Executive functions have become one of the most intensively studied targets in developmental science, and for good reason. These capacities, which allow children to resist impulses, hold information in mind, and shift flexibly between rules or perspectives, develop rapidly during the preschool years and robustly predict later cognitive, academic, and socioemotional outcomes. Longitudinal work has linked early executive function skills to achievement in mathematics and literacy, to emotion regulation, and even to social acceptance among peers. Because these abilities are so consequential, researchers have grown increasingly concerned about the possibility that a dominant feature of modern childhood, screen media, might interfere with their development during this sensitive window.

The scale of young children’s screen exposure makes the question urgent. Surveys such as the Common Sense Media census indicate that media use among children from birth to age eight has risen steadily, and studies tracking the COVID-19 pandemic documented sharp increases in screen time as families locked down and moved learning and entertainment indoors. The World Health Organization has issued guidelines recommending strict limits on sedentary screen time for children under five, and a large body of observational research has associated heavy screen use with lower developmental screening scores, attentional difficulties, and weaker self-regulation. Yet observational studies can only show correlations, leaving open the possibility that family circumstances, parenting practices, or child temperament drive both media habits and cognitive outcomes.

Experimental studies offer a sharper tool. By randomly assigning children to watch different types of content and then immediately testing their executive function performance, researchers can isolate the causal effect of the content itself. Earlier influential work, including a widely cited 2011 study by Angeline Lillard and Jennifer Peterson published in Pediatrics, found that children who watched a fast-paced fantastical cartoon performed worse on subsequent executive function tasks than children who watched slower-paced educational programming or drew pictures. That finding, and replications and extensions that followed, suggested that the architecture of a video, its pacing, its reliance on fantasy, its interactivity, might be capable of temporarily degrading young children’s self-control.

What remained unclear, Cai and Xiang argue, was how these different content characteristics relate to distinct executive function domains. Previous reviews tended to focus on screen time duration, on single components of executive function, or on one content feature at a time. To fill that gap, the researchers followed the PRISMA 2020 reporting guidelines, searching the PsycINFO, Scopus, and Web of Science databases for peer-reviewed experimental studies published between 2000 and 2025. Their search identified 13 articles reporting 19 independent experimental studies, a small but conceptually focused evidence base that allowed them to compare how educational intent, interactivity, pacing, and fantastical elements each relate to immediate executive function performance in early childhood.

The pattern that emerged across this heterogeneous literature was consistent for the riskier content features. Studies examining highly fantastical or fast-paced material generally reported poorer performance on some immediate executive function measures, with inhibitory control emerging as the domain most frequently affected. Inhibitory control, the capacity to suppress a dominant or automatic response, is precisely the skill taxed by tasks in which children must wait, resist touching an attractive object, or follow a rule that conflicts with their first instinct. Some of the reviewed studies went beyond behavioral measures, using electroencephalography to record N2 and P3 event-related potential components during go/no-go tasks, providing neural correlates of how video editing pace may alter children’s inhibitory processing shortly after exposure.

Theoretical frameworks help explain why such effects might occur. According to the limited capacity model of mediated message processing, human attentional and working memory resources are finite, and rapidly changing, overloaded media environments can consume processing capacity in ways that leave fewer resources available for subsequent self-regulation. Fantastical content, meanwhile, may violate young children’s expectations about how the world works, requiring additional cognitive effort to parse and potentially disrupting the scripts children rely on to guide behavior. A complementary account suggests that arousal and priming mechanisms are at work: exciting, unpredictable content may leave children in a heightened state that interferes with the calm, effortful control that executive function tasks demand.

Not all content effects pointed in the same direction, however. Two studies of interactive content reported more favorable outcomes on selected measures, particularly working memory. Interactivity, in which children respond to, manipulate, or make choices within the medium, may keep children actively engaged rather than passively absorbing stimulation, and could recruit rather than deplete the very memory systems that executive function depends on. The evidence for educational content was more limited and less encouraging than many parents might hope. Educational labeling alone was not consistently associated with better immediate executive function outcomes when pacing and fantastical elements were not controlled. In other words, an app or program marketed as educational does not automatically confer cognitive benefits if its underlying structure remains fast, frenetic, or heavily fantastical.

The authors are careful to emphasize the limits of the available evidence. The 19 studies included in the review employed varied samples, stimuli, exposure durations, and outcome measures, making precise meta-analytic synthesis difficult, and the overall evidence base remains small. The effects documented are immediate and short-term, measured minutes after exposure, and the review does not establish that any single viewing session causes lasting harm to children’s developing executive systems. The authors call for future research using larger and more diverse samples and ecologically valid designs that better reflect how children actually use media in their homes, across devices and over extended periods.

