The effects of information presentation mode (i.e. visual access vs. not access to an ordered array of cue combinations) and learning paradigm (i.e. selection vs. reception of cue-combinations) on learning of two-cue probabilistic inference tasks were studied in two 2 × 2 factorial experiments representing two levels of memory strain. The results showed that information presentation mode affected performance regardless of amount of memory strain, while learning paradigm affected performance only under conditions of high memory strain.