The research, published in BMC Psychiatry by a team led by Yujue Wang and Ciqing Bao, with senior authors Zhijian Yao and Qing Lu, examined a property of brain organization known as structural-functional connectivity coupling, often abbreviated SC-FC coupling. The concept is elegant in principle. Structural connectivity refers to the physical white matter highways that link one brain region to another, mapped using diffusion imaging that traces the direction of water movement along nerve fibers. Functional connectivity refers to the statistical synchrony between activity patterns in distant regions, typically measured from fluctuations in the blood oxygenation level dependent, or BOLD, signal captured by functional MRI. In a healthy brain, these two architectures are tightly interwoven: regions that are strongly wired together tend to activate together, and the precise alignment between structure and function follows a characteristic spatial gradient across the cortex.

That alignment is not merely a technical curiosity. Neuroscientists have proposed that SC-FC coupling reflects how efficiently a region converts its anatomical infrastructure into coordinated computation. In typical adults, coupling is strongest in primary sensory and motor regions, where function depends heavily on direct structural input, and weakest in association cortices such as the prefrontal and parietal networks that support abstract thought, self-reflection, and flexible cognition. Deviations from this expected pattern have been reported in schizophrenia, aging, and depression, making coupling a promising candidate marker for psychiatric illness. What remained unclear, until now, was whether suicidal ideation in depressed youths leaves its own specific imprint on this structure-function relationship.

To find out, the researchers recruited participants aged 15 to 30, a window spanning adolescence through early adulthood when depression frequently first appears and when suicide risk climbs steeply. The cohort included 97 patients with major depressive disorder who were experiencing suicidal ideation, 44 depressed patients without such thoughts, and a group of healthy controls. Crucially, all patients were drug-naive, meaning they had never received psychiatric medication. This detail matters enormously for imaging research, since antidepressants can themselves alter brain activity and connectivity, and studying medication-free individuals allows investigators to observe the illness in a relatively pristine state. Participants were assessed with established clinical instruments, including the Hamilton Depression Rating Scale, the Hamilton Anxiety Rating Scale, the Ruminative Responses Scale, and the Beck Scale for Suicide Ideation, which measured both past-week and worst-point suicidal thinking.

On the technical side, the team combined diffusion-weighted imaging analyzed with generalized q-sampling imaging, a method that resolves crossing nerve fibers more effectively than simpler diffusion tensor approaches, with resting-state functional MRI collected while participants simply lay still. White matter connections between regions were quantified using measures of fiber integrity, and functional connections were derived from correlated BOLD time series across 90 cortical and subcortical regions defined by the Automated Anatomical Labeling atlas. For each region, the researchers computed the correlation between its pattern of structural links and its pattern of functional links to all other regions, yielding both a global coupling measure and a local one for every brain area. Group differences were tested with analysis of covariance, and statistical rigor was enforced through false discovery rate correction for multiple comparisons, with additional sensitivity analyses to guard against the confounding effects of head motion, depression severity, and demographic differences.

The results were striking. Across the three groups, SC-FC coupling differed significantly both globally and locally, but the most informative findings emerged in two left-hemisphere regions: the left superior temporal gyrus and the left Heschl’s gyrus, a small structure tucked inside the lateral sulcus that serves as the brain’s primary auditory cortex. Patients with suicidal ideation showed the weakest negative SC-FC coupling in the left Heschl’s gyrus compared with both non-suicidal depressed patients and healthy controls, a difference that survived correction and reached a corrected p value below 0.001. A similar, though slightly less pronounced, reduction in negative coupling in the left superior temporal gyrus distinguished the suicidal group from healthy controls, with a corrected p value of 0.003.

The word negative deserves careful attention here, because it carries the theoretical weight of the finding. In healthy brains, some regions display an inverse relationship between structural and functional connectivity, a pattern thought to reflect a mature, specialized form of organization in which functional integration is not slavishly dictated by anatomy. Weaker negative coupling in the suicidal patients therefore suggests that this finely tuned inverse relationship has been blunted or disrupted. In other words, in young depressed people with suicidal thoughts, the normal counterbalance between the brain’s wiring diagram and its activity dynamics appears loosened specifically in temporoparietal auditory and language-related cortex. Notably, the abnormality in Heschl’s gyrus separated suicidal from non-suicidal depressed patients, indicating that the signature is tied to suicidal ideation itself rather than to depression in general.

Even more compelling was the clinical correlation. Within the suicidal group, the degree of coupling disruption in the left Heschl’s gyrus was negatively correlated with rumination, the repetitive, self-focused pattern of thinking in which distressing thoughts churn endlessly, and with scores on the suicidal risk measures. Rumination is one of the strongest psychological predictors of suicidal behavior, and linking it to a measurable property of brain organization in the superior temporal region is intriguing given that this cortex participates in language processing, self-referential thought, and the integration of social and auditory information. One plausible interpretation, offered cautiously by the findings rather than proven by them, is that disrupted structure-function coordination in this circuitry contributes to the maladaptive loops of internal dialogue that characterize ruminative despair.

The study is cross-sectional, capturing a single moment in time, so it cannot establish whether the coupling abnormality is a cause of suicidal thinking, a consequence, or a shared vulnerability marker. The samples, while well characterized, differ in size across groups, and suicidal ideation itself is heterogeneous, spanning passive wishes to be dead and active plans with means. Replication in larger and more diverse cohorts, and ideally longitudinal tracking of patients as their symptoms evolve, will be essential before SC-FC coupling can be considered a clinical tool. The authors and their funders, supported by the National Natural Science Foundation of China and several provincial programs, frame the measure as a candidate neurobiological correlate rather than a ready-made diagnostic test, an appropriately restrained claim.