Even with those caveats, the review carries a practical message for parents, educators, and app designers at a moment when screens are woven into early childhood across the globe. The findings suggest that caregivers worried about screens should look beyond the clock and scrutinize the content itself, favoring slower-paced, realistic, and genuinely interactive material over rapid-fire fantastical entertainment, and treating educational labels as insufficient guarantees. For policymakers updating screen time guidance and for developers designing children’s media, the work provides an evidence-based rationale for considering pacing and fantasy content when evaluating what young children consume. As Cai and Xiang conclude, the characteristics of screen content appear to be genuinely relevant to children’s immediate executive function performance, and understanding those characteristics may prove more productive than counting minutes alone.

Subject of Research: The effects of screen content features on executive functions in early childhood

Article Title: Screen Exposure and Executive Functions in Early Childhood: A Systematic Review of Experimental Studies

Article References: Cai, X., & Xiang, J. (2026). Screen Exposure and Executive Functions in Early Childhood: A Systematic Review of Experimental Studies. International Journal of Early Childhood. https://doi.org/10.1007/s13158-026-00547-4

Image Credits: AI Generated

DOI: 10.1007/s13158-026-00547-4

Keywords: screen exposure, digital media, executive functions, early childhood, inhibitory control, working memory, cognitive flexibility, fantastical content, pacing, interactive media, systematic review, preschoolers

The stakes of the question are hard to overstate. Executive function skills develop rapidly during childhood, and performance on executive function assessments in early life predicts academic achievement, social competence and behavioral adjustment years later. Meta-analyses cited by the authors link early executive function to reading and science outcomes, to long-term academic attainment and even to reduced risk of psychopathology. Because these skills correlate with so many positive life outcomes, they have become one of the most attractive targets for intervention in developmental science, education policy and public health. Programs ranging from preschool curricula to computerized brain training have been launched on the premise that executive function is highly malleable. Yet the review notes that the experimental evidence for far transfer, meaning improvements that carry over from trained tasks to real-world outcomes, remains weak, and meta-analytic work on cognitive training generally concludes that gains rarely generalize beyond the trained tasks themselves.

The authors begin by examining how executive function is defined and measured, and they identify this as a foundational problem. In the classic latent-variable framework, executive function decomposes into inhibition, working memory updating and shifting, with a common underlying factor. But in young children, tasks designed to tap these components often fail to show the clean separations seen in adults, and performance depends heavily on how familiar children are with task demands, labels and materials. Studies showing that familiar labels help children engage proactive control, or that multidimensional reasoning boosts performance on the dimensional change card sort task, suggest that what looks like a general executive capacity is often knowledge and context bound. The review argues that executive function is best understood as a set of skills that are shaped by experience and deployed in specific situations, rather than as a fixed, domain-general cognitive muscle that any training program can strengthen.

Measurement is where this conceptual ambiguity becomes a practical crisis. The most common assessments, including conflict tasks such as the flanker and day-night paradigms, card sorting tasks and delayed gratification measures, were largely developed in Western, educated, industrialized, rich and democratic populations. Cross-cultural work reveals striking variability: cognitive flexibility patterns differ across cultural settings, and studies in Jordan, Kenya, Brazil, South Africa and The Gambia show that task performance and its environmental predictors do not map neatly onto the models built in North American and European samples. Parent and teacher rating scales, such as the Behavior Rating Inventory of Executive Function, capture yet another construct, one that correlates imperfectly with laboratory performance measures. The authors highlight emerging alternatives, including group-based classroom assessments and observational measures that embed executive function demands in real-world activities, which may better capture how regulation skills operate where children actually live and learn.

Turning to socioeconomic status, the review acknowledges one of the most consistent findings in the literature: children from lower socioeconomic backgrounds tend to score lower on executive function assessments, an association confirmed by meta-analysis. But the authors are emphatic that this correlation is causally ambiguous and often context-dependent. Socioeconomic status is a composite construct encompassing income, education, occupation and neighborhood resources, and critics cited in the review argue that treating it as a unitary variable obscures the specific mechanisms at work. Longitudinal studies point to candidate mediators, including cognitive stimulation, language development and environmental predictability, with neuroimaging work suggesting that cognitive stimulation is linked to neural function supporting working memory. Genetic confounding also looms large, as studies of maternal education and prenatal smoking show that inherited factors account for a substantial share of the apparent environmental effects. The review warns against deficit framing, urging researchers to consider how children’s skills may represent adaptations to the specific environments they inhabit.

