npj Science of Learning

Children exhibit remarkable variability in their mathematical problem-solving abilities, yet the cognitive, metacognitive and affective mechanisms underlying these individual differences remain poorly understood. We developed a novel Bayesian model of arithmetic problem-solving (BMAPS) to uncover the latent processes governing childrens arithmetic strategy choice and efficiency. BMAPS inferred cognitive parameters related to strategy execution and metacognitive parameters related to strategy selection, revealing key mechanisms of adaptive problem solving. BMAPS parameters collectively explained individual differences in problem- solving performance, predicted longitudinal gains in arithmetic fluency and mathematical reasoning, and mediated the effects of anxiety and attitudes on performance. Clustering analyses using BMAPS parameters revealed distinct profiles of strategy use, metacognitive efficiency, and developmental change. By quantifying the fine-grained dynamics of strategy selection and execution and their relation to affective factors and academic outcomes, BMAPS provides new insights into the cognitive and metacognitive underpinnings of childrens mathematical learning. This work advances powerful computational methods for uncovering latent mechanisms of complex cognition in children.

Academic performance relies, in part, on intelligence; however, intelligence quotient (IQ) is limited in predicting academic success. Furthermore, while the search for the biological seat of intelligence predates neuroscience itself, its findings remain conflicting. Here, we assess the interplay between IQ, academic performance, and brain connectivity with behavioral and functional MRI data collected from undergraduate students as they completed an active learning or lecture-based semester-long university physics course. IQ (i.e., full-scale WAIS scores) increased significantly pre-to post-instruction, were associated with physics knowledge and reasoning measures, but were unrelated to overall course grade. IQ was related to brain connectivity during physics-related cognition, but connectivity did not mediate IQs association with task performance. These relations depended on students sex and instructional environment, providing evidence that physics classroom environment and pedagogy may have a gendered influence on students performance. Discussion focuses on opportunities to improve physics reasoning skills for all students.

AbstractResearch has implicated spatial ability as a robust predictor of aptitude, interest, and choice in STEM education and career pursuits. We address three under-explored questions regarding the role of spatial ability in STEM. First, can spatial ability consistently predict STEM success beyond other cognitive skills? Second, what aspects of spatial ability, if any, can predict success in STEM more accurately? Third, to what extent can genetic and environmental factors account for these predictions? We addressed these questions by leveraging data from the Twins Early Development Study (N = 3,936; age range = 16-22) and using 16 tests that assessed three domains of spatial ability: navigation, object manipulation, and visualization. Results show that all three domains are highly predictive of STEM educational outcomes, especially STEM degree choice. These associations persisted after accounting for verbal and general cognitive abilities (g), albeit attenuated. Associations were strongest for tests of object manipulation (e.g., 2D and 3D drawing, pattern assembly and mental rotation). Genetic factors accounted for most of the observed associations between spatial ability and STEM outcomes (62% - 86%) --genetic variance was mostly shared with g ([~] 40%) and, to a lesser extent, verbal ability ([~] 25%). Our findings highlight the potential utility of spatial ability as a specific predictor of success in STEM education and career choice beyond other cognitive abilities. Screening for and training spatial skills is likely useful for identifying potential, fostering talent, and improving outcomes in STEM. Significance StatementLeveraging a comprehensive battery of 16 spatial ability tests across multiple domains, we show that spatial ability has specific utility for predicting success in STEM education. Spatial skills predict success in STEM above and beyond other cognitive abilities, particularly when it comes to STEM engagement and pursuing further STEM education. Thus, spatial skills may be a fruitful target for policymakers, stakeholders, and industries looking to develop interventions, identify and foster talent, and reduce outcome disparities.

Physics is a challenging academic pursuit in which university students regularly struggle to achieve success. Female students tend to perform negatively on introductory physics conceptual assessments compared to their male peers; however, active-learning classroom curricula are known to broadly improve performance on these tests. Here, we used fMRI to delineate physics-related brain activity in 107 students and probed for changes following a semester of active-learning or lecture-based physics instruction. Large-scale reorganization of brain activity accompanying learning occurred in a mixed frontoparietal and default mode network. Sex differences were observed in frontoparietal, default mode, and primary visual areas before and after instruction. Regions showing significant pedagogy, sex, and time interactions were revealed during physics retrieval, suggesting the type of class students complete may influence sex differences in how students retrieve information. These results reveal potentially elucidating sex and pedagogy differences underlying the neural mechanisms supporting physics learning.

