Determinants of retrieval solutions during cognitive skill training: source confusions.

Diverse outcomes, both facilitative and disruptive, have been reported for the effect of interpolated item recognition tests on the acquisition of a cognitive skill. We collected data from a repeated set of 12 artificial arithmetic problems, soliciting compute/retrieve strategy reports after every trial. In one condition, a recognition test was administered after every three blocks of training. Recognition testing was found to depress retrieve frequencies in both younger and older adults, particularly for newly acquired items. Pairing training items with similar recognition foils mitigated these effects. This pattern of results could be explained by assuming that the participants based compute/retrieve decisions on item familiarity or frequency, tracked across both skill trials and recognition trials, and on a threshold influenced by source confusion. Variations in the threshold parameter could lead to depressed reports of item retrieval (our findings) or to elevated retrieval decisions, as has been shown in some other studies.