On induction

by Neil Rickert

Induction, its use and its problems, is a thread that runs through philosophy (epistemology and, particularly, scientific epistemology).  I am an induction skeptic.  John Wilkins has a recent blog post on the topic: Phylogeny, induction, and the straight rule of homology.  I’ll comment on that post as a way of indicating where and why I disagree.

Phylogenetic classification is a form of induction. It enables us to infer the properties of an as-yet unobserved member of a clade with a very high degree of likelihood, as was pointed out by Gary Nelson in the 1970s.

I disagree already.  But let’s continue reading and looking for other examples of the problem.  A little further down, while discussing the “grue” problem, Wilkins says:

It must be noted that this is not a claim that emeralds will change color. It is about what we can infer of unseen members of a class.

This is implicitly creationist.  It presupposes that God created the classes, and it is up to us to discover what we can about the members of those classes.  The alternative is that classes are human constructs, and are part of our pragmatic enterprise to organize the world for our own use.  In the latter case, the projectible predicates are precisely those that we use to define our organization into classes (our classification scheme).

You might ask “Why should it matter whether the classes are created by God or defined by humans?  Don’t we still have the same problem of identifying projectible predicates for those classes?”

At first glance that might seem to be a reasonable question.  But it does not stand up to analysis.  If the classes are created by God, then they are immutable.  There is nothing that we can do about the classes, except discover projectible predicates for them.  If, however, the classes are pragmatic human constructs, then the classes are not immutable.  We can modify our classification scheme whenever new empirical evidence shows that it would be to our advantage to do so.  Induction, as usually described, is a truth seeking system for immutable classes.  If science is a pragmatic enterprise, then it can change its classification as needed, and that does not require induction.

And yet, biology is not deeply troubled by grue problems, even though it is precisely the science that should be. While the colour of the swan’s plumage turned out not to be projectible, the new black swan was not placed in a new order or class. It was recognised to be a swan nevertheless, and placed into the existing genus, hitherto a monotypic genus. Although philosophers, who anyway tended then as now to rely upon folk taxonomic categories for their examples, were shocked in the manner of Captain Renault, biologists simply shrugged, reported the new species, and added it to the existing taxonomy. The reason is quite obvious by now: the swan was not defined, but classified upon the overall affinities it exhibited, and the fact that one homolog differed in character state from the rest was not crucial, any more than if it had a different shaped beak from the rest of the genus.

And that should be the evidence that science (in this case, biology), does not actually depend on induction.  We see in what Wilkins said there, that biologists are not working with an immutable set of classes.  Rather, they are working with an extensible and refinable classification system, and they adjust that system as needed to accommodate new evidence.

If the universe were such that properties correlated by chance, it would not work, but in the cases of the special/paletiological sciences, properties correlate due to a shared productive cause.  If the universe lacked appreciable structure of this kind, then no search method would deliver knowledge (consider Wolpert’s and Macready’s “No Free Lunch” theorem).

And there we see the implicit creationism again.  There’s an assumption there about an immutable structuring of the universe.  But if the structuring comes from the way that we humans organize the universe, and if we organize it on a pragmatic basis that assures that it will work for us, the problem of “it would not work” simply does not arise.

It should not be thought we are supposing that natural classifications are in any way certain, or that any given homolog will exemplify the same states in each taxon or object classified.

To be clear, I am not objecting at all to the expression “natural classification.”  I am merely saying that we should take that expression as referring to a human designed classification system based on what is observed in nature, rather than as an immutable classification system handed to us by nature.

3 Comments to “On induction”

  1. Neil, I don’t understand the “creationist” comment here. Are you saying, by extension, that if a physicist says that there is a class of fundamental particle (say, quarks), that is creationism? He/she has merely described and named the structure to be found in the world. It is realist, perhaps, but not creationist in any sense.

    Of course our classifications are conceptual and constructed. Nobody denies this – whether classifications are natural or conventional, however, depends upon whether our classifications have been empirically forced (note that the post you link to is part of a book in progress in which we define what we mean by “natural classification” – it’s not what Peirce or Duhem meant). An empirically forced classification of the world is basically a summary of observation.

    The mere fact that scientists refine their categories is, I would have thought, an argument in favour of induction, not an argument against it. Induction is not stuck at some observational sentence level. One can be inductive about classes as well (and indeed ought to be). You have set up a strawman version of induction I fear, with your claim that it seeks immutable classes. Where does that come from? No literature on the topic I know.


    • I’m using “creationist” in the ID sense, in this case the implicit assumption that there is a design.

      For the physicists – probably some are implicitly creationist and some aren’t. But it doesn’t actually matter, as long as the physicist does his physics well. Similarly, it doesn’t actually matter whether there is a platonic mathematical reality, as long as the platonist mathematician does his mathematics well.

      It does matter in philosophy and epistemology. It’s an issue of how we connect language to reality. As long as the physicist adequately connects language to reality, we can do the science. But epistemology purports to be a theory about how we connect language to reality, and that is the principle issue.

      I should add that I am not intending my use of “creationist” to be an insult. But I do see epistemology, AI and cognitive science as on a wrong track.


    • I answered in a hurry, and missed some of your points.

      I am saying that classification is not forced, but is a pragmatic choice. Or, if you like, if it is forced, then that force is from history and tradition rather than from reality. If our only interest in biology were in seeing organisms as a food source, we might well have chosen very different ways of classifying.

      In some sense, classification is prior to observation. We cannot observe that “the cat is on the mat” before we have both a cat class and a mat class. In practice, we might start with a tentative classification, use that for observations, then revise the classification based on those early observations.


%d bloggers like this: