Later Maturation of the Beneficial Than the Detrimental Effect of Selective Memory Retrieval

Method

Design and procedure

The experiment had a 2 × 2 × 3 design, with a within-participants factor of cue condition (remember, forget) and between-participants factors of prior nontarget retrieval (present, absent) and age group (second graders, fourth graders, seventh graders). In both cue conditions, the 12 items on List 1 were presented successively on index cards, one every 4 s, in random order. Two study cycles were conducted to avoid potential floor effects in the youngest age group. Thereafter, the interlist cue (i.e., the instruction to remember or to forget) was provided. In the remember condition, participants were told that List 1 should be remembered for an upcoming memory test. In the forget condition, participants were told that List 1 had been presented erroneously and thus should be forgotten (see Aslan, Staudigl, Samenieh, & Bäuml, 2010). Then, List 2 was presented in exactly the same way as List 1.

In both cue conditions, after a 90-s distractor task, the children were asked to recall and write down the items from List 1. Target and nontarget words were presented on separate sheets of paper. For each target word, the initial letter was given, followed by a blank line (e.g., “A____”; see Fig. 1). For each nontarget word, the word stem was listed, followed by a blank line (e.g., “Unc___”). ¹ On each test sheet, the order of the initial letters (for targets) or word stems (for nontargets) was random. Only children who were asked to recall the nontarget words were given the sheet listing the word stems. The children had 1 min to complete each sheet but were given extra time if they needed it. List 2 items were tested subsequently, but the results are not reported here. After a short break, the second cue condition started. The order of the remember and forget conditions was counterbalanced across participants, as was the assignment of lists to the two conditions (see Bäuml & Samenieh, 2010, 2012).

OPEN IN VIEWER

Fig. 1.

Overview of the experimental procedure. In the study phase, children studied two word lists, each containing 12 words. After study of List 1, the children received a cue to either forget or continue to remember that list. In the test phase, the children were asked to recall predefined target words (boldface type) from List 1 (e.g., “Apple,” “Table”). For each target word, the initial letter was given, followed by a blank line. The target words were tested with or without prior retrieval of the list’s remaining (nontarget) words (e.g., “Uncle,” “Sugar”). When the children were cued to retrieve nontargets, each nontarget’s word stem was listed, followed by a blank line. For this illustration, German words were translated to English.

Results

Recall rates for target words are shown in Figure 2. A 2 × 2 × 3 analysis of variance with the factors of cue condition (remember, forget), prior nontarget retrieval (present, absent), and age group (second graders, fourth graders, seventh graders) revealed significant main effects of cue condition, F(1, 138) = 10.4, MSE = 0.037, p = .002, η _p ² = .07, and age group, F(1, 138) = 20.6, MSE = 0.063, p < .001, η _p ² = .23. These main effects reflect both reduced overall recall in the forget condition compared with the remember condition and higher overall recall in older compared with younger children. There was also a significant two-way interaction between prior nontarget retrieval and cue condition, F(1, 138) = 24.1, MSE = 0.037, p < .001, η _p ² = .15, reflecting the fact that, with the data collapsed across age groups, prior retrieval of nontarget words reduced recall of to-be-remembered target words and enhanced recall of to-be-forgotten target words.

OPEN IN VIEWER

Fig. 2.

Percentage of recalled (a) to-be-remembered and (b) to-be-forgotten target words as a function of age group and the presence or absence of prior nontarget retrieval. The error bars represent +1 SE.

Note, however, that there was also a significant three-way interaction, F(1, 138) = 3.3, MSE = 0.037, p = .040, η _p ² = .05, suggesting that the beneficial effect of retrieval develops later than its detrimental effect. Indeed, whereas the detrimental effect on recall of to-be-remembered target words was present in all three age groups (second graders: 16.7%, p = .010; fourth graders: 12.5%, p = .032; seventh graders: 16.7%, p = .019), the beneficial effect on recall of to-be-forgotten words was present in seventh graders (20.8%, p = .006) but absent in the two younger age groups (second graders: −7.3%; fourth graders: 7.3%; both ps > .240). (For additional analyses of the data, see the Supplemental Material available online.)