The home environment and parenting emerge as a second major experiential domain, and one where the evidence is similarly suggestive but rarely decisive. Household chaos, characterized by noise, crowding and unpredictable routines, is associated with poorer executive function, an effect documented in meta-analysis and partially buffered by high-quality childcare. Home literacy environments, parental scaffolding, autonomy support and attachment security all show positive associations with children’s self-regulation in numerous studies spanning the United States, China, Chile, Korea and Côte d’Ivoire. Experimental work adds encouraging signal: an experimental study found that autonomy-supportive interactions improved preschoolers’ self-regulation, and a randomized clinical trial showed that an early parenting intervention accelerated inhibitory control development among children involved with child protective services. Still, the review stresses that most of this evidence is correlational, that effect sizes are typically modest, and that gene-environment correlation means children both shape and are shaped by their families in ways that standard designs cannot untangle.

Schooling, the third focal domain, offers some of the strongest quasi-experimental evidence that experience shapes executive function. School cutoff designs, which compare children born just before and just after enrollment deadlines, indicate that a year of schooling improves cognitive control and even alters associated patterns of brain activation. Differential growth in working memory across school-year and summer months suggests that classrooms actively promote executive function development rather than merely tracking maturation. The quality of teacher-child interactions matters as well, with meta-analytic evidence linking classroom interaction quality to children’s executive function gains. Curricular interventions tell a more complicated story. Programs such as Tools of the Mind generated early enthusiasm, but large rigorous evaluations have produced mixed results, while games-based approaches such as Red Light, Purple Light have shown benefits for school readiness in some low-income samples, including trials in Kenya. The Chicago School Readiness Project stands out for demonstrating longer-term impacts on behavioral regulation that persisted into late adolescence.

Why, then, has the intervention literature so often fallen short of its promise? The review offers a synthesis: interventions have typically treated executive function as a generic capacity to be exercised like a muscle, rather than asking what specific experiences, in specific contexts, help specific children regulate their behavior toward specific goals. The authors draw on a growing contextual perspective in developmental science, one that recognizes culture as constitutive rather than incidental. Culturally organized practices such as autonomy and helping, Indigenous frameworks of connectedness and culturally meaningful forms of self-regulation all suggest that the skills valued and cultivated in one community may differ from those assumed by standardized assessments and imported curricula. Ethical concerns raised by anthropologists about parenting interventions exported to low- and middle-income countries reinforce the point that interventions must be grounded in local meanings, values and strengths rather than framed around supposed deficits.

The review closes by outlining four key directions for future work. First, researchers need better conceptualizations of environmental quality and experience, moving beyond coarse socioeconomic categories to measure the specific features of environments, such as cognitive stimulation, predictability and stress, that plausibly shape developing regulation skills. Second, the field must improve the measurement of executive function itself, developing contextually grounded assessments that are validated across cultural and linguistic groups and that capture regulation as it unfolds in classrooms, homes and everyday activities. Third, studies must be designed to support stronger causal inference, leveraging natural experiments, randomized designs and genetically informed methods while remaining ecologically valid. Fourth, the authors call for greater attention to diversity and equity in who is studied, how findings are interpreted and who benefits from the resulting interventions, including genuine partnerships with communities in majority-world settings.

For a field with such high public stakes, the message of this review is both sobering and generative. Executive function skills matter enormously, they are demonstrably linked to experience, and yet the science of exactly how experience builds them remains incomplete in ways that have limited the success of interventions designed to improve children’s life chances. By demanding sharper concepts, better measures and culturally informed designs, the authors are not dismissing decades of work but redirecting it toward the questions that matter most. If the next generation of research can specify how experiences get under the skin to strengthen children’s regulation of thought and action, the promise of executive function science, from closing achievement gaps to designing smarter educational policies, may finally be kept.

