How children acquire language in just a few years, from babble to grammar, remains one of the most compelling puzzles in cognitive science. Two challenges have long constrained progress: getting the data, that is, capturing language learning as it happens in children’s rich, messy, naturalistic environments, and making sense of it, via manual annotation that is slow, costly, and impractical at scale. Over the past decade, technological advances have begun to reshape both challenges at once. The transformation has been rapid and wide-ranging, spanning the hardware that captures children’s everyday environments, the algorithms that extract meaning from raw signals, and the scientific questions that can now be addressed as a consequence. In this review, we examine the latest work, published between 2024 and 2026, on ML both as a tool, enabling behavioral measurement at scales previously unattainable, and as a model, allowing researchers to instantiate and test explicit hypotheses about learning mechanisms.
Sense of agency is the subjective feeling of being in control over one’s own actions and their sensory outcomes. Here, we propose that a key feature of sense of agency is its link to intrinsic motivation, which is a key driver for engagement with the environment without external reinforcement. We argue that intrinsic motivation is highest when the sense of agency is experienced at an intermediate level between full control and no control at all. In consequence, we propose that human-operated, artificial intelligence-enabled robotics needs to be designed such that the balance between human sense of agency and the autonomy of the assistive system is achieved, for optimizing the users’ motivation to use the technology, and for improvement of performance. We describe various examples of assistive technologies, ranging from robots in work environments to healthcare robotics, and we discuss how sense of agency is a crucial factor that needs to be accounted for in the engineering endeavor of developing such technologies.
The hippocampus plays a critical role in generalization, enabling us to flexibly repurpose prior experiences to perform novel tasks. Here, we suggest that data augmentation — a machine learning strategy to improve generalization by refactoring prior experience — offers a useful framework to conceptualize and model hippocampal function. We begin by outlining how data augmentation operates across two timescales: the traditional ‘offline’ setting, where refactoring training data yields more general representations, and an ‘online’ setting, where retrieved experiences can be flexibly refactored at test time to support zero-shot inference. We suggest that these ‘offline’ and ‘online’ computational strategies map onto functions supported by the hippocampus. Critically, we argue that these computational tools can be leveraged to develop formal ‘linking functions’ between experimental evidence and theoretical claims, such that a unified modeling approach can be used to predict the diverse behaviors that depend on the hippocampus — from navigating in high-dimensional sensory environments to more abstract inferences. We hope that this perspective, and the modeling strategies it makes available, will support new efforts to formalize and evaluate theories of hippocampal function.
Planning is often described as the use of a world model to evaluate possible futures before choosing an action. In this review, we ask how such models are acquired, utilized, and revised in the brain. We frame the world model as the combination of a transition model, which predicts possible future latent states, and an outcome model, which assigns decision-relevant consequences to states and trajectories. We organize the review according to four complementary perspectives. Intuitive physics contributes to transition-model learning by providing a physics prior over object-centered latent dynamics. Model-free reinforcement learning contributes to outcome-model learning by providing structured teaching signals over reward distributions, task-relevant features or states, and temporal horizons. Successor representation supports flexible outcome-model updating by learning predictive maps that combine with reward functions to generalize value, adapt policy, and refine outcome estimates. Model-based reinforcement learning then deploys the acquired world model during deliberation and, through metacognitive control, evaluates whether it should continue to be queried, trusted, revised, or replaced. Together, these perspectives suggest that biological planning may depend not on a single monolithic representation, but on coordinated predictive resources that constrain possible transitions, shape outcome expectations, support flexible outcome update, and govern model usage. This framework also suggests that brain-inspired artificial intelligence should consider not only whether an agent has a world model, but what kind of predictive organization supports its use.
The rapid rise of large language models (LLMs) has sparked intense debate across multiple academic disciplines. While some argue that LLMs represent a significant step toward artificial general intelligence or even machine consciousness (inflationary claims), others dismiss them as mere trickster artifacts lacking genuine cognitive abilities (deflationary claims). We argue that both extremes may be shaped or exacerbated by common cognitive biases, including cognitive dissonance, wishful thinking, and the illusion of explanatory depth, which distort reality to our own advantage. By showcasing how these distortions may easily emerge in both scientific and public discourse, we advocate for a measured approach: a skeptical, open mind that recognizes the cognitive abilities of LLMs as worthy of scientific investigation while remaining conservative concerning exaggerated claims regarding their cognitive and moral status.
