back to

Seminar: 4/20 - Michael Strevens, New York University

11:00 to 12:30 PM      at:  5101 Tolman

Concept Acquisition without Definitions

This paper is about "unsupervised concept learning", that is, learning in which a thinker, rather than trying to acquire a personal copy of a concept that already exists in their language or thought community, creates an entirely novel concept to organize the world around them in a new way. Scientific innovators build concepts in this way, but so, presumably, do ordinary people confronted by new colors, new organisms, or new kinds of weather events -- at least sometimes. Insofar as there is a standard account of the way in which novel concepts make their way into the mental inventory, it involves a kind of definition or stipulation by the concept's creator: a new "mental term" is defined or stipulated to pick out a certain kind of thing or property. Such stipulations bring into existence empirically irrefutable truths about the category (or property, or thing) corresponding to the new concept. But there appear to be very few or no such truths. It therefore seems worth looking for an alternative to the standard account. I will propose such an alternative.