توسعه فرآیندهای مرتبط اتوماتیک و حافظه کاذب کودکان

We investigated children’s ability to generate associations and how automaticity of associative activation unfolds developmentally. Children generated associative responses using a single associate paradigm (Experiment 1) or a Deese/Roediger–McDermott (DRM)-like multiple associates paradigm (Experiment 2). The results indicated that children’s ability to generate meaningful word associates, and the automaticity with which they were generated, increased between 5, 7, and 11 years of age. These findings suggest that children’s domain-specific knowledge base and the associative connections among related concepts are present and continue to develop from a very early age. Moreover, there is an increase in how these concepts are automatically activated with age, something that results from domain-general developments in speed of processing. These changes are consistent with the neurodevelopmental literature and together may provide a more complete explanation of the development of memory illusions.

A well-established finding in the literature on children’s memory is that younger children are frequently more susceptible to misinformation effects than older children and adults. This age effect has been reliably shown in studies that have used suggestibility or misinformation manipulations to taint children’s memories (Ceci, Ross, & Toglia, 1987). Thus, the message from suggestibility research is that false memories decrease with age during childhood. In contrast, over the past years, studies that implemented the Deese/Roediger–McDermott (DRM) paradigm (Deese, 1959 and Roediger and McDermott, 1995) have shown that children’s spontaneous false memories increase with age ( Brainerd et al., 2006, Dewhurst et al., 2007, Dewhurst and Robinson, 2004 and Howe et al., 2009). In the DRM paradigm, participants study lists of words that are associated with a nonpresented word, namely the “critical lure”. For example, hot, snow, warm, winter, ice, and so forth all are associated with the critical lure cold. Despite never hearing the word cold, participants falsely recollect cold along with correctly remembering list items that were presented. What the findings from these different paradigms reveal is that there exists one set of conditions under which false memories increase with age (DRM paradigm) and another set of conditions under which false memories decrease with age (suggestibility paradigm). A recent study by Ceci, Papierno, and Kulkofsky (2007) may provide some insight into why these different trends have emerged across these two paradigms. These authors investigated how 4- and 9-year-olds’ individual associations for concepts influenced their suggestibility. Specifically, when a suggested distractor is judged to be strongly associated with an original item’s representation (and these judgements differed between 4- and 9-year-olds), children’s susceptibility to misinformation increased compared with when that information was less strongly associated regardless of age. Apparently, there is an important relation between knowledge representation that is age dependent and suggestibility. That is, suggestibility is greater when it is mapped onto age-appropriate knowledge representations and the task encourages associative processing. What these results suggest is that the discrepancy between the findings for suggestibility and spontaneous false memories may be more apparent than real. Indeed, studies to date seem to show that at least one hallmark of children’s false memories is the (often spontaneous) formation of associations and that the nature of these associations changes with age. To extend this line of inquiry and study these underlying associative processes in greater detail, we adopted a DRM-like procedure because this has quickly become the sine qua non for research on associative processing in both children and adults (see Gallo, 2006). One problem with using the DRM paradigm in child research is that the word lists that are typically used are those derived from adult word association norms (e.g., Nelson, McEvoy, & Schreiber, 1999). Research on children’s memory organization, like that just reviewed (Ceci et al., 2007), suggests that although associative links between concepts appear at a relatively young age, they are further strengthened and refined with increases in both knowledge and experience (Bjorklund, 1987 and Bjorklund, 2005). Because children’s knowledge base is different from adults’ knowledge base, using adult normed word lists may explain why children have fewer false memories than adults. Recently, three studies addressed this issue and used children’s associations to construct word lists (Anastasi and Rhodes, 2008, Carneiro et al., 2007 and Metzger et al., 2008). Interestingly, all three studies found that although developmental trends were considerably attenuated, false memories still tended to increase with age. These findings suggest that the increase in false memories with age cannot be attributed solely to changes in the content or organization of children’s knowledge base. Rather, as suggested in the associative activation theory (AAT) (Howe, Wimmer, Gagnon, & Plumpton, 2009), it is likely that these additional increases in false memory rates with age are due to increases in children’s ability to automatically activate and use associative relations much in the same way as adults (Kimball & Bjork, 2002). In particular, AAT suggests that false memory development is the result of increases in the number and strength of associative relations in children’s knowledge base as well as the speed and automaticity