Second-order conditioning in human predictive judgements when there is little time to think

Abstract

Associative accounts uniquely predict that second-order conditioning might be observed in human predictive judgements. Such an effect was found for cue X in two experiments in which participants were required to predict the outcomes of a series of training trials that included P + and PX−, but only when training was paced by requiring participants to make a prediction within 3 s on each trial. In Experiment 1 training on P + ended before training was given on PX − . In Experiment 2 trials with P+, PX−, T + and other cues were intermixed. In the unpaced group inhibitory learning was revealed by a summation test, TX versus TM, where M was a control stimulus. These results suggest either that pacing interferes with learning successive associations more than with learning simultaneous associations or that lack of time to think interferes with inferential processes required for this type of inhibitory learning.

Acknowledgments

We wish to thank Evan Livesey for programming the computer tasks for the experiments in this paper. This research was supported by an Australian Postgraduate Award and a Postgraduate Research Grant from the School of Psychology, University of Sydney to D.M.K. and a grant from the Australian Research Council to R.A.B.

Notes

¹ The mean time per trial was calculated by dividing the total time for a participant's training session by the number of trials. This average exceeded the nominal time set for the paced condition (5 s) because one self-paced screen at the beginning of the training phase was included in the total training time. This screen was identical in both groups.