We evaluated the cognitive and academic effects of a closed-loop video game delivered in virtual reality (VR) and tablet formats, at two different dosages, in a school-based setting. A total of 158 children aged 8-9 with a range of attention abilities completed 30 training sessions over 10 weeks. Compared to an expectancy-matched control group, both VR and tablet training led to significant improvements in teacher-rated inattention, performance-based attention tasks, and eye-tracking measures. While both VR and tablet versions of the intervention showed benefits on specific attention-related and academic outcomes, the VR version showed select advantages in both regards. Notably, intervention dose did not significantly moderate outcomes, suggesting that efficacy may depend more on reaching a threshold of engagement than on total duration. These findings demonstrate the utility and benefits of using each type of technology to enhance measures of cognitive and academic abilities as part of a regular school curriculum.

Noncognitive skills such as motivation and self-regulation, are partly heritable and predict academic achievement beyond cognitive skills. However, how the relationship between noncognitive skills and academic achievement changes over development is unclear. The current study examined how cognitive and noncognitive skills contribute to academic achievement from ages 7 to 16 in a sample of over 10,000 children from England and Wales. Noncognitive skills were increasingly predictive of academic achievement across development. Twin and polygenic scores analyses found that the contribution of noncognitive genetics to academic achievement became stronger over the school years. Results from within-family analyses indicated that associations with noncognitive genetics could not simply be attributed to confounding by environmental differences between nuclear families and are consistent with a possible role for evocative/active gene-environment correlations. By studying genetic effects through a developmental lens, we provide novel insights into the role of noncognitive skills in academic development.

Musical training has long been argued to boost early phonological and reading abilities. Cortical tracking of speech (CTS) has been proposed as a mechanism for this music-to-literacy transfer. In this study, we examined how musical training shapes CTS in young readers and whether it facilitates literacy benefits. In a sample of 57 children aged 5-9, musical training was linked to enhanced reading and phonological awareness (PA). EEG during story listening revealed that higher left-hemispheric and lower right-hemispheric CTS were also associated with higher reading scores. However, children with higher musicality exhibited stronger reading skills at lower levels of left-hemispheric CTS, suggesting more adult-like speech analysis. Critically, PA mediated the relationship between musicality and reading: greater musicality was associated with stronger PA, which in turn predicted higher reading performance, independent of demographic and cognitive factors. These findings indicate that musical training supports literacy by enhancing PA and shaping left-lateralized speech processing.

Platos Republic, Einsteins Theory of relativity, Vilvadis Four Seasons are all remarkable examples of humans unique ability to create and manipulate complex abstract structures, whether in language, mathematics or music. Yet the mechanisms by which children develop such abstract thinking, and the role of education and structured experiences such as musical practice in shaping these abilities remain unclear. To explore these questions, we conducted cross-sectional behavioral experiments with 566 children aged 4 to 8, spanning four educational grades, half of whom participated in a violin training program since the age of four. Two experiments examined how children encode, process and compress auditory sequences and visual patterns, while a third examined their sensitivity to geometric regularities. Our results reveal the emergence of symbolic reasoning as early as the start of formal schooling, yet with deeper abstraction as a function of grade. By first grade, children encoded complex auditory sequences within a Language of Thought (LoT) similar to adults. Additionally, when confronted with quadrilaterals, children showed increasing sensitivity to geometric regularities, suggesting a developmental transition from perceptual to symbolic reasoning. However, we did not observe significant impact of musical practice on abstraction abilities across any of the domains tested. We discuss whether and how the impact of education and extracurricular activities such as music could be enhanced.

Humans can extract regularities from their environment, enabling them to recognize and predict sequences of events. The process of regularity extraction is called statistical learning and is generally thought to occur rapidly and automatically; that is, regularities are extracted from repeated stimulus presentations, without intent or awareness, as long as the stimuli are attended. We hypothesized that visual statistical learning is not entirely automatic, even when stimuli are attended, and that the learning depends on the extent to which viewers process the relationships between stimuli. To test this, we measured statistical learning performance across seven conditions in which participants (N=774) viewed image sequences. As task instructions across conditions increasingly required participants to attend to relationships between stimuli, their learning performance increased from chance to robust levels. We conclude that the learning observed in visual statistical learning paradigms is, for the most part, not automatic and requires more than passively attending to stimuli.