Loneliness has been associated with perinatal mental illness, which is shown to negatively affect infant mental health. This review outlines current understandings of prevalence and negative impacts, and summarises what is currently known about contributing factors, highlighting multidimensional experiences of loneliness in the perinatal period (pregnancy and two years following birth). The socioecological framework is used to explore evidence on individual, interpersonal, organisational, community and societal factors that contribute to perinatal loneliness. Parents who experience intersectional inequalities are most at risk, including parents who are on low incomes, young, LGBTQ+, from ethnically marginalised populations, or affected by poor health. Most interventions to reduce perinatal loneliness operate at the individual or interpersonal level. To fully address perinatal loneliness, it is also important to further explore organisational factors and adopt a spatial and social justice approach that addresses the community and societal drivers such as poverty, discrimination and unsupportive social policies.
Successful action selection and execution requires fine-grained control across a continuous and complex range of movement parameters and contexts. Behavioral decisions are rarely simple binaries. While indexing and selecting between individual discrete actions has been an attractive model in reductionist conditions, the myriad state-action values an animal must keep track of make biological bookkeeping untenable. Here, we lay out the key problems a brain must solve when choosing and optimizing how to interact with our diverse world. We then review several behavioral and physiological studies that suggest a reappraisal of how the brain, and specifically the basal ganglia, solves these problems.
The ability to mentally navigate beyond observed experience is a defining feature of human cognition. Converging evidence now indicates that the neural codes supporting physical navigation also underpin goal-directed navigation in nonphysical domains. In this review, I argue that the default mode network (DMN), together with the hippocampal-entorhinal circuitry, constitutes a domain-general system for the organization and mental navigation of relational knowledge. Recent studies demonstrate navigational neural codes not only in the medial temporal lobe, but also across key DMN regions during navigation of both physical and abstract spaces, including social hierarchies, conceptual knowledge, and value-based decision-making. Based on these findings, I propose that the DMN facilitates mental navigation through three interrelated operations, including association, abstraction, and anticipation, that collectively enable the construction and deployment of topologically structured abstract cognitive maps across multiple domains of knowledge. This framework reconciles the diverse functional associations and clinical vulnerabilities of the DMN under a unifying principle: mental navigation as a default computational mode in human cognition.
We embed the default mode network (DMN) within a predictive processing architecture, arguing that its core function is to coordinate high-level predictions in the brain’s hierarchical generative model. Rather than mapping DMN regions onto broad psychological categories such as memory, social cognition, or emotion, we propose that the network plays a unified computational role. At a global level, the DMN occupies a privileged position in large-scale cortical organization, supporting abstract, multimodal representations that drive predictions across extended spatial and temporal scales. At a local level, this role is structured by systematic functional-anatomical variation within the network itself. We describe three broad axes of DMN organization grounded in computational neuroanatomy. First, a longitudinal axis differentiates a posterior complex that is more tightly coupled to high-dimensional sensory systems from an anterior complex with stronger connectivity to visceromotor subcortical structures and longer integration timescales. This axis reflects graded differences in predictive scope rather than a simple external–internal divide. Second, hierarchical gradients of a generative architecture support increasing abstraction, not only between the DMN and other networks, but also within DMN complexes. Third, patterned laminar connectivity enables bidirectional predictive signaling between DMN complexes. Together, this framework links the DMN’s global computational role to its internal structural organization and provides a mechanistic account of its role in large-scale brain function.
