Part 2:Memory Monitoring And Control in Japanese And German Preschoolers
Mar 20, 2022
For more information:ali.ma@wecistanche.com
Click to organic Cistanche product for memory
Results
All data are included as a supplementary information file. The preliminary analyses found that there was no main effect of item order or gender, nor did they interact with any other main factors. Thus, we did not consider these factors for the rest of the analyses. In all reported analyses, we compared the models with and without Age – Age was dropped if the model including Age did not significantly improve the overall model fit.
Memory performance We frst asked whether our manipulation of memory strength during memory encoding was efective. Figure 2 presents the mean proportion of trials in which children accurately remembered items as a function
Fig. 2 Mean proportion of trials in which children accurately remembered items as a function of Encoding type and Country. Error bars indicate standard errors
of Encoding type and Country. Children remembered the repeated items (M = .90, SD = .16) more accurately than the non-repeated items (M = .82, SD = .17), F (1, 98) = 33.650, p < .001 but after controlling for Age (F (1, 96) = 12.994, p < .001) it was no longer signifcant F (1, 96) = 2.439, p = .122. The model including Age signifcantly improved the model ft (-2LL = -181.351 vs. -195.936). No other efects were signifcant including Country F (1, 96) = .330, p =
.567 or Country X Encoding F (1, 96) = .283, p = .596.
Monitoring memory accuracy and strength Importantly, we asked whether children’s confidence judgments corresponded to their memory accuracy and strength. Figure 3 presents children’s mean confidence scores (score 0- low confidence; score 1- medium confidence; score 2- high confidence) as a function of Country and Item types (Repeated- accurately remembered items vs. Non-repeated accurately remembered vs. Inaccurately remembered items). Because we reasoned that memory strength would not be differentiated between repeated and non-repeated conditions for the
Fig. 3 Mean confdence as a function of Item type and Country. Error bars indicate standard errors
inaccurately remembered items we combined these condi- tions for the inaccurately remembered items. The model including Age (F (1, 92.535) = 4.728, p = .032) did not signifcantly improve the model ft (-2LL = 257.274 vs. 269.702). Item type was signifcant, F (2, 84.568) = 19.010, p < .001. Children rated the inaccurately remembered items (M = 1.19, SD = .74) as less confdent than either the non- repeated-accurately remembered (M = 1.66, SD = .39) (t (75.697) = 5.987, p < .001) or the repeated-accurately remembered items (M = 1.70, SD = .37) (t (77.324) = 6.309, p < .001). Country was not signifcant F (1, 96.762) = .005 p = .945, nor was Country X Item type F (2, 84.568) = .277,
p = .759.
Sorting decisions An important question was whether chil- dren’s confdence judgments corresponded to their sorting decisions. Figure 4 presents children’s mean confdence scores as a function of Country and Sorting decision. The model including Age (F (1, 73.55) = 4.638, p = .035) did not signifcantly improve the model ft (-2LL = 246.056 vs. 246.743). Children’s confdence was higher for the items they chose for later evaluation (open-eyes box) (M = 1.71, SD = .39) than for the items they did not choose for later evaluation (closed-eyes box) (M = .94, SD = .71), F (1,
74.601) = 63.072, p < .001. Sorting X Country was not signifcant F (1, 74.601) = .027, p = .870 as well as CountryF (1, 73.316) = .359, p < .551.
Next, we asked whether children’s sorting decision based on their confdence level as demonstrated by the previous analysis corresponded to their memory accuracy. Figure 5 presents children’s mean memory scores as a function of Country and Sorting. The model including Age (F (1, 49.385) = 5.310, p = .025) did not signifcantly improve the model ft (-2LL = -18.622 vs. -20.397). There was a signifcant interaction of Sorting X Country F (1, 96.108) = 8.260, p = .005 as well as Country F (1, 103.280) = 9.491, p = .003 and Sorting F (1, 96.108) = 30.523, p < .001.
Fig. 5 Mean accuracy as a function of Sorting decision and Country. Error bars indicate standard errors
Japanese children sorted accurately remembered items more frequently into the open-eye box (M = .88, SD = .16) than into the closed-eye box (M = .60, SD = .33) F (1, 96.967) = 37.803, p < .001 whereas German children sorted accurately remembered items equally into the two boxes (open-eye box: M = .89, SD = .15; closed-eye box: M = .81, SD = .24) F (1, 95.362) = 3.293, p = .073. German children sorted accu- rately remembered items more frequently into the closed-eye box than Japanese children F (1, 161.946) = 14.785, p < .001 whereas there was no group diference with respect to the open-eye box F (1, 159.822) = .198, p = .657.
