Information versus reward in binary choices

Abstract

This binary prediction experiment isolates the rewarding characteristics of outcomes from their informative characteristics. On each of 1000 trials S had to choose between 2 acts. He could guess which of 2 lights was correct, whereupon a nickel was added to or subtracted from a sum to be paid to him after the experiment, but he received no immediate feedback. Or he could observe which light was correct with no financial consequences. All Ss deviated from optimal strategy by seeking far too much information. Nonstationary instructions hindered performance, hypothetical predictions improved it. (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1966 American Psychological Association.