Do preliterate children spontaneously employ spatial coding for serial order in working memory?

Abstract

The ability to memorize arbitrary sequences contributes to cognitive faculties like language and mathematics. Research suggests that in literate adults, serial order in verbal working memory (WM) is grounded in spatial attention and is mentally organized according to our reading habits, that is, from left-to-right in Western cultures. Currently, it is unknown whether spatialization is a consequence of literacy, or whether the ability already exists early in life but is shaped by literacy in "calibrating" the initial individual differences in the orientation of spatial coding. Here, we investigated the spatial coding of serial order in WM in 5-year-old children who did not yet enter formal literacy education. At the group level, no systematic spatial coding was observed. To investigate whether this absence was due to subjects with reliable but opposing effects, we determined the prevalence of spatial coding at the individual level. This analysis revealed that 36% of the children systematically associated serial order to space, with approximately half of them coding from left-to-right and the rest from right-to-left. These results indicate that a subgroup of preliterate children associate serial order with space and suggest that reading and writing experience calibrates the orientation of spatial coding with reading habits.

Spatialization in working memory: can individuals reverse the cultural direction of their thoughts?

“…This pattern almost completely reversed in the present study where seven (among 29) participants showed a right-to-left SPoARC effect (two of them showed a left-to-right SPoARC). In addition, the pattern we present here was close to that reported in Guida et al, 36 where six (among 19) right-to-left reading participants exhibited a rightto-left SPoARC and two a left-to-right SPoARC (see the bootstrapping analysis run by van Dijck et al 47 ), despite the fact that both experimental setups were different (i.e., in Guida et al, 36 participants had to memorize the color of four squares).…”

“…This is repeated 5000 times. Subsequently, we established confidence intervals based on the range in which mid 80% of the bootstrap slopes were located (we used the same confidence level than van Dijck et al 47 . to allow direct comparison with this study).…”

“…Here, we used the H1 bootstrapping variant, which checks how the slope changes depending on which trials are included for aggregating. This method has recently been applied to the SPoARC effect by van Dijck et al 47 In this method, all of the RTs of each participant for the different item position × hand combinations are randomly sampled (with replacement) and are used to calculate the slope. This is repeated 5000 times.…”

“…This is repeated 5000 times. Subsequently, we established confidence intervals based on the range in which mid 80% of the bootstrap slopes were located (we used the same confidence level than van Dijck et al 47 to allow direct comparison with this study). In the last step, we checked whether such a confidence interval for a given participant contains zero (no evidence for reliable effect) or not (evidence for a reliable effect; see Ref.…”

“…We performed a Fisher's exact test (suitable for a 2 × 2 contingency table with small samples) between the variable experiment and a new variable stemming from the individual analysis, named spatialization direction (modalities: right-to-left spatializer versus left-to-right spatializer), that led to a significant P = 0.002 value, allowing us to argue that we observed significant inversed patterns of spatialization between the two studies. For comparison, a bootstrapping analysis of the results reported by Guida et al 36 (testing rightto-left reading participants) run by van Dijck et al 47 showed that six (among 19) right-to-left reading participants exhibited a right-to-left SPoARC (two a left-to-right SPoARC). If one compares the pattern in Guida et al 36 (six right-to-left spatializers and two right-to-left spatializers) and the data pattern presented in the present paper (seven right-to-left spatializers and two right-to-left spatializers) using a Fisher's exact test, the difference is nonsignificant.…”