One step is not enough: making better use of association norms to predict cued recall

Cued recall is strongly affected by the strength of the preexisting connection between the test cue and the information to be recalled, the target. In all past work, preexisting cue-to-target strength has been measured by the probability that the cue produced the target in free association. This paper presents four experiments showing that this use of such norms underestimates the strength of the connection and that a more accurate estimate can be obtained by incorporating indirect as well as direct connections in the estimate. Experiments 1 and 2 showed that in extralist cued recall both the strength and number of two-step indirect connections facilitate recall. Experiment 3 showed that three-step connections have negligible effects. Experiment 4 used an intralist task in which cue and target are first studied together, and the results showed once again that indirect connections can affect recall. In all of these experiments, indirect connections had an effect on recall that was larger when direct cue-to-target strength was weak than when it was strong. Implications for using association norms in research are described, and an algorithm for using association norms to measure cue-to-target strength is proposed.