Nonreinforced trials in concept identification: Presolution statistics and local consistency

Examined the effects of nonreinforced (blank) trials during presolution of a concept identification task. 83 undergraduates served as Ss. A partial reinforcement group received 50% randomly administered blank trials; a control group was reinforced on all trials. Presolution statistics, number of called errors, and trials to trial of last error indicated that Ss did not resample hypotheses following blank trials. The degree of local consistency, measured by matching of responses on Trial n to actual or inferred feedback on Trial n-1 was weaker following uncalled successes and errors than following both called successes and called errors (p < .01 in all cases). Consistency was greater after called successes than called errors (p < .01). Results are explained through a learning-performance distinction and the notion of a passive state. (PsycINFO Database Record (c) 2006 APA, all rights reserved).