to test those theories and try those experiments out. The same goes for expectations, habits and all other kinds of knowledge. And, in case you hadn’t spotted this, all our evidence is conjectural too and sometimes a supposed observation is refuted.

Problem of induction 1 – How can we get theories from evidence?

Nobody gets theories from evidence. Evidence doesn’t imply that any particular theory is right. So whenever a person claims to have got a theory from evidence what he actually did was come up with a conjecture that explained the evidence.

Problem of induction 2 – How can we work out from evidence that X causes Y?

This is a special case of problem 1, so I won’t say anything more about it.

Problem of induction 3 – How does evidence confirm universal theories?

It doesn’t. Confirmation is sometimes said to be some sort of objective confirmation, i.e. – the theory is more likely to be true in some objective sense. That is, if we guess some theory is true in the light of evidence we are less likely to get it wrong. This is obviously nonsense. The theory is either right or wrong. That’s all there is to it. Most philosophers now say that theories increase subjective confirmation. It helps us to decide what to work on, or what theory we like best, or something like that. I think that evidence has very little to do with how scientists and philosophers decide what to work on. They pick whatever interests them, or whatever has implications that conform to their metaphysical or moral ideas, and sometimes use evidence to weed out some of their worse ideas. If you want to call conjectures that are sometimes controlled by evidence confirmation then go right ahead, but that’s going to confuse matters and you would be better to state clearly what you’re really doing.

Problem of induction 4, aka the Goodman problem, after Nelson Goodman the author of “Fact, Fiction and Forecast” in which Goodman, being a good inductivist, points out a load of problems with inductivism and then fails to spot their implications.

What’s the Goodman problem? Goodman states that some predicates are confirmed by observations and others are not. What’s a predicate? I hear your cry. A predicate is part of a sentence that modifies the subject of the sentence. So if I type “My computer is white.” the computer is the subject of that sentence and white is the predicate. So Goodman worries about the following problem. He states that when a scientist looks at an emerald he thinks that this confirms the theory that emeralds are green. However, his observations are consistent with the idea that emeralds are grue, that is green up to some time t and then blue thereafter.

When I look at the emerald I have some expectations. If I know a lot about emeralds the expectation might be a result of thinking that I understand the chemistry of emeralds. If not, then I might just think emeralds look pretty solid and solid things don’t usually change colour unless you do something to them, or something like that.

You might say “oh, but we’re talking about predicates now, not theories, or knowledge.” To this, I answer that if you put lipstick on a pig, it’s still a pig, and if you call a piece of knowledge a predicate, it’s still a piece of knowledge. The difference between words, including predicates, and explicitly stated theories is that words are shorthand for particular pieces of knowledge about things we see or use or whatever. So the word “green” is shorthand for a particular sensation, the word predicate is shorthand for a part of grammar, the word lipstick is shorthand for “waxy stick thing that people use to decorate their lips” and so on.

Problem of induction 5 – How do we work out from evidence that the future will resemble the past?

We don’t. Our best theories say that the world changes over time and so that the future will not resemble the past, except by virtue of obeying the same laws of nature. We try to work out those laws by conjecture and criticism.

Philosophers sometimes talk about another variety of induction. They ask what you can work out from the evidence plus a universal theory and they call this induction. For example, John Manchak a philosopher at the University of Washington, writes:

Cosmologists often use certain global properties to exclude “physically unreasonable” cosmological models from serious consideration. But, on what grounds should these properties be regarded as “physically unreasonable” if we cannot rule out, even with a robust type of inductive reasoning, the possibility of the properties obtaining in our own universe?

This is just using a theory, some evidence and logic to try to work out how much you can tell about what the world is like if you assume the theory is true. Now of course, anybody has the right to use any word he likes to describe what he is doing, but just working what a theory implies about the world given some evidence has very little to do with the problems listed above, or with the bad theories they presuppose.