Discussion

We found in all three age groups a detrimental effect of prior nontarget retrieval on retrieval of to-be-remembered target material, which is consistent with results from prior work (e.g., Ford et al., 2004; Zellner & Bäuml, 2005). More important, we found a beneficial effect of selective memory retrieval, with children’s prior nontarget retrieval improving their retrieval of to-be-forgotten target words. Unlike the detrimental effect, this beneficial effect was present only in the oldest age group tested. These results reveal a developmental dissociation between the two effects of selective memory retrieval, indicating later maturation of the beneficial than of the detrimental effect.

The two faces of memory retrieval have been attributed to different mechanisms: Inhibition underlies the detrimental effect, and context-reactivation processes underlie the beneficial effect (Bäuml & Samenieh, 2012). Consistent with this view and with recent evidence for deficient inhibitory processes in kindergartners compared with second graders (Aslan & Bäuml, 2010), the results suggest that the detrimental effect of memory retrieval develops shortly after the time of school entry, and that children from this age on can rely on retrieval-induced inhibitory processes to reduce interference from related memories. The beneficial effect of memory retrieval develops after elementary school; only then can children capitalize on retrieval-induced context-reactivation processes to make forgotten memories accessible again.

Context reinstatement can involve associations of items with the context itself (e.g., Howard & Kahana, 2002); thus, older (though not younger) children’s retrieval of nontarget items may reactivate contextual information, which then guides retrieval of target items. Alternatively, or in addition, context reinstatement may involve the use of direct interitem associations, such that retrieved nontarget items themselves serve as a type of (episodic) contextual cue for target items (e.g., Raaijmakers & Shiffrin, 1981). Because younger children generally encode items in a less associative manner than older children or adults do (Schneider & Pressley, 1997), part of the beneficial effect of retrieval that we observed may have been due to age-related improvements in encoding processes. Future research may address the issue of the exact role of encoding and retrieval processes in the development of the beneficial effect of selective memory retrieval.

Our results indicating that young children do not yet show retrieval-induced context reactivation is consistent with Lehmann and Hasselhorn’s (2010) findings regarding the contiguity effect, which is typically attributed to context-reactivation processes and is reduced, if not absent, in 8-year-olds. Moreover, our results do not conflict with research in which young children were found to use and profit from contextual information (e.g., Dietze & Thomson, 1993; Wilkinson, 1988). In these previous studies, either the contextual information was present throughout the recall phase (as in the case of physical context reinstatement), or specific mental-reinstatement instructions or questions were introduced before the recall phase to reactivate the contextual information. Neither was the case in the current study, in which reactivation processes could be triggered only by the selective retrieval of some of the previously encoded items. Thus, whereas dedicating exceptional effort to reactivate contextual information may be critical for the reinstatement of context in younger children, it does not seem to be necessary in older children.

A forget cue can trigger similar processes in children and adults (Aslan et al., 2010; Harnishfeger & Pope, 1996), which indicates that in both groups, the presentation of a forget cue deactivates the study context (Geiselman, Bjork, & Fishman, 1983; Sahakyan & Kelley, 2002). Therefore, the fact that we did not find beneficial effects of young children’s nontarget recall in the forget condition suggests that these children did not engage in context-reactivation processes. Alternatively, young children may have intact context-reactivation processes, but their directed forgetting may be mediated by processes other than context deactivation. However, no support for such a claim has arisen to date.

In sum, this is the first study to examine the detrimental and beneficial effects of selective memory retrieval in children. Whereas second, fourth, and seventh graders showed the detrimental effect of memory retrieval, only seventh graders showed the beneficial effect. From a more theoretical perspective, the results indicate later maturation of the beneficial than the detrimental effect of selective memory retrieval. From a more applied perspective, the results suggest that in eyewitness situations, in which access to the original encoding context is often impaired, interrogators may indeed trigger children’s memory for critical information through guided retrieval of related information. Yet our results also suggest that mainly older children would benefit from such guided retrieval, and that other techniques, such as mental-reinstatement techniques (Dietze & Thomson, 1993), might be necessary to improve eyewitness testimony in younger children. The present findings thus provide important clues and restrictions for the further development and refinement of age-appropriate interrogation techniques for children.