Subject of Research: How experience such as socioeconomic status, parenting and schooling shapes the development of childhood executive function skills

Article Title: Understanding how experience supports the development of executive function skills

Article References: Doebel, S., Chevalier, N., Lipina, S. J., Rabii, V., & Ahmed, S. F. (2026). Understanding how experience supports the development of executive function skills. Nature Reviews Psychology. https://doi.org/10.1038/s44159-026-00614-6

Image Credits: AI Generated

DOI: 10.1038/s44159-026-00614-6

Keywords: executive function, child development, socioeconomic status, parenting, home environment, schooling, cognitive control, self-regulation, working memory, inhibitory control, cognitive flexibility, intervention

For years, a dominant narrative in autism research has held that autistic people respond more cautiously, demanding more evidence before committing to a decision. That idea grew out of studies applying diffusion decision models, a class of cognitive models that decompose two-choice task performance into distinct components. The models describe noisy sensory evidence accumulating over time from a starting point toward one of two decision boundaries. The key parameters are drift rate, which reflects how efficiently task-relevant information is extracted; boundary separation, which captures response caution or how much evidence is required before responding; the starting point, which encodes any preexisting bias toward one option; and non-decision time, which covers sensory encoding and motor execution outside the decision itself.

Previous findings using this framework had been strikingly inconsistent. Several studies, including work by Pirrone and colleagues with adults and with children and adolescents, reported wider boundary separation in autistic participants during visual orientation tasks, suggesting a general preference for certainty. Similar patterns appeared in implicit association tasks, go/no-go tasks, numerical addition and flanker tasks. Yet other studies found no such difference, and at least one study of autistic adolescents during face processing reported narrower boundaries instead. The research team, led by Lou Thomas and Catherine Manning, set out to resolve this confusion by systematically testing three candidate explanations: that group differences depend on the task, that they depend on the modelling approach, or that they hinge on the instructions given to participants.

The answer came from two large, carefully matched studies involving 50 autistic and 50 non-autistic children aged 6 to 14 years. In the first study, children played a child-friendly game called InsectLand in which they judged the tilt of striped Gabor patches relative to a vertical reference line, distinguishing easy trials tilted 1.3 degrees from vertical from hard trials tilted only 0.6 degrees. Crucially, no explicit instructions about speed or accuracy were given. In the second study, children judged the direction of coherent motion among randomly moving dots, first under instructions to respond as accurately as possible and then under instructions to respond as quickly as possible, in a counterbalanced order.

The modelling was deliberately rigorous. Analyses were conducted by team members blind to group membership, who made all decisions about outlier removal and prior selection on a dataset in which diagnostic status and trait scores had been randomly permuted. The team fitted Bayesian hierarchical diffusion models, which preserve uncertainty at both the participant and group level, and followed up with conventional two-step non-hierarchical fits, extracting individual parameter estimates before running group statistics, mirroring the approach of earlier studies that reported group differences. Across both tasks, both modelling approaches, and all model variants that controlled for age and intelligence, modelled drift rate variability, or accounted for contaminant responses, the results were the same: no conclusive evidence of group differences in drift rate, boundary separation, starting point or non-decision time.

The headline result emerged from the motion task. When told to emphasise accuracy, all children widened their decision boundaries and lengthened their non-decision times; when told to emphasise speed, they narrowed boundaries and shortened non-decision times. The evidence for this within-participant modulation was overwhelming, with Bayes factors exceeding 100, and the effect appeared equally in the autistic group alone. Drift rates, by contrast, were untouched by instructions, exactly as theory predicts. This is the first demonstration that autistic children can strategically modulate the speed-accuracy trade-off in perceptual tasks on the basis of explicit instruction, and it directly challenges theoretical accounts that portray autistic cognition as inflexible, including executive dysfunction frameworks and predictive coding proposals of rigidly high precision weights.

The findings also complicate the popular idea that autistic people are uniformly more cautious decision-makers. Even without any speed or accuracy instructions in the orientation study, autistic children did not show the wider boundary separation previously reported. The researchers suggest that increased caution may emerge only under specific real-world conditions. A consultation with five autistic community members after data collection suggested that caution might depend on the number of choice options, the stakes of the decision, and contextual factors, and a recent narrative review similarly proposes that group differences in decision-making arise mainly in complex metacognitive and value-based tasks rather than simple perceptual ones.

Beyond the group comparisons, the team explored how individual differences relate to decision-making parameters, examining ADHD traits, sensory processing, coordination skills and reading ability. A few relationships surfaced, though they were task-specific and sensitive to modelling choices. In the motion task, children with stronger sight-word reading efficiency on the Test of Word Reading Efficiency showed narrower boundary separation and shorter non-decision times, echoing previous links between reading and motion processing consistent with magnocellular accounts of dyslexia. In the orientation task, sensory underresponsivity was negatively related to non-decision time once age and performance IQ were controlled. The hypothesised relationship between hyperactivity and impulsivity traits and drift rates in autistic children, based on earlier work, was not supported; in fact, there was conclusive evidence against it.

