In the years that immediately preceded the recent surge of developmental DRM studies, we reported several experiments on children's false memory for words using other paradigms (Brainerd & Reyna, 1996, 1998; Brainerd, Reyna, & Brandse, 1995; Brainerd, Reyna, & Kneer, 1995; Brainerd, Stein, & Reyna, 1998). In all of these studies, the measure of children's false memory was the false-alarm rate to unpresented words (which are usually called distractors or lures) that were related, usually semantically related, to words that had been presented on the list (which are usually called targets). Actually, the measure of false memory was the difference between the false-alarm rate for distractor words that were related to target words and distractor words that were unrelated to targets. This controls for the fact that there is always some baseline tendency to misrecognize distractors as old on the basis of guessing or other irrelevant strategies, a tendency that is normally higher in younger children than in older ones (e.g., Brainerd, Reyna, & Kneer, 1995) and is usually called response bias. Thus false memory, in the sense of erroneously remembering information that is related to information that was experienced, can only be said to be present if the false-alarm rate for related distractors exceeds the level of response bias. It is this difference, called the false-recognition effect, that is the standard index of false memory in recognition.
Although age levels and procedural details varied from study to study, all of the aforementioned experiments used a core design that emulates the standard technique that Underwood (1965) introduced for investigating false memory in adults. First, children were exposed to a fairly long list of familiar words, typically at least 30 words (e.g., couch, tree, snake, cat, happy, shirt, car, sky). To control for age differences in reading ability, the list was always presented orally. It was also presented at a rather leisurely pace (no faster than 2 seconds per word). Finally, the list had no obvious structure—each successive word seemed to be pretty much unrelated to its predecessors, unlike the categorized lists and DRM lists that were described above. Next, children responded to an orally presented old/new recognition test, typically under instructions to say "yes" whenever the experimenter said a word that they had just heard on the list and to say "no" otherwise. Further, children were cautioned not to fall prey to new words that were similar to old words in some way. The test list itself consisted of the same three types of probes that were administered in studies of children's narrative false memories (i.e., targets, related distractors, and unrelated distractors). Across our experiments, the type of false memory that was measured was varied by varying the relations between related distractors and targets. False memories based on all of the following semantic relations were studied: antonymy (e.g., sad is an antonym of happy), synonymy (e.g., sofa is a synonym of couch), within-category (e.g.,pants is an exemplar of the same taxonomic category as shirt), superordinate category (e.g.,,furniture is the name of the category to which couch belongs), subordinate category (e.g., oak is an exemplar of the tree category), and association (e.g., blue is a strong associate of sky). Considering that study and test lists were orally presented, we also investigated false memory for distrac-tors that were phonologically related to targets in some experiments (e.g., sat rhymes with cat).
The overall patterns that emerged from these experiments, which serve as background for recent studies of the DRM paradigm, ran as follows. First, children of all ages displayed false memory for all of the aforementioned target-distractor relations; false-alarm rates were always higher for related than for unrelated distractors. Second, the pattern of age change was usually the intuitive "memory gets better" trend: the tendency for children to recognize related distractors as old declined with age. (The overall age span in our experiments was 5 years to early 20s.) Generally speaking, although the amount of age change depended on the specific target-distractor relation, of course, the initial decline in false memory, across the elementary grades, was more marked than the subsequent decline. According to the distinctions that were discussed earlier, this developmental pattern has a straightforward interpretation: children's performance in the generic paradigm that we described is chiefly controlled by age improvements in the verbatim-processing abilities that suppress false alarms to related distractors. That this paradigm would not, instead, be highly sensitive to age improvements in the gist-processing abilities that foment such false alarms is apparent from two considerations. As mentioned, the lists that children heard were composed only of familiar words whose meanings were well understood by even the youngest children, leaving little latitude for age improvements in sheer meaning comprehension to increase false memory. Second, also as mentioned, the lists were constructed in such a way that there were no salient semantic relations between successive targets, leaving no way for age improvements in children's ability to connect meaning across different words to increase false memory. Thus, the developmental data argue that variations in false memory on the standard Underwood-type task are dominated by variations in verbatim processing rather than gist processing.
