حافظه واقعی (اما نه حافظه کاذب) در معرض بازیابی ناشی از فراموش کردن در کودکان

Veridical and false memories of children aged 6 to 15 years were studied in two experiments with the retrieval-induced forgetting paradigm. Using the Deese–Roediger–McDermott (DRM) false memory word lists, children’s reports of true, but not false, memories showed evidence of retrieval-induced forgetting. These differences were observed across delays as long as 2 days following word list presentation. The lack of observation of retrieval-induced forgetting in children’s false memories provides evidence that a key assumption in the theory of retrieval-induced forgetting, cue independence, might not consistently apply. These experiments underscore the need for both practical and theoretically motivated study of true and false memories.

The long-standing debate regarding the nature of false memories in children has substantial practical and theoretical implications (see Brainerd, Reyna, & Ceci, 2008). The sometimes marked differentiations between veridical and false memory representations are increasingly being explored across a variety of retrieval paradigms that both inform us about how such memories are represented and can test the assumptions of associated theories (e.g., Howe, 2005). In the current experiments, we explored children’s true and false memories using a particular partial retrieval paradigm—the retrieval-induced forgetting paradigm. Retrieval-induced forgetting Cuing with part of the to-be-remembered material can be an effective method of facilitating recall. Indeed, partial retrieval cues are suggested regularly in investigative interviewing techniques with both adults and children as a way to assist with recollection of important event details (e.g., cued recall questions; Lamb, Orbach, Hershkowitz, Esplin, & Horowitz, 2007). However, there is a well-developed area of research that has demonstrated that partial retrieval can increase the difficulty in accessing associated memories not initially targeted for retrieval. This retrieval-induced forgetting (RIF) phenomenon is proposed to be a result of targeted memories inhibiting semantically related memories that compete for retrieval. These suppressed non-retrieved memories are subsequently more difficult to access (Anderson, Bjork, & Bjork, 1994). Forgetting associated with partial retrieval is especially concerning in a forensic context. Suppose that during a police interview of a witness only a limited number of specific questions are asked. The theory of RIF predicts that this partial retrieval may suppress future recall of other strongly associated details. The negative effects of partial retrieval have been observed beyond basic laboratory tasks (i.e., word lists), in children’s autobiographical memory (Phenix & Price, 2012), and in eyewitness memory for salient events (e.g., MacLeod, 2002, Migueles and Garcia-Bajos, 2007 and Shaw et al., 1995). Researchers who have studied partial retrieval phenomena such as RIF have often done so with the use of word lists. For instance, during the initial study phase of the RIF paradigm, participants study a list of CATEGORY–exemplar pairs (e.g., FRUIT–apple, FRUIT–banana, DRINKS–rum). Later during the partial retrieval practice phase, they practice retrieving half of the exemplars using word stems (e.g., retrieve and say aloud “apple” when presented with A P _ _ _) from half of the categories. During the final test phase, participants attempt to recall all exemplars from all categories shown during the original study phase. Retrieval-induced forgetting is found when the unpracticed exemplars (e.g., banana) from practiced categories (e.g., FRUIT) are recalled less often than the baseline condition of non-retrieved exemplars (e.g., rum) from unpracticed categories (e.g., DRINKS). This difference in performance between non-retrieved exemplars from practiced and non-practiced categories is the RIF effect and is theorized to result from the practiced exemplars (e.g., apple) inhibiting related non-retrieved exemplars (e.g., banana) relative to a baseline (e.g., rum). The RIF phenomenon is remarkably robust, although it is subject to several boundary conditions. For instance, when items are conceptually integrated, or strong interconnections between items are formed, RIF is attenuated (Goodmon & Anderson, 2011). Another factor that may attenuate RIF is the duration of delay between the practice and test phases. RIF does not appear consistently across long delays, with some researchers finding it to be a very short-lived phenomenon (Chan, 2009, MacLeod and Macrae, 2001 and Saunders and MacLeod, 2002) and others finding RIF after as long as 24 h or a week (Garcia-Bajos et al., 2009, Migueles and Garcia-Bajos, 2007 and Storm et al., 2006). It is likely that the nature of the to-be-remembered details affect the durability of any memory phenomena. The RIF literature focuses mainly on adults’ memory. However, the work with children may be particularly interesting because RIF is theorized by many (e.g., Anderson et al., 1994; but see, e.g., Jonker, Seli, & MacLeod, 2013) to be a result of inhibition stemming from competition between items. The development of inhibitory control during childhood has implications for children’s performance on many memory tasks because controlling initial temptations to provide a particular response (e.g., one that is socially appropriate or well-practiced) may mean that children will instead be more likely to rely on memory to make accurate decisions. Because of children’s developing inhibitory control, researchers have posited that even when RIF is observed, the mechanisms underlying RIF may differ across different developmental spans (e.g., Aslan & Bäuml, 2010). False memories and the DRM paradigm To examine whether reports of true and false memories differ in children as a function of partial retrieval practice, we decided to use the well-established Deese–Roediger–McDermott (DRM) paradigm known for generating false memories (Deese, 1959 and Roediger and McDermott, 1995). This paradigm involves the presentation of several lists of strongly related words for study and, during a later memory test, often results in false recall of words that are related but not on the studied list (often referred to as lure words). The DRM paradigm is useful for studying retrieval-induced forgetting because the word lists have a strong associative structure among items, as evidenced by the high rates of falsely recalled lure words ( Stadler, Roediger, & McDermott, 1999). With strong associative relations among items, false items should be activated during cuing and produce intense interference that should consequently be strongly suppressed. The inhibitory account