HMeta‑ d We found no group diference for children’s mem- ory accuracy-confdence as HDI of diference of posterior group estimates overlapped with 0: [-1.866 ~ .569]. By con- trast, there was a group diference with respect to children’s memory accuracy-sorting: [.000123 ~ 1.4264]. See Fig. 6. Note that there was no group diference in terms of memory performance t(98) = -1.63, p = .106, confdence level t(98) = -.283, p = .778 or sorting level t(98) = .228, p = .820. 1
Discussion
In the present research, we investigated 3.5 ~ 5-year-old Japanese and German children’s memory monitoring and control. As noted in the Introduction, the empiri- cal data are scarce concerning metacognitive abilities in young children growing up in non-WEIRD cultures. To our knowledge, the present research provides initial evidence both for cross-cultural similarity and diversity in
1 Although our interest was a direct group-level comparison of Mratio, we also provide values of Meta-d’: For confidence-accuracy: Meta-d’ = 1.67 for Japanese children, Meta-d’ = 1.75 for German children; for sorting-accuracy: Meta-d’ = 2 for Japanese children, Meta-d’= 1.47 for German children.
young children’s memory monitoring and control. Specifically, we found, first, that children rated accurate memory responses more confidently than inaccurate responses. Importantly, this was similar for both cultural groups. Second, both cultural groups similarly sorted out for prize attribution the items they felt more confident about. How- ever, Japanese children’s sorting decisions corresponded more closely to their memory accuracy than German children’s decisions. Below we discuss these findings in more detail.
Memory monitoring Both Japanese and German children rated their accurate memory responses more confidently than their inaccurate responses. This finding is consistent with prior research showing that young children are able to monitor their own uncertainty (e.g., Coughlin et al., 2015; Goupil & Kouider, 2016, Lyons & Ghetti, 2013).
Memory control Both Japanese and German children tended to exclude their less confidently rated responses for later prize evaluation. However, Japanese children’s sorting corresponded to their memory accuracy, whereas German chilresorting did not.
In summary, we found similarities in Japanese and German children’s memory monitoring and control. Accurately remembered items were rated as more confident than inaccurately remembered items. Additionally, more confidently rated items were appropriately selected for later evaluation. Kim et al. (2020) also reported similarities between Japanese and German children in their metacognitive ability to assess their own knowledge. The findings of cultural similarity in the present study together with those of Kim et al. (2020) may indicate the universal importance of metacognitive monitoring and control in adaptive learning and behaviors, and thus inhuman and nonhuman survival. Indeed, basic memory monitoring and control are present even in preverbal infants (e.g., Goupil & Kouider, 2016).
In the Introduction, we discussed several related reasons for cultural differences of metacognitive monitoring and control appearing early in childhood. If different cultural practices surrounding teaching and instructions as well as parental sensitivity in parent-child interactions – and in relation to different notions of self (interdependent vs. independent) – contributed to metacognitive monitoring and control even in early childhood, then we would have observed more pronounced and consistent differences between the two cultural groups targeted in our study. On the contrary, the cultural difference emerged only with respect to the sorting decision: Japanese children’s sorting for evaluation corresponded more closely to the accuracy of their responses than German children’s. These findings were further confirmed by HMeta-d controlling for potential confounds (also note that both groups of children had a similar level of memory performance and of response biases (confidence or sorting level)).
How can we explain our funding of a cultural difference in the sorting decision? Cross-cultural studies demonstrate that internal motivation, as well as behaviors and even individual well-being, are modulated by a culturally diferent construal of the self – whether the self is understood independently of others (as in Germany) or interrelated to others (as in Japan). For example, when primed by public eyes (thus eliciting social evaluation) during an IQ test, Japanese American adults scored higher than Caucasian American adults, whereas a reversed pattern was found in the absence of priming conditions (Na & Kitayama, 2012). This suggests that Japanese American participants' internal motivation to perform and behave is more likely to be influenced by others’evaluation than participants of European descent because the relation to others is a more salient dimension in selfhood in an interdependent than in an independent conception of selfhood. In this research, the sorting task required children to sort the items that might qualify for a reward. Thus, it is possible that even at this young age, children’s performance is modulated by a cultural difference in self-construal related to internal motivation and public evaluation (see also Iyengar & Lepper, 1999). Japanese children might have a stronger motivation to perform well in general – accompanied by higher social expectations –compared to German children. Alternatively, German children may be more prone to risk aversion (over-estimating the risk of error when evaluating their chances of getting a prize) – and this might be because German children are under higher pressure to perform rather than Japanese children.
