Hannah Whitehead

In the Minority World, children are exposed to number symbols early in life (e.g., television, toys, store signage). In these contexts, both symbolic (e.g., 3 vs. 4) and nonsymbolic (e.g., ∵ vs. ∷) magnitude processing, at both behavioural and neural levels, relate to mathematics skills, with stronger associations between symbolic magnitude and mathematics. However, little is known about the neurocognitive development of numerical processing in contexts where children have minimal exposure to number digits prior to formal schooling. We used functional near-infrared spectroscopy (fNIRS) to investigate the behavioural and neural predictors of mathematics development in rural Cote d’Ivoire (N=142, Mage=9.23, SDage= 1.44). Overall, mathematics skills were low, including in the oldest children (11-13) who still struggled with single- and double-digit addition and subtraction. Despite minimal exposure to number digits prior to formal schooling, at the behavioural level, symbolic skills were predictive of mathematics. At the neural level, age-related neural sensitivity to magnitude was related to how children leveraged symbolic skills for mathematics. More specifically, greater magnitude sensitivity in the right angular gyrus in younger, but not older, children was associated with stronger links between symbolic skills and emergent mathematics. Results suggest developmental sensitivity for successfully leveraging symbolic skills for mathematics. In line with a neuroconstructivist framework, our findings emphasize the complex interplay between development, neural processing of magnitudes, and symbolic magnitude skills in supporting mathematics.

Significance: The first three years of life represent a period of heightened neural plasticity, during which early adversity—such as poverty and maternal stress—can shape long-term developmental trajectories. However, little is known about how these adversities are linked to neural network organization in infants and toddlers living in low-resource settings.Aim: This study examined how prenatal exposure to maternal poverty and mental health symptoms relates to neural network organization in early childhood.Approach: We conducted a longitudinal study with 42 children (6–36 months old) from economically-vulnerable households in rural Côte d’Ivoire. Maternal multidimensional poverty and mental health symptoms were assessed around conception and during pregnancy. Two years later, children’s development was measured using the Caregiver Reported Early Development Instruments (CREDI), and resting-state brain activity was recorded using portable functional near-infrared spectroscopy (fNIRS). Graph-theoretic metrics of functional connectivity, including segregation, integration, and small-worldness, were computed to characterize neural network organization.Results: Greater exposure to poverty was associated with reduced network segregation and integration. In contrast, higher maternal mental health symptoms were linked to increased segregation and integration. These findings suggest that deprivation and maternal stress differentially relate to early neural network architecture.Conclusions: Our results align with dimensional models of adversity, and suggest that deprivation (related to poverty) and threat/unpredictability (related to maternal mental health and caregiving) may become biologically embedded. These findings underscore the importance of addressing both structural poverty and maternal mental health during pregnancy and early childhood to promote healthy brain development in high-adversity contexts.

We present qualitative insights from teachers (N=25) following a 2-year implementation of a targeted instruction program, Programme d’Enseignement Ciblé, in rural Côte d’Ivoire. We interviewed teachers with high (above mean) versus low (below mean) student test scores to gain insights into what factors may account for differences in student outcomes. Teachers discussed challenges in program implementation from an administrative perspective (e.g., delayed implementation, lack of teacher-student continuity). Teachers with higher student performance exhibit higher professional agency than those with lower student performance. Four cross-cutting themes emerged among the full sample of teachers: (1) teacher motivation changes over time, (2) teacher agency is low overall, (3) the need for additional training and materials, and (4) pre-existing barriers to children’s learning. The results highlight significant challenges to successful program implementation, which, if addressed, may lead to greater program effectiveness.

Literacy and numeracy are correlated throughout development, however, our understanding of this relation is limited. We explored the predictors of literacy and numeracy covariance (i.e., shared fluency between literacy and numeracy) in children (N = 1167, girls = 563) in rural Côte d'Ivoire, with specific focus on how developmental timing of instruction may relate to covariance. Many Ivorian children experience late enrollment and grade repetition, leading to variation in age-for-grade; participants were between grades 1 to 6, but their ages ranged from 5 to 15 (M = 9.19, SD = 2.07). Phonological awareness, numerical magnitude, ordinality, working memory, and inhibitory control were cognitive predictors of covariance. Age-for-grade was negatively related to covariance suggesting that covariance is related to timing of instruction.