Perhaps the most sobering implication concerns methodology. The team found that results shifted between conclusive and inconclusive depending on whether hierarchical or two-step modelling was used, and depending on whether age and performance IQ were partialled out. Two-step approaches that ignore participant-level uncertainty may exaggerate evidence in either direction. The authors also note that many earlier studies failed to report the speed-accuracy instructions given to participants at all, an omission this work shows to be consequential. They call for future studies to use larger samples, preregistered and blinded analysis pipelines, clear reporting of task instructions, and shared experimental code, recommendations the current studies themselves modelled through open data on the UK Data Service and materials on the Open Science Framework.

For autistic children and their families, the practical message is quietly powerful: when instructions are clear, autistic children calibrate their decisions as responsively as anyone else. Cognitive flexibility, so often framed as a core deficit, appears intact in this domain, suggesting that differences reported in earlier studies may reflect experimental design rather than fundamental differences in how autistic minds weigh evidence, set thresholds and commit to a choice.

The diffusion decision model has a long pedigree in cognitive psychology, tracing back to work by Ratcliff and colleagues in the late twentieth century that formalised how noisy evidence accumulates toward a response threshold. Its appeal lies in separating what a simple accuracy score or average reaction time conflates: a slow response could reflect cautious threshold-setting, sluggish sensory encoding, or weak perceptual evidence, and these possibilities carry very different theoretical implications. Applying the model to children adds further complexity, since parameters such as boundary separation and drift rate change systematically over development, which is why the current studies controlled for age in several of their model variants.

The broader scientific backdrop also helps explain why the task choice mattered so much to the researchers. Psychophysical studies of autism have long produced a mixed picture, with some reporting enhanced orientation discrimination thresholds consistent with theories of heightened local perceptual processing, and others reporting reduced sensitivity to coherent motion, in line with proposals about dorsal stream vulnerability. Because orientation judgments and motion coherence judgments arguably tap different visual pathways, comparing the two tasks within a single modelling framework offered a way to test whether decision-making differences, if they existed, would follow these task-specific predictions. They did not, which itself is informative about the generality of any perceptual decision-making differences in autism.

Sensory processing differences also carry diagnostic weight in their own right. The current edition of the International Classification of Diseases defines autism partly through persistent hypersensitivity or hyposensitivity to sensory stimuli, making the observed link between sensory underresponsivity and non-decision time in the orientation task a potentially meaningful bridge between questionnaire-based sensory profiles and the latent stages of a decision model. Non-decision time encompasses processes such as sensory encoding and motor execution, so a relationship with underresponsivity invites speculation about how atypical early sensory processing might feed into the timing of overt responses, though the authors are careful to note the association was task-specific and modest.

The reading findings add another dimension. Links between reading ability and motion processing have been debated for decades under magnocellular accounts of dyslexia, and the observation that children with more efficient sight-word reading showed narrower boundaries and shorter non-decision times in the motion task is consistent with that tradition, even though the study was not designed as a test of reading theory. More broadly, the pattern of trait correlations underscores the authors’ argument that autism rarely presents in isolation: attentional traits, literacy, sensory processing and motor coordination all vary across children and may shape decision-making parameters in ways that a binary diagnostic comparison can obscure.

Finally, the open science infrastructure surrounding the work deserves note. De-identified data are archived with the UK Data Service, and experimental code, analysis scripts and preregistrations are publicly available, allowing other teams to reanalyse the same datasets under alternative modelling assumptions, which is precisely the kind of cross-lab scrutiny the authors argue the field needs.

Subject of Research: Perceptual decision-making and speed-accuracy adjustment in autistic and non-autistic children using diffusion decision models

Article Title: Autistic and non-autistic children’s perceptual decision-making in visual orientation and motion tasks and the effect of task instructions

Article References: Thomas, L., Scerif, G., Yusuf, H., Laird, M., Taylor, G., Evans, N. J., & Manning, C. (2026). Autistic and non-autistic children’s perceptual decision-making in visual orientation and motion tasks and the effect of task instructions. Attention, Perception, & Psychophysics, 88(7), Article 188. https://doi.org/10.3758/s13414-026-03330-8

Image Credits: AI Generated

DOI: 10.3758/s13414-026-03330-8

Keywords: autism, perceptual decision-making, diffusion decision model, speed-accuracy trade-off, visual orientation, motion coherence, Bayesian hierarchical modeling, cognitive flexibility, ADHD traits, response caution, drift rate, boundary separation

The finding challenges the familiar view of the prefrontal cortex as an all-purpose engine of intelligence and adaptability. In humans, the region is associated with working memory, planning, decision-making, emotional regulation and cognitive flexibility. But the new research suggests that its influence can be context-dependent. Rather than always promoting flexible behavior, the medial prefrontal cortex may sometimes stabilize decisions based on previous experience, making it harder to respond to information arriving in the present moment.

