Monday, November 5, 2007

Induction and the nature of fallacies

After I had read about induction and deduction in the Skeptic's Dictionary (my own take here), I started to recognize a pattern among some of the logical fallacies. The reason that some logical fallacies are compelling to some people is that they are actually inductive arguments. Inductive arguments, though they do not guarantee the conclusions, can be quite compelling under certain circumstances. Therefore, it stands to reason that there might be some cases where a fallacy is not a fallacy at all.

For a first example, take ad hominem and the related argument from authority. Oh yes, it's true that if crackpot makes an argument, our judgement on the crackpot ultimately has no bearing on our judgement of the argument itself. However, if there are time-constraints that prevent you from scrutinizing every single claim, would you not think it at least a little more likely that the biologists are right, and the guy who can't even understand entropy is wrong?

As a second example, take the fallacy of confusing correlation and causation. Yes, large feet are correlated with better spelling. Pirates are negatively correlated with global warming. But if atmospheric carbon dioxide levels are correlated with human output of carbon dioxide, you certainly can make a case that there is a causative link.

As a final example, consider the argument from ignorance. Just because I've never seen evidence for psychics doesn't mean that they don't exist. But, you know, I think that the absence of evidence, even after much searching, can in itself be rather compelling.

The point of this isn't at all to toss out the beloved baloney detection kit. Many fallacies may be inductive arguments, but they're called fallacies because the induction is nearly always very weak or completely invalid. Even the unusually good examples I showed above aren't nearly as powerful as other arguments I could have made for the same things. But experienced skeptics know that there are times when fallacies appear to apply when in fact they do not. So inquiring minds must know, what are the limits?

The limits are the same as those for induction. Induction is most powerful when it eliminates or weakens all competing claims. For the example with carbon dioxide, any competing claim would have to explain the apparent correlation. Yes, it can be done, but not as easily as with the claim of causation. For the psychics, my argument is not so compelling at first, since it is possible that I won't see any evidence whether psychics actually exist or not. However, if psychics did exist, we would expect that some well-designed experiments would provide evidence. So when experiments are performed, but no evidence is found, it is reasonable to argue that no psychics exist.

And of course, induction only reaches its full potential when combined with lots of other good evidence. In the carbon dioxide example, the argument would be greatly improved by a robust theory of the sources and sinks of atmospheric carbon dioxide. In the example with the biologists, the argument I provided is not very good just by itself, but may be supported by the fact that the guy not only misunderstands entropy, but also misunderstands biology, standard reasoning, and honesty. But for that argument to become truly compelling, there is no substitute for actually looking at the claims and debunking them.

So anyway, my entire point is that it is important to have a nuanced understand of fallacies, and not just apply them blindly as if they were infallible.

1 comment:

B Fay said...

You might like looking at some of the issues regarding belief scales and environmental factors that aren't often addressed when looking at the paranormal. http://podblack.wordpress.com/2008/04/27/women-and-superstitions-part-two/