SignificanceTo learn mathematics, young children require accurate interpretations of mathematics vocabulary. When school language differs from childrens home language, mathematics performance often decreases. Little is known about cortical activation during mathematics vocabulary processing in different languages. This insight will help us to better understand childrens mathematical learning in multilingual societies. Aim and approachWe investigated behavioral and brain responses (fNIRS) of 42 isiZulu and Sesotho (L1) first graders (6.75-7.83 years, 22 girls) who learn mathematics in English (L2) at school when they encounter mathematics vocabulary in L2 compared to L1; and mathematics vocabulary compared to object recognition in L1. ResultsThe results show that higher accuracy in the L1 mathematics vocabulary, as compared to the L2 mathematics vocabulary, comes with the costs of higher cognitive demands in the right superior and middle frontal gyri for first graders. Mathematics vocabulary required longer response time than object recognition and a higher activation in the right superior frontal gyrus. No parietal difference was observed between conditions. ConclusionsFirst graders with no automatization of mathematics vocabulary processing, still demand frontal cognitive resources. This study is a good example of how educational neuroimaging compliments our interpretation of behavioral outcomes and environmental factors such as multilingualism.

While traditionally considered a motor structure, the cerebellum is also involved in cognition. However, the underlying cognitive mechanisms through which the cerebellum contributes to evolutionarily novel cognitive abilities remain poorly understood. Another open question is how this structure contributes to a core unifying mechanism across domains. Motivated by the evolutionary principle of neural reuse, we suggest that a successful account of cerebellar contributions to higher cognitive domains will build on the structures established role in motor behaviors. We conducted a series of neuropsychological experiments, assessing selective impairments in participants with cerebellar ataxia (CA) compared to neurotypicals in solving sequential discrete problems. In three experiments, participants were asked to solve symbolic subtraction, alphabet letter transformation, and novel artificial grammar problems, which were expected or unexpected. The CA group exhibited a disproportionate cost when comparing expected problems to unexpected problems, suggesting that the cerebellum is critical for violation of expectations (VE) across tasks. The CA group impairment was not found either when the complexity of the problem increased or in conditions of uncertainty. Together, these results demonstrate a possible causal role for the human cerebellum in higher cognitive abilities. VE might be a unifying cerebellar-dependent mechanism across motor and cognitive domains. Significance StatementWhile the cerebellum, a phylogenetically ancient brain region, is traditionally viewed as a motor structure, evidence suggests its involvement in cognition. However, the mechanisms by which the cerebellum supports evolutionarily novel cognitive abilities remain poorly understood. In addition, despite theoretical proposals, direct evidence for the cerebellums contribution to a core unifying mechanism across non-motor domains is lacking. Drawing on the principle of neural reuse, we present neuropsychological evidence highlighting the cerebellums causal role in symbolic arithmetic reasoning, alphabet transformation, and grammar problems via violation of expectations processes. The results offer a new perspective on how, rather than merely if, the cerebellum contributes to higher cognition, suggesting a constraint on its role in cognitive domains.

The belief that learning can be modulated by social context is mainly supported by high-level value-based learning studies. However, whether social context can even modulate low-level learning such as visual perceptual learning (VPL) is still unknown. Unlike traditional VPL studies in which participants were trained singly, here we developed a novel dyadic VPL paradigm in which paired participants were trained with the same orientation discrimination task and they could monitor each others performance. We found that the social context (i.e., dyadic training) led to a greater behavioral performance improvement and a faster learning speed, compared with the single training. Interestingly, the facilitating effects could be modulated by the performance difference between paired participants. Functional magnetic resonance imaging (fMRI) results showed that, compared with the single training, social cognition areas including bilateral parietal cortex and dorsolateral prefrontal cortex displayed a different spatial activity pattern and enhanced functional connectivities to early visual cortex during the dyadic training. Furthermore, the dyadic training resulted in more refined orientation representation in primary visual cortex (V1), which was closely associated with the greater behavioral performance improvement. Taken together, we demonstrate that the social context, learning with a partner, can remarkably augment the plasticity of low-level visual information process by means of reshaping the neural activities in early visual cortex and social cognition areas, as well as their functional interplays.