There is an emerging consensus that numerical, executive function (EF), and spatial skills are foundational to children’s mathematical learning and development. Moreover, each skill has been theorized to relate to mathematics for different reasons. Thus, it is possible that each cognitive construct is related to mathematics through distinct pathways. The present study tests this hypothesis. One-hundred and eighty 4- to 9-year-olds (M_age = 6.21) completed a battery of numerical, EF, spatial, and mathematics measures. Factor analyses revealed strong, but separable, relations between children’s numerical, EF, and spatial skills. Moreover, the three-factor model (i.e., modelling numerical, EF, and spatial skills as separate latent variables) fit the data better than a general intelligence (g-factor) model. While EF skills were the only unique predictor of number line performance, spatial skills were the only unique predictor of arithmetic (addition) performance. Additionally, spatial skills were related to the use of more advanced addition strategies (e.g., composition/decomposition and retrieval), which in turn were related to children’s overall arithmetic performance. That is, children’s strategy use fully mediated the relation between spatial skills and arithmetic performance. Taken together, these findings provide new insights into the cognitive foundations of early mathematics, with implications for assessment and instruction moving forward.

Executive Functions (EF; inhibitory control [IC], cognitive flexibility [CF] and working memory [WM]) mediate the relationship between socioeconomic status and reading. However, little is known of the roles of individual EF components in mediating the socioeconomic-reading achievement gap, especially in low- and middle-income countries. In Côte d’Ivoire, children experience many socioeconomic disadvantages (i.e., fewer household resources, maternal illiteracy), and kinship fostering (child in care of extended family while parents pursue economic opportunities elsewhere) is prevalent. This study examines the relation between EF components, socioeconomic risks, and reading among 5th grade children in rural Côte d’Ivoire (N=369). Poorer WM mediated the relationship between higher cumulative socioeconomic risk (poverty, maternal illiteracy, fostering) and lower reading scores. Further, WM fully mediated the negative effects of fostering risk on reading scores. Results suggest that EF components are differentially impacted by environmental socioeconomic risks and play different roles in supporting reading development.

Technology‐based remote research methods are increasingly widespread, including learning assessments in child development and education research. However, little is known about whether technology‐based remote assessments remain as valid and reliable as in‐person assessments. We developed a low‐cost phone‐based language and literacy assessment for primary‐school children in low‐resource communities in rural Côte d'Ivoire using voice calls and SMS. We compared the reliability and validity of this phone‐based assessment to an established in‐person assessment. A total of 685 5th grade children completed language (phonological awareness, vocabulary, language comprehension) and literacy (letter, word, pseudoword, passage reading, and comprehension) tasks in‐person and by phone. Reliability (internal consistency) and predictive validity were high across in‐person and phone‐based tasks. Children's performance across in‐person and phone‐based assessments was moderately to strongly correlated. Phonological awareness and vocabulary skills measured in‐person and by phone significantly predicted in‐person and phone‐based letter, word, and pseudoword reading. Oral language and decoding skills measured in‐person and by phone significantly predicted in‐person and phone‐based passage reading and comprehension. Our phone‐based assessment was a reliable and valid measure of language and reading and feasible for low‐resource settings. Low‐cost technologies offer significant potential to measure children's learning remotely, increasing the inclusion of remote and low‐resource populations in education research.

Literacy and numeracy are correlated throughout development, however our understanding of this relation is limited. We explored the predictors of literacy and numeracy covariance (i.e., shared fluency between literacy and numeracy) in children (N=1167, girls=563) in rural Côte d’Ivoire, with specific focus on how developmental timing of instruction may relate to covariance. Many Ivorian children experience late enrollment and grade repetition, leading to variation in age-for-grade; participants were between grades 1 to 6 but their ages ranged from 5 to 15 (M=9.19, SD=2.07). Phonological awareness, numerical magnitude, ordinality, working memory, and inhibitory control were cognitive predictors of covariance. Age-for-grade was negatively related to covariance suggesting that covariance is related to timing of instruction.