Effects of information-presentation mode and task complexity on the learning of probabilistic inference tasks

Learning of two-cue probabilistic inference tasks was studied in a 2 level of information-presentation mode (visual access to an ordered array of cue values vs. no such access) by 2 levels of task complexity (linear vs. nonlinear cue-criterion relations). The results replicated earlier findings in that performance was better in the linear tasks than in the nonlinear ones, and also showed that visual access to an ordered array of cue values facilitated performance, viz. there were no interactions. The results were interpreted as evidence that at least part of the suboptimal performance in ordinary experiments of this kind is due to cognitive strain introduced by the lack of control of the sequence of instances of cue-criterion pairs.