with which these associative relations are accessed and activated. An alternative theory, fuzzy trace theory (FTT) (e.g., Brainerd, Reyna, & Ceci, 2008), suggests that children’s memory is organized by two different memory traces: a verbatim trace and a gist trace. Verbatim traces encode surface features of items such as the phonological structure of a word. Gist traces encode the meaning or overall theme of a word or list of words. Children’s false memories increase with age because gist extraction processes improve with age. The difference between the two theories is that FTT explains age increases in false memories in terms of coincident changes in gist extraction, whereas AAT explains these same changes in terms of increases in the number and speed of direct activation processes among items in a semantic network. Both AAT and FTT agree that for false memories to occur, it is necessary to have a preexisting lexicon of associations and this lexicon needs to be activated mentally (see also Gallo, 2006, for a more in-depth discussion). Somewhat surprisingly, despite this agreement, there is little research on how children’s ability to form spontaneous associations develops. The current research aimed to shed more light on this. Specifically, in two experiments, we examined the development of both number and type of word associations and their interconnections in children’s lexicons and the automaticity with which these items are accessed. It has been demonstrated that children’s false memories are produced less automatically than adults’ false memories (Howe, 2005). That is, when instructed to “forget” items that have been studied, both children’s and adults’ true memories decrease. In contrast, for false memories, only those of children—but not adults—decreased in comparison with a control condition with no instructions (Howe, 2005 and Kimball and Bjork, 2002). Thus, when using a directed forgetting paradigm, although both children and adults can suppress true memories, only children seem to be able to suppress false memories. This lack of inhibition in adults’ false memories in a directed forgetting paradigm suggests that they are generated more automatically outside of conscious awareness. For children, these items, like items on the studied list itself, can be inhibited and appear to be treated as if they are part of the episodic list. To date, however, only Howe (2005) has investigated age differences in the automaticity of false memory generation, and even that study was restricted to a single component of automaticity, namely response inhibition. Because little is known about the automaticity of children’s associations, or how children come to automatically access and activate concepts in memory, the purpose of the current research was to investigate these developments in 5-, 7-, and 11-year-olds. To be precise, we examined a different aspect of automaticity from that studied by Howe (2005), namely age changes in the speed with which children can generate associative responses. Here we used speed of processing of an associative response as an index of automaticity. In particular, if the knowledge representation of an association is strong, then associative activation will be faster and less effortful. Faster and less effortful processing indicates automaticity. Therefore, for the purposes of the current research, we used speed of processing as an index of automaticity. The study of age changes in speed of processing is well motivated from other areas of research with children such as the finding that children’s automaticity in solving cognitive tasks increases with age. For example, children’s increase in processing speed on a variety of cognitive tasks (e.g., visual and memory search, mental rotation, mental addition) has been described by a single exponential function (Kail, 1988). This finding suggests that although the development of specific areas of knowledge may be a domain-specific characteristic of children’s memory development, changes in speed of processing may be a more domain-general development that affects all areas of children’s cognitive processing, including memory and, potentially, false recollection. However, the question arises as to how the development of both is connected. That is, studies that have examined false memory development have addressed either the issue of knowledge base (e.g., Anastasi and Rhodes, 2008, Carneiro et al., 2007 and Metzger et al., 2008) or that of automaticity (e.g., Howe, 2005) but never have addressed both issues simultaneously. An exception to this is an early study by Hall (1969) that gave an indication of how children’s associative processes develop in relation to knowledge base and automaticity. Hall found that 6- and 9-year-olds have more difficulty in correctly recognizing word associates that have been produced by themselves than in correctly recognizing words that have been produced by another person. This lack of memory advantage for the involvement of the self stands in contrast to an extensive literature that suggests increased accuracy of remembering self-generated events (e.g., actions, thoughts) in comparison with external-generated events (Foley et al., 1983 and Roberts and Blades, 1998). How can this discrepancy be explained? These inconsistent findings of a memory advantage for the self versus a memory disadvantage for the self suggest that these responses are generated in a qualitatively different fashion. Specifically, Hall’s (1969) finding that self-produced word associates are less well remembered suggests that they may be less conscious than those generated by another person. They may be less