The researchers studied a natural behavior in female mice: retrieving displaced pups and returning them to the nest. In the experiment, an adult female was placed at the base of a T-shaped maze while an artificial sound played from either the right or left arm. The sound acted as a beacon indicating where a pup would be placed. A mouse that followed the sound could reach the pup directly, but the animals initially relied on a different strategy. They returned to the maze arm where they had found a pup during the previous trial, regardless of where the sound was coming from.

This behavior is known as a win-stay strategy. It is often useful because repeating a successful action can conserve time and energy, particularly when conditions remain stable. However, it becomes inefficient when the environment changes. Over repeated trials, the mice gradually learned that the sound was a more reliable guide than memory of the previous pup location. Half of the 12 animals adopted the sound-based strategy by the fourth day of training, and all of them were using the auditory cue by the eighth day.

The experiment enabled the researchers to compare activity in two brain regions involved in the task: the auditory cortex, which processes sound, and the medial prefrontal cortex, which is involved in decision-making and behavioral control. The animals were implanted with silicon probes that recorded the firing of individual neurons while they navigated the maze. These recordings allowed the team to examine how neural circuits responded as the mice shifted from a learned habit to a strategy based on an external sensory signal.

The researchers then used chemogenetics to silence each region separately. This technique uses engineered receptors that can be activated by a specially selected drug, allowing scientists to reduce activity in targeted neurons without broadly disrupting the rest of the brain. When the auditory cortex was silenced, the mice showed impaired sound learning, although the ability was not completely eliminated. Animals that failed to form a strong sound association continued to depend on the win-stay strategy even after eight days.

The result was dramatically different when the medial prefrontal cortex was silenced. Instead of becoming confused or making random choices, most of the mice learned to follow the sound in only two or three days. In other words, disabling a region linked to executive control accelerated the adoption of a more efficient strategy. When the researchers restored medial prefrontal activity and repeated the task, the animals returned to their original preference for the familiar win-stay approach.

The findings indicate that the medial prefrontal cortex was not simply helping the mice make decisions. It was actively supporting a decision rule based on prior success, creating competition with the auditory system’s representation of the current cue. The researchers propose that this neural competition may explain why a behavior that is initially useful can become resistant to change. A circuit that emphasizes past outcomes can suppress the influence of new information, even when that information offers a faster route to the goal.

The study may offer a new perspective on human behavior, including conditions in which people have difficulty shifting attention, abandoning routines or responding to changing circumstances. The authors emphasize that the mouse results cannot be directly equated with human neurodiversity or cognitive disorders, but they may help identify mechanisms that contribute to differences in executive function. The Emory team is now studying genetically modified mice carrying markers associated with autism and is working with collaborators to test related ideas in adults using non-invasive techniques such as transcranial magnetic stimulation. The long-term goal is to determine whether carefully regulating prefrontal activity could help people overcome maladaptive habits while improving their ability to use relevant external cues.

Subject of Research: Animals

Article Title: Neural competition between prefrontal and auditory cortex constrains novel sound strategy learning

News Publication Date: 7-Aug-2026

Web References: https://doi.org/10.1126/sciadv.aeb3005

References: Science Advances, DOI: 10.1126/sciadv.aeb3005

Keywords: medial prefrontal cortex, auditory cortex, cognitive flexibility, behavioral neuroscience, sound learning, win-stay strategy, chemogenetics, neural competition, mice, executive function

Compositionality is most obvious in language. A limited vocabulary can generate an effectively unlimited number of sentences because words are assembled according to grammatical rules. The meaning of a sentence depends not only on the words it contains, but also on how those words are combined. This ability allows people to understand entirely new statements, follow unfamiliar instructions and apply old knowledge to new situations. For decades, researchers have treated this systematic flexibility as a defining feature of human cognition and a central ingredient of general intelligence.