The cultural difference of a tighter correspondence between sorting and accuracy may also be explained by a difference in executive function (defined as a range of cognitive abilities whose function is to control behaviors and other cognitive processes in order to attain a certain goal). Roebers (2017) proposed that metacognition and executive function are guided by the same underlying cognitive self-regulatory processes: executive function might be causally involved in the development of metacognitive control, especially in the early years. In the sorting task, children were asked to sort the items for later prize-winning consideration, which is likely to involve executive function. Studies report a higher executive function among children growing up in East Asian countries than among those in Western countries (e.g., Imada et al., 2013). Thus, to the extent that metacognitive control involves executive function (e.g., Koren et al., 2006) – whose development varies across cultures – we might explain our data by a cultural difference of metacognitive control. Future studies should replicate the present findings and address whether the cultural differences observed in the present study become increasingly heightened or diminished with increasing age. Nevertheless, the present study is the first step toward answering cross-cultural variations and similarities in young children’s metacognitive development. The above explanations are consistent with a dual-process theory of metacognition (e.g., Koriat, 1997; Koriat & Ackerman, 2010; Koriat & Levy-Sadot, 1999; Proust, 2013). According to the theory, two kinds of inputs are used in metacognitive evaluations: 1) Experience-based evaluations are guided by automatic inferences based on mnemonic and heuristic cues gathered from immediate feedback from task performance. 2) Information-based evaluations are guided by analytic and deliberate inferences and theories. In the present study, the tasks seem to depend primarily on experience-based evaluations (i.e., subjective feelings of confidence). While it is not excluded that children also rely upon information-based evaluations (e.g., beliefs about the task), this form of metacognition has been shown to typically develop later in development – around school age) (e.g., Schneider & Lockl, 2008). Given the observed cultural difference specific to the sorting task, therefore, it is plausible that metacognitive control is modulated by cultural factors such as social values and overall learning goals. Moreover, the difference between a prospective and a retrospective form of evaluation (respectively used in sorting one's responses and expressing a confidence judgment) might be itself sensitive to cultural influences. Indeed, diferent brain regions are involved in prospective judgments (e.g., ease of learning or feeling of knowing) and in retrospective monitoring judgments (confidence monitoring) (Fleming & Dolan, 2012). Although the sorting task may not exclusively exploit a predictive source of information diferent from confidence (because the decision was likely to be made partly on the basis of retrospective judgments of memory and/or confidence), further research still might scrutinize the existence of a potential contrast between future-oriented and past-oriented metacognitive evaluations in young children – whether they are subject to diferent external influences. Finally, the absence of cultural difference in confidence monitoring could be due to small efects in the present study.
What cannot be denied is that accuracy in metacognitive monitoring and control is beneficial not only at the individual level (e.g., Dunlosky & Rawson, 2012; Dunlosky et al., 2013; Sodian & Frith, 2008) but also at the group level (Heyes et al. 2020; Shea et al., 2014). Societies and cultures where individuals reliably report to others their own confidence and subsequently use it in their own decisions and behaviors are likely to be more successful and more efficient in their cooperative endeavors (e.g., Bahrami et al., 2010). While metacognitive monitoring and control underpin the emergence and evolution of every culture and are involved in all kinds of learning and decision-making, the ways in which individual metacognitive monitoring and control are scaffolded by social structures and practices early on in development promotes sensitivity to specific metacognitive cues and to associated epistemic norms (Heyes et al., 2020; Proust & Fortier, 2018).
Although we found culturally variable as well as similar aspects of metacognition in a perceptual memory task, future studies should address whether this finding generalizes to other types of tasks. There is no reason to expect that the development of metacognitive abilities should be aligned across cognitive tasks in a given culture. In fact, even among the same cultural population, developmental studies document diferent ages for the emergence of various metacognitive abilities. For example, in a memory task, verbal assessment of confidence in one's own memory is observed in children as young as 4 years old (Hembacher & Ghetti, 2014), whereas in a perceptual discrimination task, 3.5 -year-olds are able to opt-out from providing responses when they are uncertain (Bernard et al., 2015; Lyons & Ghetti, 2013).
Likewise, there is no a priori reason to expect that these cultural differences and similarities would generalize to other populations. In adult studies, for example, reasoning and attention styles of Central and East Europeans are more similar to those of East Asians than West Europeans (Var- num et al., 2008); similarly, South Italians are more similar to East Asians than North Italians (Knight & Nisbett, 2007). Developmentally, we may also find a diferent pattern of metacognitive abilities in other ethnic populations (e.g., Kim et al., in press). Depending on the social and cultural emphasis on specific epistemic practices and attentional guidance, metacognitive abilities might sensibly differ across cultures. Engaging in investigations of diverse human populations, therefore, will promote a better understanding of the social dimensions of human cognition as well as of human metacognition.
Supplementary Information.
Acknowledgments We thank all the families and children who participated in the present research. The research was supported by an ERC advanced grant (#269616) and ANR-17-EURE-0017 FrontCog to JP and Japanese Ministry of Education, Culture, Sports, Science, and Technology grant (#16H06301) to SI. Part of the data was presented at Budapest CEU Conference on Cognitive Development (2017), Budapest, Hungary.
