Inductionism

What a nuisance it can be, trying to refute a theory that seems to have as little evidence against as it does for. At times like this, one feels tempted or even justified in asking for some positive evidence.

However, to do so would be to delve into the black art of induction; not always an invalid option, but one where it pays to take a little care.

Remembering all that "If A then B" stuff, induction is the process by which such rules are discovered in the first place. (Deduction is when you go on to say something like "not B; therefore not A". (Forget Sherlock Holmes - he lied.)) Though essential, induction is a very private and personal function, as well as being apallingly inaccurate much of the time, though statistical investigations often help with accuracy and consensus. In this way, the study of characteristic claw curvatures of tree and ground dwelling birds allowed us to induce more precise rules such as "If claw curvature such and such, then high likelihood of terrestriality".

But the vast majority of the laws we use when thinking haven't been based on a formal statistical study, just our memory of what tended to happen in the past. They may also involve situations sparse in evidence, for example rules where certain patterns make us suspect ghost lineages exist. They may also be tempered by other rules such as "ghost lineages are more likely amongst small, rare when present, forest living animals than with large common when present marine forms". When people come up with notions apparently lacking in direct positive evidence, they are often using a combination of rules, most of which will be probabilistic, and due to their own particular experiences, will be unique. It may well be worth identifying these rules and giving them a formal statistical investigation. It may be very hard for the individual to identify what rules he used, but the process of extracting them is now an everyday occurrence in industry.

They constitute positive evidence, they may be accessible to others, and even when they aren't, they may still be terribly useful insights. Surely a good scientist should try to make use of all knowledge, no matter how probabilistic? In other words, it's often wrong to say there is no positive evidence, but even when there seems to be very little positive evidence, that's no reason to say the theory's wrong.

(I hope this goes some way towards addressing Ross MacPhee's posting.)




Back to main web page