This brings us back to the main topic of this section: developmental DRM studies. We summarize the results of those studies in two waves. First, we consider initial baseline work, so called because it established and replicated the core developmental trends and ruled out possible artifac-tual explanations of those trends, which was published between 2002 and 2004. Next, we consider studies that have been subsequently published. These studies are concerned with more refined questions that are connected to theoretical explanations of false memory.
A few years before developmental DRM studies began to appear, Brainerd and Reyna (1998) and Ceci and Bruck (1998) noted that FTT made the counterintuitive prediction that false memory would increase with age in tasks that met certain theoretically specified conditions. Brainerd, Reyna, and Forrest (2002) reported the first experiments that evaluated this prediction in connection with the DRM paradigm. They noted that this paradigm differs from Underwood-type tasks in two key respects, both of which have ramifications for how false memory should be found to change with age. One of the differences is obvious: although the words that compose DRM lists are mostly familiar ones whose meanings will be understood even by young children, all of the words share meaning, so that listening to a DRM list repeatedly cues certain meanings. Owing to the way that the lists are constructed, the word that shares more of these repeatedly cued meanings than any other is a critical distractor, of course. So, in the terminology of FTT, a hallmark of DRM lists is that they cause very strong gist traces to be stored by adults ("I heard a list with lots of medical words"; "I heard a list with lots of furniture words") because adults are adept at connecting meaning across distinct exemplars (e.g., Seamon et al., 2002). The other difference between the DRM paradigm and Underwood-type tasks is more subtle. The high level of meaning connection among targets makes the use of verbatim processes to suppress false memories far more problematical than on Underwood-type tasks. Returning to the illustrative music list, suppose that a subject who is responding to a recall or recognition test is able to retrieve clear verbatim memories of having heard the words song, piano, band, horn, and symphony but cannot retrieve a clear verbatim memory of having heard music. This does not constitute compelling evidence that music was not presented and therefore should not be recalled or recognized for adults because adults, by virtue of their excellent meaning-connection abilities, are well aware that they heard several music-related words other than these five, which they also cannot clearly recollect (Brainerd, Reyna, Wright, & Mojardin, 2003).
Brainerd et al. pointed out that, taken together, these two features of the DRM paradigm, coupled with FTT's distinctions, lead to a clean prediction of the counterintuitive developmental trend that, as we saw earlier, constructivism made for narrative false memory: false memory for critical distractors ought to increase with age. Why? One reason is that DRM lists encourage a form of gist processing—specifically, the spontaneous connection of meaning across several words—that is far more difficult for young children than it is for older children or adults. Note particularly that we did not say that meaning connection was impossible for young children or utterly absent in young children, which would conflict with certain types of evidence. We merely said that it is far more difficult, which is demonstrated by the fact that they usually fail to exhibit the standard adult signs of connecting meaning across different words, such as semantic clustering in free recall (e.g., Bjorklund & Jacobs, 1985) and semantic proactive inference (e.g., Bjorklund & Hock, 1982). Thus, young children will be less likely than older children or adults to engage in this particular form of gist processing, which means that the strong gist memories that result from such processing will be less likely to be available to support intrusions and false alarms to critical distractors. The other reason for the age-increase prediction is that although verbatim memory for DRM targets will be improving with age, it will not have much effect on performance. More concretely, the fact that, say, 10-year-old children can remember more of the targets from the music list than 5-year-old children can does not necessarily put the 10-year-olds in a better position to suppress music, because they are also more likely to know that there are still remaining words that cannot be clearly remembered. In short, according to FTT, the DRM paradigm is an example of a procedure in which performance variations are primarily controlled by gist processing, both because a form of gist processing (connecting meaning across words) is encouraged that leads to particularly strong gist memories and because verbatim processing is not very helpful in suppressing false memories. Hence, the DRM paradigm is also a procedure that should reveal age increases in false memory, because the focal form of gist processing is known to increase with age (Reyna et al., 2007).