While social media offers opportunities to increase social capital and to stay in contact with close ties, studies show social media use is associated positively and negatively with loneliness, indicating there is a dynamic relationship. To fully understand the association factors that influence the relationship need to be explored. This review updates an earlier proposed model incorporating recent research into influencing factors and identifies gaps in the evidence and methodological issues, suggesting focusses for future research. User characteristics and features of social media platforms influence the ways in which people use social media. Although there have been some important criticisms of the active-passive social media use dichotomy, there is strong support that passive social media use is associated with high loneliness, which may be due to this type of use being less likely to lead to interactions with others. Motivations for using social media, interactions and responses from others influence whether social gains are realised. Those with high social capital are most likely to experience the social gains from social media. The evidence in this area is still largely cross-sectional, so future research should use longitudinal and experimental designs to examine causality. Multi-level analyses will be important to identify between- and within-person effects. Future research will need to examine in more detail mediating and moderating factors, and as age/generational dependent effects are evident, particularly in studies examining user motivation, these should also be considered.
Although the default mode network (DMN) is consistently implicated in social cognition, its functional contributions to social thought and behavior have been challenging to characterize. To date, scientific understanding of DMN function has mainly been informed by neuroimaging studies of task-based activation. In this review, we adopt a complementary approach, capitalizing on the observation that exogenous cognitive load tasks reliably deactivate the DMN. In a variety of dual-task studies, an exogenous cognitive load task (e.g. digit string memorization) has been paired with a concurrent task of interest (e.g. personality trait attribution). Although typically conducted for other reasons, we propose that such studies offer a means to ask what happens to behavioral performance across a range of tasks when DMN activity is temporarily perturbed, helping to constrain hypotheses about the cognitive functions of the DMN. Here, we review patterns of behavioral performance on a variety of social and nonsocial tasks under concurrent cognitive load, treating load as a temporary DMN perturbation. Although results are mixed when organized by task type, a clearer pattern emerges when organized by how much uncertainty reduction demand (URD) tasks imposed on participants: tasks involving higher URD, but not lower URD, are reliably disrupted under exogenous cognitive load. This evidence tentatively suggests uncertainty reduction as an overarching framework through which to view the contributions of the DMN to (social) cognition, reconciling seemingly contradictory evidence from past studies and opening new avenues for future research.
In the last quarter of a century, the default network (DN) has become a major focus of scientific research. Researchers have sought to understand its functional properties and relationship to various psychological, clinical, and social variables. Work on the DN has occurred alongside a tension between considering the DN as a unitary system versus treating it as a heterogenous system comprised of subcomponents or networks. A potential difficulty researchers face when recognizing the heterogeneity of the DN is knowing how these different fractionations compare to one another. In this review, we provide an overview of the evidence for heterogeneity of organization within the DN. We begin by reviewing the first group-level fractionations of the DN into different subsystems. Next, we consider recent individual-level fractionations of the DN that reveal organizational heterogeneity missed by group-level approaches. Lastly, we discuss how different approaches to brain network estimation may influence the measurement of DN organization. We conclude with a discussion of what researchers in the field of network neuroscience can do to increase the adoption and appreciation of DN heterogeneity.
Episodic memories showcase the complexity of internal thought: they bind together people, places, and events into coherent narratives, and they carry a subjective quality that sets them apart from other mental representations. In this review, I argue that episodic memory provides a powerful lens for understanding the functional organization of the default mode network (DMN). Recent research reveals that memory-related functions of the DMN are organized along at least two key dimensions— content and specificity. Distinct areas support memory for different types of content (e.g., object, spatial, and social information) and with varying levels of specificity (e.g., detailed episodes vs general schemas). Together, DMN pathways can support the diversity of episodic thought, imbuing multidimensional representations with subjective experience. By delineating how these pathways correspond with memory features, we can gain insight into how the DMN builds and updates internal narratives across cognitive domains.
Racial disparities in K-12 education funding have a negative impact on student outcomes and have been documented by researchers for decades. Yet, these disparities persist in many states. In this paper, I survey empirical evidence on racial disparities in K-12 funding and their impact. Then, I draw on insights from QuantCrit and recent historical scholarship to examine how racial funding gaps get created, maintained, and misunderstood in research and policy. I discuss how the traditional explanation offered for these disparities in research and policy — the underfunding of poor school districts through race-neutral lawmaking — is untenable. While the allocation of fewer resources to poorer school districts contributes to funding inequities, I describe how recent work shows the marginalization of poor districts by ostensibly race-neutral policies cannot, on its own, explain race-based funding differences. Instead, I suggest that researchers consider the role of ‘racist causality’ — a term coined by historian Leah Gordon to describe the failure of social science to address racism as an explanatory variable during the 1960s and 1970s — when studying the distribution of education funding.