of RIF, therefore, predicts that retrieval of targeted memories from the DRM lists will induce forgetting of the related false lures associated with each list. In fact, any activated items, including both false memory lures and veridical items, that interfere with the targeted memory during the partial practice phase should be suppressed during recall. Thus, it is hypothesized that veridical and false memories will be similarly affected by inhibition, specifically that both false and veridical memories that are related to memories targeted for retrieval will be suppressed. There is evidence in research with adults that both veridical and false memories are subject to RIF. For instance, Bäuml and Kuhbandner (2003, Experiment 2) found that partial retrieval practice suppressed reports of false memories from DRM lists (see also Kimball and Bjork, 2002 and Starns and Hicks, 2004). Bäuml and Kuhbandner (2003) argued that false recall was a result of activation of false memory during study and that the resulting RIF effect was observed because the falsely recalled items were not practiced during the initial partial practice phase and were consequently suppressed by the retrieved items. Interestingly, when the word lists included the lure words (i.e., no false memories for lures were possible; Experiment 1), no RIF was observed for list words in high false memory rate lists. The authors hypothesized that integration of list items was enhanced substantively with the inclusion of the lure words and ultimately concluded that veridical recall, but not false recall, shows integration effects (see also McDermott, 1996 and McDermott and Watson, 2001). Thus, although false memories were subject to RIF in adults, there was some indication that false and veridical representations differed. False memories in children Theoretical explanations for the development of false memories are varied, but a prominent theory that has been used to explain these processes in children is fuzzy trace theory (Brainerd & Reyna, 1995). Fuzzy trace theory proposes that each experience lays down two parallel memory traces; a verbatim trace represents the precise details of the experience, and a gist trace represents the more general meaning of the event. Fuzzy trace theorists have hypothesized that false memories develop when false details are introduced that are consistent with a gist representation. If verbatim memories are not accessible to allow for rejection of this new inaccurate information, the consistency between the new false information and the gist representation allows for the incorporation of the false memory to the gist-consistent representation. Thus, fuzzy trace theory proposes that veridical and false memories are created differently (see also Howe, 2005, for a hypothesis that veridical and false memories differ at encoding). Although retrieval-induced forgetting for DRM list words has been demonstrated in adults (e.g., Bäuml and Kuhbandner, 2003 and Starns and Hicks, 2004), there is reason to suspect that children’s experience and monitoring of false memories may be qualitatively different from that of adults. Theoretical support for this idea can be found in what appears to be a counterintuitive increase in false memory reports in older children relative to younger children. That is, whereas we have largely accepted that younger children are typically more susceptible to suggestion than older children (e.g., Bruck & Ceci, 1999), false memory development in some contexts is consistently more likely in older children than in younger children. In their thorough review and analysis, Brainerd and colleagues (2008) distinguished between spontaneous and implanted false memories (see also Otgaar & Candel, 2011, for an experimental comparison of these two processes). Implanted false memories are often observed in suggestibility paradigm research and are the source of the oft-cited conclusion that younger children are more suggestible than older children (e.g., Bruck & Ceci, 1999). The focus in the current work was on spontaneous false memories for which false memory is driven by endogenous processes (“meaning-driven reconstruction”). Brainerd and colleagues (2008) argued that such false memory development is enhanced by procedures that increase reliance on gist. For example, in connected meaning paradigms, category-consistent details are presented together and a false memory is likely to develop for related, but not experienced items. Thus, the connected meaning promotes reliance on gist, which increases the likelihood that false memories will develop. Because older children are better able to make meaning connections, false memory development will be more likely as children develop (into young adulthood) as long as the development of the false memory relies on making meaning connections between true and false memories. The body of research supporting this developmental reversal in false memory is robust across a number of different paradigms and age comparisons (Anastasi and Rhodes, 2008, Brainerd et al., 2002, Connolly and Price, 2006, Howe et al., 2004 and Sugrue and Hayne, 2006). There is also direct empirical evidence for developmental differences in false memories specifically created through the DRM paradigm. For instance, Dewhurst and Robinson (2004) argued for a developmental shift from phonological (rhyming) to semantic false memories with age. Furthermore, Dewhurst, Pursglove, and Lewis (2007) found that although 5-year-olds reported fewer false memories than 8- or 11-year-olds in response to DRM word lists, when the words were embedded within a story context, the 5-year-olds reported more false memories than the older children. The authors concluded that the story context facilitated young children’s ability to connect the lure word to the gist (i.e., increased meaning connections) but that their underdeveloped ability to access verbatim traces at retrieval relative to older children resulted in higher rates of false recall.

In applying a partial retrieval paradigm to the study of children’s true and false memories, we found retrieval-induced forgetting effects for true memories but not false memories. These effects were present after varying delays, including lengthy delays that are much longer than those in much of the extant work. The current experiments indicated that research with RIF in adults might not be consistently replicable in child populations. Furthermore, the lack of observation of RIF in children’s false memories provides evidence that a key assumption of RIF, cue independence, might not apply in some circumstances. These findings indicate a need for further exploration of the RIF paradigm with a focus on developmental and practical considerations.