Last week in considering the realist and conventionalist arguments about the nature of classification, I ended up putting a weak – or perhaps even silly! – argument in the conventionalists mouth. Namely, I had the conventionalist claim that the following is an ampliative inference: “upon encountering a Dewey decimal for a previously unknown (at least for our investigator) book, if its code starts with an 8, one will be able to infer that it is a work of literature, and if with 88 a work of Greek literature and so on.” Here I wish to bolster the conventionalist’s case, and make less controversial the analogy between biological classification and library classification. A thought experiment is, alas, to be had.
Consider the following scenario. Humankind has faced its extinction, but before said extinction it was decided that all books be classified and marked using the Dewey decimal system. A race of alien beings happen upon Earth to find the vestiges of a lost civilisation. Among the artifacts they discover weird sheets of pulped wood bound together bearing differing symbols. They come to see these objects as a means of sharing information in a textual form. But one thing baffles them: why do all of the books have this strange string of numbers in their cover page? To their great misfortune, the aliens never discover a book detailing the Dewey system, so they are forced to start building theories about the strings of numbers. So theories they construct. They make bold ampliative leaps: books with strings of numbers beginning with ‘8’ seem to contain fiction, so they surmise that any further works with an ‘8’ in the beginning of their string will be works of literature, and so on. Sometimes such inferences will be robust, but other times they will fail, and the alien theorists will be forced to revise their system. This, the conventionalist adamantly claims, is ampliative reasoning at its finest. The aliens’ theory of classification will allow for (fallible) inductive generalisations!
So, given the above, the realist and the conventionalist will continue to defend their intuitions and to pump their interlocutors’ intuitions with clever arguments and intuition pumps. And maybe what is needed is not a resolution for this dispute, but rather adoption of new, better intuitions.
Finally, why care about our intuitions when it comes to classification – or anything else for that matter? Science is built on highly regimented everyday thought and observation, and as such it still carries residues of our intuitions and fallacies. Further, in order to make sense of what our best science tells us, we need a way to – partially, fallibly – translate scientific claims into ordinary, everyday language. This will require satisfying at least some of our intuitions – for now, until we are able to develop new, better intuitions, ones that are better suited to our scientific worldview. This is Sellars’ project of negotiating between the scientific and the manifest images. We cannot do away with our intuitions any more than we can do away with language. We can, however, change our intuitions and our language for the better. For this reason it is important to look at our intuitions and critically evaluate and compare them. This pair of blog posts has hopefully gone some way towards that goal.