Summary

Two experiments were conducted to examine what effect inducing retrieval inhibition has on recall of both studied items (veridical recall) and critical, non-studied items (false recall) in the DRM pradigm. Different methods of inducing retrieval inhibition led to different patterns of results. The authors tried to relate the results to real world situations.

Experiment 1

This experiment employed the directed forgetting paradigm. In this paradigm subjects are presneted two lists for free recall, and after the first list they are told either to forget the first list and remember the second, or to continue to remember both. Subjects in the control conditions learn only one list. They engage in a filler activity in place of the other list. At recall, subjects are instructed to recall either list 1 or list 2.

If, when compared to instructions to remember, the instructions to forget have an effect, that is they reduce the access to L1, then two outcomes should result. First, recall of L1 should be reduced. Second, recall of L2 should be increased because of reduced proactive interference from L1. Results of previous studies are reasonably consistent in producing these outcomes.

The question of interest in this study was, what is the effect of directed forgetting instructions on recall and intrusion of critical items in the DRM paradigm. The authors made predictions based on five different theories each of which utilizes episodic and semantic information to explain the effect of directed forgetting instructions.

The predictions from three theories, (Fuzzy trace, Source Monitoring, and Attribution theory) all rely on the assumption that semantic information is unaffected, but use of episodic information is impaired, thus decreasing subjects’ reliance on episodic information. These theories would predict that directed forgetting instructions would decrease studied-item recall, and increase intrusions of critical items.

According to Spreading Activation theory, because recall is determined solely by semantic activation, there should be no effect of DF instructions on CII. Finally, according to the author’s interpretation of IAR theory, once the CI has been elicited during acquisition it will behave in the same fashion as a studied item. Thus CIIs should decrease with DF instructions.

Method

The design was a 2 X 3 factorial with Instructions (forget, remember, control) and Recall List (L1, L2) manipulated between subjects. Subjects were presented two DRM lists (15-items each) using SOP for those lists. Items presented on slides at 2-s rate.

Results

Studied items. Recall in the Control condition was higher than in either F or R conditions. Recall in F was not different from R. There was interference, both RI and PI. There was also a significant List X Instructions interaction, shown below. (Look only at R and F conditions.)

Percent Recall

List 1 List 2

Remember 58 51

Forget 46 60

Control 69 60

Critical Item Intrusions. Each subject only contributed one item, so the percent of CII is also the percent of subjects who produced a CI. In general, as recall of studied items decreased, the percentage of CIs increased. The Instructions X List interaction was significant.

Percent CII

L1 L2 Total

Remember 37 60 48

Forget 70 42 56

Control 43 23 33

Experiment 2

In this experiment the authors used the Part-list cueing paradigm to reduce recall of the list. In this procedure, after the study phase, subjects are given some of the items from the list as "cues" for recall of the remaining items. Presenting some portion of the list as cues results in lower free recall when compared to a non-cued condition. In this experiment the authors varied the number of words from the DRM list which served as cues (4, 8), and the associative strength of those items (strongest 4 or 8, weakest 4 or 8). Because all manipulations were within subject, each subject learned and recalled 12 different lists.

The predictions for this experiment were based on the authors’ extension of Rundus’ retrieval competition theory, with the following predictions: a) Increasing the number of cues (0, 4, 8) would result in decreasing recall of non-studied items; b) Increasing the number of cues (0, 4, 8), would result in decreasing frequency of Critical Intrusions; c) Increasing the strength of the associates used as cues should lead to a decrease in the number of CIIs; d) There were no predictions as to the effect of strength of the associations on recall of studied items.

Method

The subjects were 115 high-school and college students run in small groups. The words in each of the 12 lists were presented auditorialy at the rate of 1.5sec per word. Subjects were given 90sec for written recall. Eight cues were presented at the top of each recall sheet, with 0, 4, or 8 being words, and 8, 4, or 0 being a string of ##s.

Results

The authors always evaluated recall of the 7 never cued items, to reduce any item selections effects.

Percent correct recall

Cue type Weakest Items Moderate Items Strongest

Uncued 53 43 55

S4 48 S4 38 W4 51

S8 43 W4 37 W8 49

Percent Critical Item Intrusions

These comparisons represent the primary comparisons across conditions. Comparisons are to be made within a colunn.

Cueing Number of Cues Strength of Association

Uncued 54 Uncued 54 Uncued 50

Cued 44 4 cues 47 Weak 51

8 cues 40 Strong 38

In general, the results agreed with the authors’ predictions from the retrieval competition theory. Increasing the number of cues reduced recall of non-cued items and the frequency of CIIs. Increasing the associative strength (serial position) of the items used as cues had no effect on recall of studied items, but did decrease the percent of CIIs.