Abstract:
The Problem: Forty-eight human Ss were presented with a four choice probability learning-task. Two groups of 24 Ss each were placed under different incentives - Equitable Wage (EW) and Gambling Wage (GW). It was anticipated that a transfer of learning effect would be present. Also, it was anticipated that the test in which the most frequently and least frequently presented alternatives were in opposite positions from that presented during pre-training would be the most difficult test presented.
Procedure: All Ss were presented with 120 trials during pre-training and 60 trials in each of the four probability schemes during the test session. During one of the four tests, each alternative was the most frequently presented alternative. In each test, the four alternatives were presented in the following percentages: 56.25; 18.75; 18.75; and, 6.25.
Findings and Conclusions: The GW group (greater incentive given) performed significantly more accurately than the EW group (lesser incentive given). Although both incentive groups approximated probability matching choice behavior, Ss in the GW group made more correct responses by distributing more responses to the most frequently presented alternative and fewer responses to the least frequently presented alternative than did the EW group. No support was present for a transfer of training effect nor the expected task difficulty.