The new analysis questions whether compositional thought requires an explicit symbolic architecture. Traditional theories often describe intelligent reasoning as the manipulation of discrete symbols according to formal rules, much like operations performed by a computer program. Under that view, a system must represent objects, concepts or actions as separate symbolic units before it can recombine them. But modern artificial intelligence has complicated this picture. Large language models, trained on enormous datasets and built from neural networks, can produce strikingly novel combinations without being given an explicit grammar or a manually programmed symbolic system.

These models do not simply retrieve sentences from memory. Their behavior suggests that statistical learning across vast numbers of examples can create internal representations that support generalization. A language model may respond appropriately to a combination of words, concepts or instructions that it has never encountered in precisely that form. The researchers emphasize that this does not settle the question of how compositionality works, but it raises a crucial possibility: some compositional behavior may emerge from scale, learning and distributed representations rather than from clearly identifiable symbolic components.

Neuroscience is revealing comparable complexity in animal brains. Studies of animals performing tasks that require flexible reasoning have found evidence for compositional neural codes. In such codes, separate features of a situation—such as an object’s identity, its location, an action or a goal—can be represented in ways that allow them to be recombined when circumstances change. An animal that has learned what an object is and where an action is useful may be able to apply that knowledge to a new combination, even without having experienced the exact situation before.

The technical challenge is determining how neural circuits achieve this flexibility. Biological neurons rarely function as isolated symbols. Instead, information is encoded through patterns of activity distributed across populations of cells. A concept may be represented by the coordinated firing of many neurons, with the same neural population participating in multiple tasks. Compositional computation can therefore arise when these activity patterns are organized so that particular features remain sufficiently stable while other features can be recombined. The result is a neural system that can preserve structure without storing every possible combination separately.

The authors propose studying these mechanisms through a “compositionality continuum,” defined by two interacting properties: the expressivity of computation-specific building blocks and the complexity of the rules used to recombine them. At one end, a system might rely on highly specialized components combined through simple operations. At the other, it might use broad, flexible representations whose interactions are learned through complex distributed dynamics. Between these extremes lies a wide range of possible biological and artificial implementations.

This framework could reshape how researchers compare brains with machine-learning systems. Rather than asking whether an animal, neural network or language model is compositional, scientists could ask what kind of compositional mechanism it uses. Does the system contain reusable representations? Can it apply learned relationships to unfamiliar combinations? Are its recombination rules explicit, implicit or distributed across many units? And how much performance depends on the architecture itself compared with the quantity and diversity of training data?

Answering those questions will require experiments that connect behavior to neural computation. High-density recordings from animals engaged in compositional tasks can reveal how populations of neurons represent individual elements and how those representations change when elements are combined. At the same time, reverse engineering artificial neural networks can identify the internal circuits and activity patterns responsible for flexible behavior. Comparing the two may show whether similar computational principles appear in systems built from biological neurons and systems built from mathematical units.

The debate has implications far beyond language or artificial intelligence. If general intelligence depends on explicit symbolic compositionality, researchers may need to design machines with more structured internal operations. If sophisticated compositional behavior can emerge from large-scale learning and distributed neural dynamics, then increasing model capacity and improving experience may be sufficient to produce abilities once thought to require symbolic reasoning. The proposed continuum does not choose between these possibilities. Instead, it offers a way to measure them, turning a philosophical question about intelligence into a testable problem in neuroscience and machine learning.

Subject of Research: The neural and computational mechanisms underlying compositionality in biological brains and artificial intelligence systems.

Article Title: The compositionality continuum as a principle for studying the neural basis of intelligence

Article References: Riveland, R., Pouget, A. & Driscoll, L. The compositionality continuum as a principle for studying the neural basis of intelligence. Nat Neurosci (2026). https://doi.org/10.1038/s41593-026-02382-1

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41593-026-02382-1

Keywords: compositionality, intelligence, neuroscience, neural networks, artificial intelligence, large language models, neural codes, cognition, symbolic reasoning, general intelligence

Transitive inference, the cognitive ability to deduce a relationship between two items based on their relation to a third item, has long been considered a hallmark of higher-order reasoning in both humans and animals. However, prior models often assumed symmetric learning processes, wherein the strength of learned associations adjusted uniformly across different relational directions. Graham and Spitzer challenge this assumption, presenting compelling evidence that learning is inherently asymmetric and tailored to environmental contingencies.