Acquiring sequential information is of utmost importance, e.g., for language acquisition in children. Yet, the long-term storage of statistical learning in children is poorly understood. To address this question, 27 seven-year-olds and 28 young adults completed four sessions of visual sequence learning (Year 1). From this sample, 16 seven-year-olds and 20 young adults participated in another four equivalent sessions after a 12-month-delay (Year 2). The first three sessions of each year used stimulus set-1, while the last session used stimulus set-2 to investigate transfer effects. Each session consisted of alternating learning and test phases in a modified artificial grammar learning task. In Year 1, seven-year-olds and adults learned the regularities and showed transfer to stimulus set-2. Both groups retained their final performance level over the one-year-period. In Year 2, children and adults continued to improve with stimulus set-1, but did not show additional transfer gains. Adults overall outperformed children, but transfer effects were indistinguishable between both groups. The present results suggest that long-term memory traces are formed from repeated sequence learning which can be used to generalize sequence rules to new visual input. However, the present study did not provide evidence for a childhood advantage in learning and remembering sequence rules.

Human cognitive abilities depend on flexible coordination across distributed cognitive control systems within the frontoparietal network, yet the causal architecture linking executive functions (EFs) to higher cognition remains debated. We combined psychometric modeling and neuromodulation to test causal contributions of EFs to distinct cognitive abilities. In a psychometric study, we modeled Unity-Diversity EF components (updating, inhibition, shifting) against four Cattell-Horn-Carroll (CHC) group factors. Fluid reasoning (Gf) was uniquely related to both Common EF and Updating-specific variance, whereas visual processing (Gv) and processing speed (Gs) were primarily linked to Common EF; crystallized ability (Gc) proved to be largely independent of EFs. To causally probe this architecture, in neuromodulatory study we applied anodal transcranial direct current stimulation (tDCS) over prefrontal and parietal hubs of the frontoparietal network. tDCS selectively enhanced working memory updating, with right-hemisphere stimulation improving Gv but reducing Gf performance. Mediation analyses revealed that working memory updating mediated tDCS effects on higher cognition, yet the direction of effects varied across hemispheres and ability domains, exposing indirect, updating-driven facilitative and compensatory mechanisms within the frontoparietal network. Together, these findings bridge psychometric and neuromodulatory approaches to advance mechanistic understanding of how EFs support higher cognition.

Many learning opportunities of mathematical reasoning in school encourage imitative learning procedures (algorithmic reasoning, AR) instead of engaging in more constructive reasoning processes (e.g., creative mathematical reasoning, CMR). Here, we employed a within-subject intervention in the classroom with pupils in upper secondary schools followed by a test situation during brain imaging with fMRI one week later. We hypothesized that learning with CMR compared to AR should lead to a CMR-effect, characterized by better performance and differential brain activity during test. We observed higher brain activity in key regions for mathematical cognition such as left angular gyrus and left inferior frontal gyrus on tasks previously learnt with CMR compared to AR. The effects remained when controlling for individual differences in cognitive abilities, as well as performance and response time differences between the two conditions. Encouraging pupils to engage in constructive processes when learning mathematical reasoning might thus have lasting beneficial effects.

Educational achievement is a key predictor of later-life outcomes, including financial security, social mobility, health, and mortality. Knowing its familial determinants, such as genetic predispositions, is crucial for understanding intergenerational educational mobility and addressing educational inequality. Here, we disentangle direct and indirect genetic effects on educational achievement using trio genome-wide complex trait analysis (Trio-GCTA). Leveraging up to 23,200 genotyped parent-offspring trios from the Norwegian Mother, Father, and Child Cohort Study (MoBa), we find that direct genetic effects explain 15-21% of the variance in educational achievements, including 8th-grade national assessments in Math, Reading, and English, and 10th-grade grade point average (GPA). Notably, indirect genetic effects additionally explain 7-15% of the variance. Both mothers and fathers exert notable influences, exceeding those suggested in previous studies. Moreover, positive gene-environment correlations between parents and offspring show that the intrafamilial environment and childrens genetic predisposition are dependent and mutually reinforcing.

