Abstract:

The part-list cuing effect refers to the phenomenon that when providing a subset of previously learned items as retrieval cues, people’s recall performance for the remaining items is often worse compared to when retrieval cues are absent. Memory research also showed that items with high value are generally better remembered than items with low value. However, it is unclear how the values of items affect the part-list cuing effect and its persistence. Through two experiments, this study investigated the influence of item value on the part-list cuing effect.

Experiment 1 employed a part-list cuing paradigm in a value-directed memory task. During the learning phase, participants were asked to study 80 category exemplars which were assigned different values (1 or 10 points). Then, a distractor task of either 60s or 100s (in the part-list cuing or no-part-list cuing condition, respectively) was then given. Participants in the part-list cuing condition were then presented with 20 high- or low-value cues for 40 seconds, and were told to read these items aloud and use them to mentally recall the remaining items. Finally, participants completed an old/new recognition task. The procedure can be seen in Figure 1.

The repeated-measures ANOVA of accuracy showed a significant main effect of cuing condition, F(2, 58) = 8.99, p < 0.001, η2 p = 0.24, and a significant main effect of targets value, F(1, 29) = 18.99, p < 0.001, η2 p = 0.40. An interaction cuing condition × targets value effect was found, F(2, 58) = 4.93, p = 0.01, η2 p = 0.15. A further simple effect analysis revealed that for the high value targets (F(2, 28) = 4.16, p = 0.026, η2 p = 0.23), the recognition accuracy under the no cues condition was significantly higher than that of high value (p = 0.048) and low value (p = 0.045) cues conditions; whereas for the low value targets (F(2, 28) = 7.48, p = 0.002, η2 p = 0.35), the recognition accuracy under the no cues condition (p = 0.002) and low value cues condition (p = 0.004) was significantly higher than that of high value cues condition (Table 1).

Experiment 2 further manipulated the encoding condition (i.e., 1-study encoding vs. 2-study-test encoding) and the test schedule (i.e., immediate test vs. final test). In the 1-study condition, participants received only one study cycle, but went through two study-test cycles in the 2-study-test condition. The immediate test phase is the same as Experiment 1; the final test involved a final recognition test after a 5min distractor task. The procedure can be seen in Figure 2.

For the 1-study encoding condition, the repeated-measures ANOVA of accuracy showed a significant main effect of cuing condition, F(2, 72) = 23.29, p < 0.001, η2 p = 0.39, a significant main effect of targets value, F(1, 36) = 12.68, p < 0.001, η2 p = 0.26, and a significant main effect of test schedule, F(1, 36) = 16.16, p < 0.001, η2 p = 0.31. An interaction cuing condition × targets value × test schedule effect was found, F(2, 72) = 3.89, p = 0.025, η2 p = 0.10. A further simple effect analysis revealed that for the high value targets, the recognition accuracy under the no cues condition was significantly higher than that of high value (p < 0.001) and low value (p = 0.008) cues conditions in the immediate test (F(2, 35) = 13.50, p < 0.001, η2 p = 0.44), the recognition accuracy under the no cues condition (p = 0.001) and low value cues condition (p = 0.014) was significantly higher than that of high value cues condition (F(2, 35) = 8.53, p = 0.001, η2 p = 0.33) in the final test; for the low value targets, the recognition accuracy under the no cues condition (p = 0.004) and low value cues condition (p = 0.017) was significantly higher than that of high value cues condition (F(2, 35) = 6.45, p = 0.004, η2 p = 0.27) in the immediate test, the recognition accuracy under the no cues condition (p < 0.001) and low value cues condition (p = 0.010) was significantly higher than that of high value cues condition (F(2, 35) = 10.59, p < 0.001, η2 p = 0.38) in the final test.

For the 2-study-test encoding condition, the repeated-measures ANOVA of accuracy showed a significant main effect of cuing condition, F(2, 70) = 6.07, p = 0.004, η2 p = 0.15, and a significant main effect of targets value, F(1, 35) = 7.75, p = 0.009, η2 p = 0.18. An interaction cuing condition × targets value effect was found, F(2, 70) = 4.09, p = 0.021, η2 p = 0.11. A further simple effect analysis revealed that, for the high value targets (F(2, 34) = 2.67, p = 0.084), the recognition performance among no cues, high value cues and low value cues were not significant, whereas for the low value targets (F(2, 34) = 4.95, p = 0.013, η2 p = 0.23), the recognition accuracy under the no cues condition (p = 0.015) and low value cues condition (p = 0.020) was significantly higher than that of high value cues condition (Table 2).

Results from the two experiments collectively showed both the assigned values of cued and test items affected the item recognition performance: cue items with high value resulted in poorer target item recognition performance than those with low value; however, the recognition accuracy was higher for target items with high- than low-value, and the high-value target items were more sensitive to the presentation of part-list cuing. The emergence and persistence of part-list cuing was also modulated by item values. Under the 1-study condition, the high-value cues led to worse target item recognition regardless of the values of the target items, and this detrimental effect was observed in both immediate and final tests. In contrast, the low-value cues only caused poorer recognition of high-value targets in the immediate test. Under the 2-study-test condition, only high-value cues caused recognition impairment of the low-value targets in both immediate and delayed tests. The above results partially validate the two-mechanism account of part-list cuing, and also are a key supplement to this hypothesis: the role of part-list cuing on memory retrieval is not necessarily manifested as a lasting impairment in the low associative coding condition, or a transient impairment in the high associative coding condition, and the item value also influences the strength and persistence of the role of part-list cuing, and it is also necessary to take into account the role of item value when defining the role of part-list cuing on memory retrieval from the perspective of item associative encoding.

Key words: part-list cuing, item value, encoding conditions, test schedule