Reward information and cue selection following multiple-cue probability learning

Told 48 college-age Ss in a probability-reinforced, 2-choice verbal discrimination problem either to assume that the outcomes within a pair of cues were mutually exclusive or that the outcomes within a pair were independent. Each cue had an independent probability of being correct (r), and within a pair these values were identical; however, between pairs, the reward probabilities differed. On transfer tests involving all possible pairings of the 6 alternatives, Ss told that the outcomes were independent preferred the higher r values, and Ss told the outcomes were mutually exclusive displayed much less (but still significant) selection. The training data tended to support a pattern, rather than a linear model, for learning. (PsycINFO Database Record (c) 2006 APA, all rights reserved).