Induction and Science

Sure. And we would investigate that, in an attempt to find a cause. We would not just assert a generalization as truth. Of course this depends on what one means by "regularity." The word, based on its root, ought to mean "rule following". I am saying that nature is not obliged to follow any rules.The universe is not uniformly homogeneous stuff. It is lumpy. If the world were uniformly homogeneous, that would be strong evidence of regularity. It's lumpiness is evidence of irregularity.Given a lumpy world, it is up to us to make some kind of sense of that world. So we invent rules. But they are not rules that the world follows. They are rules that we follow in making sense of the world. Those rules, human inventions, are what we call "scientific laws".Take Newton's famous laws of motion. Prior to the investigations of Galileo and Newton there was no concept of mass. What we currently call "mass" and what we currently call "weight" were at one time merged into a single broad concept. Similarly, the ancient folk concept of force was different from Newton's concept of force. Newton introduced the new concept of mass, changed the concept of force, used his calculus to precisely define acceleration. Newton's laws were very much about defining and using these concepts. The laws and concepts together were an invention, a new way of slicing up the world so as to make better predictions. The rules are rules for scientists to follow, not rules that nature is required to follow. "F=ma" was pretty much the definition of how to measure mass, and necessarily true by virtue of being a definition. An interesting example. We use these studies to identify elements present on stars. To the best of my knowledge, we have not visited a single star to verify this. If this use is an induction, then it is a generalization based on zero observations. If our laws arose by induction, then we could not trust the apparatus. It is because our laws are part of our inventions, that we know we can trust them.The traditional view is that we start with minute facts, and use induction to form more general laws. I am saying that, for the most part, we start with general laws as accounts of empirical methodology, and we use those to slice up the world into smaller piece so as to allow the collection of the minute facts.

According to textbook induction, one starts with facts and jumps to generalizations as conclusions.Textbook induction ignores the use of of experimentation for critical tests of theories.Textbook induction presupposes that all concepts exist a priori, and one gets facts by simple observation. It ignores the fact that scientific investigation leads to new concepts, and many of the laws have to do with the newly invented concepts.Textbook induction presupposes that observation is simple. It ignores the difficulty of measurement. It ignores the extent to which new concepts and new laws come from inventing new measuring methods.

That's a pretty good description of what is wrong with textbook induction.

I am arguing that it is often the other way around. That is, we start with general principles, and use those to enable us to find particular facts. The general principle known as Ohm's law (V=IR) is built into most standard electrical measuring equipment to make it possible for us to measure electrical quantities. Ohm's law itself is pretty much true by definition, not assumed true as a result of inductive inference.

There is a big difference in what is taken to be already known and what is taken to be unknown.

"Validity" implies that induction leads to truth. That's what I am questioning.Observed patterns lead to interesting ideas that can be further investigated. Without additional investigation, one cannot take an observed pattern as truth (as something that will always occur).All Boeing 737 passenger aircraft have arrived safely at their destination. By induction, all such aircraft will arrive safely. Oops, one of them crashed, so the induction has failed. Send in the NTSB (National Transportation Safety Board) to investigate.We actually learn a lot from work by groups such as the NTSB. We make scientific advances by investigating apparent induction failures. In this case nobody would have really believed the 737 induction anyway, because there were known risks. We can have generalized statements which we know to be false, even if a good approximation. We improve our science by the extra specialization that comes from investigating the failure of these generalized statements.

We invent our own axiom systems that fit reasonably well, and call them "scientific laws". They don't have to be axiomatic principles of the universe. They only have to fit, in the sense that we can use our axioms (scientific laws) to do useful things.

No, but there is a trial-and-error component.Suppose you want to repair Notre Dame cathedral. If you can find the original architect's blueprints, you have your axioms. Failing that, you would usually build a scaffolding around the building, and work from there. The structure of the scaffolding is a kind of axiom system that approximates the cathedral.I'm suggesting that scientific laws are the scaffolding we build to approximate the world.Scientific change is much like evolution. Scientific theories change, usually by small increments. There is a trial and error phase, where new ideas are tested -- that's a bit like mutation. At some stages, we get the sudden change, similar to punctuated equilibrium. That's when a bunch of previously tested alternative ideas are put together in a coherent package to replace the old scaffolding with a better fitting one.

Are you suggesting that there are problems when the mechanic tests the tuning of your car? Or are you saying that testing the tuning of your car amounts to induction?If that's your point, then okay. But then what you mean by "induction" is very different from what is described in the literature.

Great. We finally have something we can agree upon. Another agreement. That's two agreements in one message. We are really making progress. Oops! You just broke the chain of agreement.You are assuming that a scientific law is a generalization. That's precisely where our disagreement lies.Which specific observations did Newton's law of gravity generalize? As far as I can tell, the first specific observation, of which it could possibly be said to be a generalization, was the measurement made by Cavendish in 1798. That's 111 years after Newton proposed his law of gravity.Which specific observations were generalized by Ohm's law, V=IR? To observe a specific instance you would have to already know the resistance R, but there was no way of determining R without presupposing V=IR.Which specific observations were generalized by Darwin, in his "Origin of the Species"? As the creationists are fond of reminding us, Darwin had observed zero instances of an origin of a species. It was intended to be problematic for the view you expressed in Message 56, that testing shows induction is being used. It is an example of testing a mechanism. I'm arguing that scientific laws are parts of mechanisms, or proposed mechanisms, and are not generalizations. Again, not a problem for me. I am not suggesting that the mechanic is using induction. I was using the example to question your apparent view that testing of a mechanism is an instance of induction.

That a statement is general does not make it a generalization.There can be general statements that are not generalizations. In fact, I just made such a general statement.That a statement is a generalization does not make it an inductive generalization. Mathematicians are often generalizing, but they are never making inductive generalizations. You may have misunderstood my comment about mechanisms. I do not suggest that scientific laws are mechanisms of reality. There might not even be such mechanisms. I am saying that scientific laws are empirical mechanisms, part of the empirical machinery we use to effectively manage our interactions with reality.

Darwin's "Origin of the Species" attempted to explain his observations at Galapagos. It was not a generalization of those observations. He did not observe any species originate.

He certainly thought he was accounting for the succession of species. He wasn't explaining origins in the sense of abiogenesis. As far as I know, the specific instances of selection he observed were of artificial selection, and were not on Galapagos. But his theory was based on far more evidence than that. It just was not an inductive generalization of the particular observations.

He doesn't talk about speciation, as I recall. He probably didn't think it needed discussion. But he does discuss the succession of fossils in the geologic record, so it seems he assumed he was accounting for the succession of species. If it is an inductive generalization, then the specific evidence ought to be particular cases of the general statement, and not just something that can be inferred from the general statement.

Based on this thread, there seem to be several principles used to argue for induction.

1. Make bare assertions (many messages in this thread)
2. When challenged for evidence (Message 60), ignore the challenge and continue to make bare assertions (Message 62).
3. Assert, wrongly, that your opponent has conceded (Message 72).