Stored memories are useless unless they are available for retrieval. Thus, investigating different ways to modulate retrieval is crucial. Novelty has been extensively studied as a modulator of memory. In this study, we investigated whether exposure to a novel event, an innovative neuroscience lesson, can enhance memory retrieval and divergent thinking in high school students. Across three experiments, we assessed the timing and mechanisms underlying these effects. In experiment 1, we found that memory retrieval was enhanced when the novel lesson occurred immediately before a memory test, but not when it was presented one hour earlier. In experiment 2, we found that the same immediate novelty exposure improved divergent thinking performance. Finally, in experiment 3, we explored potential shared mechanisms using a competition protocol and revealed that novelty improved divergent thinking regardless of its timing relative to memory retrieval. However, memory retrieval benefited only when tested immediately before the divergent thinking task. These results suggest that novelty boosts both memory retrieval and divergent thinking, but through partially distinct mechanisms. Our findings demonstrate that a simple, real-world classroom intervention can effectively enhance key cognitive functions in students. Significance StatementStored memories are only valuable if they can be retrieved, and memory retrieval plays a key role in creative thinking. Here, we tested whether a simple, novel event, a neuroscience lesson, could enhance memory retrieval and creative thinking in a real-world classroom setting. We found that novelty improved both memory retrieval and divergent thinking, an aspect of creative thinking, when presented immediately before the task. Finally, we revealed a non-reciprocal competition effect between memory retrieval and divergent thinking. These findings highlight a practical, low-cost intervention to boost key cognitive functions in students, demonstrating that brief, well-timed novel experiences can support both learning and creative thinking in educational environments.

Everyday activities often require learning sequences that necessitate the involvement of both the declarative and the procedural memory domains. Previous research has shown that a learning structure that is common across tasks from different domains can improve learning and resistance to interference. However, it remains unknown whether such shared learning structure can enhance longer-term memory retention. To address this question, forty-eight healthy adults participated in a pre-registered study in which they learned both an object sequence task (declarative learning) and a motor sequence task (procedural learning) in two separate sessions separated by 4h. Participants were assigned to either an associated group, where the two tasks shared a common learning structure - that consisted of a specific mapping between finger movements and object categories across learning sessions - or an unassociated group with no such cross-domain shared structure. Memory retention was assessed with a 24h retest session on both tasks. Contrary to our predictions, a shared higher-order structure between tasks from different domains did not enhance memory retention. Exploratory follow-up analyses revealed that the order the tasks were learned (i.e., object or motor first), rather than their structural overlap, influenced performance. Specifically, learning the motor task before the object task impaired the consolidation of the object task irrespective of whether the tasks shared a common learning structure or not. This effect was unidirectional as learning the object task before the motor task had no effect on the consolidation of the motor task. Altogether, the current findings suggest that the order of cross-domain learning experiences rather than their structure influences memory consolidation.

A persons preferred timing of nocturnal sleep (chronotype) has important implications for cognitive performance. Students who prefer to sleep late may have a selective learning disadvantage for morning classes due to inadequate sleep and circadian desynchrony. Here, (1) we tested whether late-type students perform worse only for morning classes, and (2) we investigated factors that may contribute to their poorer academic achievement. Chronotype was determined objectively in 33,645 university students (early, n=3,965; intermediate, n=23,787; late, n=5,893) by analyzing the diurnal distribution of their logins on the universitys Learning Management System (LMS). Late-type students had lower grades than their peers for courses held at all different times of day, and during semesters when they had no morning classes. Actigraphy studies (n=261) confirmed LMS-derived chronotype was associated with students sleep patterns. Nocturnal sleep on school days was shortest in late-type students because they went to bed the latest and woke up early compared with non-school days. Surveys showed that late-type students had lower self-rated health and mood (n=357), and lower metacognitive self-regulation (n=752). Wi-Fi connection data for classrooms (n=17,356) revealed that late-type students had lower lecture attendance than their peers for classes held in both the morning and the afternoon. Our findings suggest that multiple factors converge to impair learning in late-type students. Shifting classes later can improve sleep and circadian synchrony in late-type students but is unlikely to eliminate the performance gap. Interventions that focus on improving students well-being and learning strategies may be important for addressing the late-type academic disadvantage.

Rest and sleep not only strengthen existing memories but also reorganise memories to generate new knowledge that extends beyond direct experience. However, it remains unclear both how memories are reorganised and the effect of this reorganisation on behaviour. Here, we designed a novel protocol to casually manipulate memory consolidation during rest using awake, contextual targeted memory reactivation (TMR). We found that promoting memory consolidation during rest qualitatively reorganises memories by forming shortcuts between memories which have not been experienced together. These shortcuts in memory extend beyond direct experience to facilitate our ability to make novel inferences. A series of control tests indicate that inference performance cannot be explained by quantitative strengthening of the experienced component links but are rather explained by qualitative changes in the cognitive map which involve formation of new shortcuts. Interestingly, we show that representing a shortcut may come with limitations, as shortcuts cannot be readily updated in response to rapid changes in the environment. Together, these findings reveal how memories are reorganised during awake rest to construct a cognitive map of our environment, while highlighting the constraints set by a trade-off between efficient and flexible behaviour.