Central to their investigation is the concept of relational structure—how items or concepts are arranged in hierarchical or networked frameworks that dictate learning pathways. The researchers designed sophisticated experimental paradigms to manipulate relational structures dynamically, allowing them to observe how subjects adjust their inference strategies when these underlying structures change unpredictably. This approach brings to light the brain’s remarkable adaptability and the selective updating of specific relational links over others.

The findings reveal that asymmetric learning is not a flaw or limitation but a highly adaptive strategy. When confronted with altered relational structures, the cognitive system preferentially enhances or suppresses specific directional associations to maintain coherent inference chains. This selective plasticity ensures that learned information remains relevant and minimizes interference from outdated knowledge, thereby optimizing behavioral responses.

Underpinning this discovery is a sophisticated computational model that integrates asymmetric learning rates across relational dimensions. The model quantitatively captures how individuals weigh evidence differently depending on whether it supports upward or downward inference along a hierarchy. It further accounts for the variable impact of new information on strengthening or weakening existing associations, reflecting a dynamic balance between stability and flexibility in knowledge representations.

To empirically validate their theoretical framework, the authors employed a combination of behavioral testing and rigorous statistical analyses. Participants engaged in tasks requiring transitive inference across shifting relational networks, with performance metrics illustrating rapid adjustments aligned with model predictions. Notably, the degree of asymmetry in learning correlated with enhanced adaptability, suggesting potential avenues for targeted cognitive training or rehabilitation.

Beyond cognitive psychology, the implications of this research ripple through multiple scientific domains. In artificial intelligence, algorithms inspired by asymmetric learning principles could bolster machine learning systems’ capacity to adapt fluidly to evolving data environments. By mimicking human-like inference adaptability, AI agents might better handle tasks involving hierarchical categorization or relational reasoning under uncertainty.

Neuroscientifically, the study invites a reevaluation of neural circuit models supporting learning and plasticity. Preliminary neuroimaging evidence points to differential activation patterns in brain regions implicated in relational processing, such as the prefrontal cortex and hippocampus, when learning asymmetrically. Understanding these neural substrates could illuminate pathways to treating cognitive disorders where inference and adaptability are impaired.

The research also challenges long-standing theoretical paradigms, prompting debates on the nature of logical reasoning and associative learning. Asymmetric learning posits a more nuanced mechanism, where the cognitive system does not merely statically encode relationships but actively prioritizes certain inference trajectories over others based on context. This insight demands a reconsideration of educational strategies to nurture flexible reasoning skills adaptable to complex, real-world scenarios.

Importantly, the study’s innovative methodology—combining theoretical modeling, experimental manipulation, and real-time behavioral observation—sets a new standard for research exploring cognitive adaptability. Such integrative approaches enable researchers to dissect complex mental operations with precision, capturing the dynamic interplay between learning mechanisms and environmental demands.

Graham and Spitzer’s work also raises intriguing questions about individual differences in asymmetric learning capacities. Variability in learners’ adaptability could reflect underlying genetic, developmental, or experiential factors, offering rich territory for future research into personalized cognitive enhancement and neuroplasticity.

Moreover, the concept of asymmetric learning may extend beyond transitive inference into broader realms of decision-making and social cognition. Humans often navigate environments laden with hierarchical and relational complexities, from social hierarchies to conceptual taxonomies, suggesting that asymmetric processing may be a ubiquitous cognitive strategy.

The paper concludes with a call for interdisciplinary collaboration to expand the applicability of asymmetric learning frameworks, proposing cross-talk between psychology, neuroscience, artificial intelligence, and education science. Such synergy promises not only to deepen fundamental knowledge but also to translate findings into practical innovations for technology and health.

As society increasingly depends on adaptive systems—whether human cognition or artificial agents—the importance of understanding how learning asymmetries facilitate flexibility cannot be overstated. This seminal study offers a transformative lens through which to view learning and inference, reshaping the future landscape of cognitive science and beyond.

The results herald a new era where the brain’s selective treatment of relational structures is recognized as a core feature of intelligent behavior. By embracing asymmetry, researchers and practitioners alike can harness the power of adaptability, improving how we learn, reason, and interact with ever-changing worlds.

Subject of Research: Cognitive adaptability and asymmetric learning during transitive inference in changing relational environments.

Article Title: Asymmetric learning and adaptability to changes in relational structure during transitive inference.

Article References:
Graham, T.A., Spitzer, B. Asymmetric learning and adaptability to changes in relational structure during transitive inference. Commun Psychol 3, 155 (2025). https://doi.org/10.1038/s44271-025-00352-0

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s44271-025-00352-0