In three experiments, Brainerd et al. (2002) confirmed the predicted age increases with both recall and recognition tests, and the increases for recall, in particular, were dramatic. The initial experiment was an exceedingly simply one. They administered a series of DRM lists to a sample of 5-year-olds, using standard free-recall procedures from the adult literature (e.g., Deese, 1959; Roediger & McDermott, 1995). Children studied and recalled a total of 10 lists, each of which consisted of 12 words. The task was not difficult; 5-year-olds were able to recall a quarter of the list words after hearing them only once. False recall of the critical distractors was virtually nonexistent, however. Critical distractors are falsely recalled 35%
to 45% of the time when adults study these lists, but children recalled them only 6% of the time. Children falsely recalled other words, however, and when Brainerd et al. examined those words, they noted qualitative differences between what young children and adults remembered about DRM lists. The adult data show, not surprisingly, that adults "get the gist" of a DRM list. When they recall an unpresented word, over 80% of the time it is the critical word (e.g., music), and when they falsely recall other unpresented words, these intrusions are consistent with the meaning of that same list (e.g., drums, guitar). In sharp contrast, although 5-year-olds recalled words that were not on the list, their intrusions seemed to show that they did not get the gist of the lists. As noted, critical words did not usually intrude, and neither did other words that preserved the meaning of the list. Instead, children's intrusions consisted primarily of novel words that were unrelated to the meaning of list words (e.g., pirate following the music list) or were targets that had been presented on a previous list (e.g., nurse or sick if the doctor list was presented before the music list). (Nurse or sick are not gist errors, because if children get the gist of the music list, they will know that only music-related words should be recalled.) Naturally, all of this is congruent with the larger literature on memory development that, as we said, points to young children's limitations in connecting shared meaning across different words.
The other two experiments replicated and extended the finding of age increases in false memory. In Experiment 2, Brainerd et al. (2002) simply replicated the first experiment with an age manipulation. Children from two age levels, 5- and 7-year-olds, studied and recalled DRM lists. This particular age change was chosen both because the advent of formal school might affect false memory and because many developmental changes occur during these 2 years that are known to affect how children learn (e.g., White, 1970). In addition, a list manipulation was included. Deese (1959) found that although he constructed all of his lists in the manner described earlier, some of them produced much higher false recall than others, a result that was replicated many years later by Stadler, Roediger, and McDermott (1999). Half of the lists that were presented to the children were ones that produce the very highest levels of false recall, and half were lists that produce much lower levels of false recall. This was done to test the hypothesis that young children might show adult-like false memory on the "high" lists. They did not: levels of false recall of critical distractors continued to be quite low, averaging 5% for younger children and 7% for older children, and there was no difference in false recall for "high" versus "low" lists. Thus, for recall at least, the development of DRM false memory seemed to be very slow indeed.
The final experiment centered on false memory in old/new recognition, rather than recall. Three age levels were included in the design: 5-, 11-, and 20-year-olds (undergraduates). At all three age levels, the participants first studied and recalled a series of "high" and "low" DRM lists. After that, however, they responded to a single old/new recognition test for all the lists, which contained three types of probes: some of the targets from each list, the critical distractor for each list, and some distrac-tors that were not related to any of these lists. The overall patterns can be seen in Figure 9.2, where the plotted data involve the familiar signal detection statistic A' rather than raw hit and false-alarm probabilities (to control for the fact that response bias was higher at younger age levels). Contrary to the commonsense "memory gets better" thesis about development, it is starkly apparent that memory simultaneously gets better and worse: this tendency to judge targets to be old increases steadily between early childhood and young adulthood, but so does the tendency to judge critical distractors to be old. Moreover, one could not say that there is any overall improvement in memory accuracy because false alarms to critical distractors and hits to targets increased by about the same amount. With respect to false memory per se, note that false alarms to critical distractors
Was this article helpful?
Parenting is a challenging task. As a single parent, how can you juggle work, parenting, and possibly college studies single handedly and still manage to be an ideal parent for your child? Read the 65-page eBook Single Parenting Becoming The Best Parent For Your Child to find out how. Loaded with tips, it can inspire, empower, and instruct you to successfully face the challenges of parenthood.