conscious because they may have been generated automatically based on a child’s preexisting knowledge base and the links among associations. Underwood (1965) used the phrase “implicit associative responses” (IARs) to describe the pattern of associations generated spontaneously by participants in response to words that have been presented. Interestingly, Hall (1969) also demonstrated that with development, children get better at discriminating whether information was self-generated or generated by another person. Discriminating whether information is self- or other-generated is a form of source monitoring—the ability to distinguish between memories based on the origin or source of those memories (Johnson, Hashtroudi, & Lindsay, 1993). There is a large body of evidence showing that children’s discrimination of internally versus externally generated actions and words increases with age (Foley et al., 1991, Foley and Johnson, 1985, Foley et al., 1983, Lindsay et al., 1991 and Roberts and Blades, 1998). Source-monitoring errors can also be a source of false memories in adults. In particular, source monitoring has been embedded in the activation-monitoring theory (AMT) of false recollection in adults (Roediger, Balota, & Watson, 2001). Like AAT, AMT views false memories as a product of spreading associative activation processes. Unlike AAT, AMT also states that for adults whose source-monitoring skills are better developed than those of children, false memories occur because of source-monitoring errors. Specifically, false memories occur because participants cannot determine the source of the critical lure—whether it has been generated internally or was part of the episodic list. In fact, adult participants frequently believe that they accurately remember that a critical lure had been presented and do so with a high degree of confidence (Roediger & McDermott, 1995). Unfortunately, a source-monitoring explanation cannot account for children’s increase in false memories with age because it would lead to the prediction that younger children should have more false memories than older children because younger children have poorer source-monitoring skills (Johnson et al., 1993). However, there may be conditions under which source-monitoring errors can explain children’s increase in false memories with age. In particular, in conditions where the spontaneous formation of associations is encouraged (as in the DRM paradigm) and where this association formation is derived from automatic processes, a correct source judgement may be more difficult. That is, if a “false” association is formed spontaneously, then it may become part of the studied associative information and cannot be distinguished from nonpresented information. Furthermore, if false memories are generated more automatically with age, then the greater the automaticity, the more difficult a source judgment may be, resulting in a developmental reversal of source-monitoring abilities. We investigated this possibility in the current research. Specifically, we were interested in children’s developing ability to discriminate self- from other-generated information when the information is derived from either automatic or conscious associative processes. Moreover, we were interested in whether source confusions, such as those observed in adults’ false memories, can explain the increase in the number of false memories with age during childhood when those memories are generated automatically. In sum, there are several processes that develop over childhood—reorganization of knowledge, speed of processing, and source-monitoring developments—and their interplay may contribute to false memory formation. The aim of the current research was to shed more light on the development of these processes, their interplay, and their relation to developmental increases in false memories. To do this, we used a word association task in Experiment 1 to investigate the development of children’s spontaneous associative processes based on preexisting knowledge. In Experiment 2, we investigated how children form associations between multiple word associates in a DRM-like paradigm. The crucial questions concerning children’s spontaneous associations were (a) how children’s knowledge organization changes with age, (b) how spontaneous associations are formed or activated when they are encouraged, and (c) whether the automaticity of children’s associative connections increases with age. In both experiments, we investigated how the speed of associative responses increased with age by measuring the time children needed to generate their own word associations from their own knowledge bases. In addition, we were interested in how the development of associative representations and their activation relates to source-monitoring development and false memories. The key question concerning source monitoring was whether we would find that increasing automatic processing with age is correlated with increasing difficulty in judging the source of information in memory. To investigate this question, across both experiments we explored children’s ability to discriminate the source of generated information (self vs. other) as a function of whether that information is generated consciously or automatically. Investigating the development of implicit associative processes, the speed with which these associations are generated, and source-monitoring developments, as well as their relation to false memories, allows us to gain more insight into how these processes develop in isolation and in combination as well as to document their relevance for developmental increases in false memories.