Social learning and collective choice

Daniel Osherson

M. Stob

Steven Weinstein

pp. 319-347

in: Synthese 70 (3), 1987.

Abstract

To be pertinent to democratic practice, collective choice functions need not apply to all possible constellations of individual preference, but only to those that are “humanly possible” in an appropriate sense. The present paper develops a theory of humanly possible preference within the context of the mathematical theory of learning. The theory of preference is then exploited in an attempt to resolve Arrow's voting paradox through restriction of the domain of majoritarian choice functions.