ABSTRACT
Cued recall of word pairs is improved by asking participants to combine items in an interactive image. Meanwhile, interactive images facilitate serial-recall (Link Method), but even better when each item is imagined alongside a previously learned peg-word (Peg List Method). We asked if a peg system could support memory for pairs, hypothesising it would outperform interactive imagery. Tested with cued recall, five study strategies were manipulated between-subjects, across two experiments: (1) Both words linked to one peg; (2) Each word linked to a different peg; (3) Peg list method but studying as a serial list; (4) Interactive imagery (within-pairs); (5) Link Method. Participants were able to apply peg-list strategies to pairs, as anticipated by mathematical modelling. Error-patterns spoke to mathematical models; peg lists exhibited distance-based confusability, characteristic of positional-coding models, and errors tended to preserve within-pair position, even for inter-item associative strategies, suggesting models of association should incorporate position. However, the peg list strategies came with a speed–accuracy tradeoff and did not challenge the superiority of the interactive imagery strategy. Without extensive practice with peg list strategies, interactive imagery remains superior for associations. Peg strategies may excel instead in tasks that primarily test serial order or with extensive training.
Acknowledgements
The authors thank Tomi Ann A. Limcangco for assisting with earlier versions of the data analyses and Felicitas Kluger and Jeremy Thomas for helpful feedback on the manuscript. Supported by the Natural Sciences and Engineering Research Council of Canada and a studentship from the Undergraduate Research Initiative at the University of Alberta.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Although memory for pairs has been found to be successful with the method of loci, following three sessions of mnemonic training and tested with free recall; de Beni and Cornoldi (Citation1988).
2 The larger sample size in this group is because we ran a separate cohort of Link Method participants intended for a different study (unpublished).
3 Although distance functions tend to be approximately exponential, we sought a very simple way of quantifying the size of the contiguity effect, and with few degrees of freedom and requiring fewer data points given that some participants had missing data at some lags.