This review examines the role of elementary science principal decision-making in elementary science. This review offers literature with background about elementary principals as leaders who serve as decision makers, as well as five frames for equity, with a specific focus on race, to examine a specific case study. The findings are that the five frames for equity offer an expansive view of equity; however, several frames for equity have yet to be deeply engaged by administrators regarding elementary science.
Atlanta’s designation as the Black Mecca signals Black political, cultural, and economic prominence, even as the city remains deeply segregated and structurally unequal. This paper reframes the Black Mecca as a research design intervention by introducing the Black Mecca Method, a structured approach for examining how racialized systems redesign institutional conditions where equity is most expected. Drawing on 36 peer-reviewed sociological studies published between 2022 and 2024, we analyze how contemporary scholarship engages three interlocking dimensions of racial equity — material, symbolic, and temporal — and situate these within an analytic architecture grounded in ethnoraciality, systemic racism, and racial colonial capitalism. Through this framework, Atlanta serves as a high-expectation site that makes racialized infrastructures empirically legible by tracing divergence between anticipated and observed outcomes. The Black Mecca Method integrates quantitative approaches to measurement, modeling, interpretation, and research design to examine how racial inequality persists through institutional systems organized around housing, labor, governance, and spatial regulation. Rather than documenting disparity alone, we process how racial power normalizes unequal outcomes as progress. The method offers a portable analytic framework for studying racialized systems that publicly affirm Black advancement but structurally constrain it.
The goal of this review is to synthesize recent, methodologically and conceptually diverse bodies of literature addressing the cause and consequence of racial disparities in school funding and equal educational opportunity in pursuit of enhancing school funding research practices. The three bodies of relevant literature include: 1) empirical validations of the presence of racial disparities, and insights regarding empirical associations with economic, demographic, governance, and structural conditions of schooling; 2) historical origins and contemporary relevance of these influential economic, demographic, governance, and structural conditions; and 3) political and economic theories for why these systems have evolved over time and why substantive change to the underlying systems and policies that reinforce racial disparities in school funding is difficult. The studies reviewed herein demonstrate that race-avoidant policies will not sufficiently remedy racial disparities in school funding because of structurally and individually racist causes. Racism caused the disparities throughout the school funding system.
Default network anticorrelation with other neurocognitive networks has been a topic of ongoing and expanding interest in neuroscience. Advances in anticorrelation research have moved beyond debates regarding methodological artifact to reveal a more comprehensive view of brain function. We review neurocognitive, anatomical, computational, neuromodulatory, and dynamical approaches to the study of anticorrelation. We argue that anticorrelation is a robust feature of network organization, revealing a latent principle of network competition.
In this methodological-reflection paper, the authors discuss their experiences conducting science, technology, engineering, and mathematics (STEM) research that employs critical frameworks while connecting qualitative and quantitative approaches. Specifically, they reflect on their research concerning Black undergraduate women in STEM and discuss how critical lenses informed the development of a measure that articulates the race-gendered experiences of these Black women in fields where they often experience intersecting race and gender biases. The authors discuss the epistemological tensions they managed as scholars doing work that is both critical and quantitative. Moreover, they discuss limitations of their work and how it might be extended in future research.
Sustainability technologies are critical for addressing climate change, yet public opposition can hinder their adoption. Recent research shows that perceptions of naturalness influence public support for technology. Naturalness is a multidimensional psychological construct. Technologies are seen as natural if they involve minimal processing, align with ecological norms, and are culturally aligned. More natural technologies, like afforestation, solar energy, and plant-based foods, are viewed as safer, more beneficial, familiar, and positive garnering greater public support than less natural technologies like direct air capture, nuclear energy, or lab-grown meat. People with heightened aversion to altering natural systems are particularly swayed by a technology's perceived artificialness or naturalness. Naturalness is thus a core factor shaping public reactions to sustainability technologies, and understanding this concept can enhance communication strategies aimed at reducing public resistance.