Nevertheless, the work opens a genuinely new window onto one of psychiatry’s most urgent problems. If the alignment between brain structure and function in temporoparietal cortex can be measured reliably with a standard MRI session, it could eventually complement clinical interviews in flagging depressed young people at elevated risk, allowing earlier and more targeted intervention. It may also guide future treatments, whether pharmacological, neuromodulatory, or psychotherapeutic, by identifying the circuits most in need of support. For now, the study’s message is quieter but no less important: the neural signature of suicidal despair in youth may be written not in any single region, but in the delicate correspondence between what the brain is built from and what it does, and in the auditory and language circuits where the darkest thoughts are spoken internally before they are ever voiced.

Subject of Research: Structural-functional connectivity coupling abnormalities associated with suicidal ideation in young people with major depressive disorder

Article Title: Temporoparietal structural-functional coupling abnormalities in drug-naïve youths with major depressive disorder and suicidal ideation

Article References: Wang, Y., Bao, C., Bao, J., Chen, W., Zhang, Q., Yan, R., Yao, Z., & Lu, Q. (2026). Temporoparietal structural-functional coupling abnormalities in drug-naïve youths with major depressive disorder and suicidal ideation. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08608-7

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08608-7

Keywords: major depressive disorder, suicidal ideation, SC-FC coupling, structural connectivity, functional connectivity, fMRI, diffusion imaging, rumination, temporoparietal cortex, Heschl's gyrus, superior temporal gyrus, youth mental health

The research, led by Dr. Noam Goldway under the supervision of Prof. Talma Hendler and Dr. Haggai Sharon of Tel Aviv University’s Institute for Psychedelic Research at the Sagol School of Neuroscience, was published in the journal Neuropsychopharmacology. For the first time in humans, the team demonstrated that ketamine activates two distinct mechanisms in the brain: one responsible for pain relief and another responsible for the dissociative state the drug is famous for. The distinction is not merely a matter of subjective experience — it is visible in the brain’s activity patterns.

The question the researchers set out to answer is one of the most consequential in modern psychedelic medicine. In recent years, mind-altering substances have migrated from the fringes of culture to the center of serious clinical investigation. MDMA is being studied as a treatment for post-traumatic stress disorder; psilocybin, the active compound in magic mushrooms, is being tested primarily against depression; and ketamine has attracted considerable interest for its effectiveness in treating pain, depression and, more recently, PTSD. Yet a central unresolved debate shadows the entire field: is the alteration of consciousness produced by these drugs a necessary condition for their therapeutic benefit, or merely a side effect that happens to accompany it?

To address that question, the team turned to ketamine, a drug that relieves pain while simultaneously inducing dissociation — a temporary sense of detachment from one’s self, body or surroundings. For years, the working assumption had been that these two phenomena were causally linked: the more detached a patient felt, the less intense or threatening their pain would seem. If that were true, then the psychedelic experience itself would be an essential ingredient of the drug’s analgesic action, with major implications for how ketamine-based therapies should be designed.

The study enrolled 37 healthy volunteers, both women and men, each of whom participated in two separate sessions. In one session they received a ketamine infusion; in the other, a placebo. While under the influence of each infusion, participants underwent functional magnetic resonance imaging (fMRI) and were exposed to carefully controlled pain stimuli. Throughout the experiment, they reported both the intensity of their pain and the depth of their dissociative experience, allowing the researchers to track the two phenomena independently and in real time.

The behavioral results were strikingly clear. As expected, ketamine reduced pain and increased dissociation. But when the researchers analyzed the relationship between the two effects, no connection emerged. Participants who experienced strong dissociation did not necessarily enjoy greater pain relief, and those with the most substantial analgesia were not necessarily the most detached. As Dr. Goldway explained, the sense of detachment is simply not the mechanism through which ketamine reduces pain. The two effects, long assumed to be two faces of the same process, appear to run in parallel without one driving the other.

The brain imaging data reinforced the behavioral findings at the neural level. Pain relief was accompanied by reduced activity in brain regions classically involved in processing pain, including the insula and the anterior cingulate cortex — structures that map the intensity and emotional salience of noxious stimuli. Dissociation, by contrast, was related to changes in a different network altogether: the default mode network, a set of interconnected regions implicated in, among other things, the construction of the sense of self. In other words, the study did not merely document two uncorrelated subjective experiences; it revealed two different brain mechanisms, both triggered by the same drug but operating on distinct neural circuitry.

Prof. Hendler, who heads TAU’s Institute for Psychedelic Research, framed the findings within the broader scientific controversy. “In recent years, there has been a lively scientific and clinical debate over whether the subjective experience produced by psychoactive drugs is an integral part of their therapeutic effect,” she said. “Our study shows that, at least in the case of ketamine and pain relief, the answer is apparently no. We were able to show that the drug’s effect on pain is separate from its effect on consciousness — both behaviorally and in terms of brain activity.” She was careful, however, to draw a boundary around the conclusion: “It is important to note that this cannot be generalized to other psychedelic drugs, such as magic mushrooms. In this case, the burden of proof still rests with us, the scientists.”

Dr. Sharon, Deputy Head and Medical Director of the Institute and a specialist in pain medicine, emphasized what the results reveal about the brain itself. The study, he noted, provides further evidence of the complexity of the brain’s representation of the pain experience and of the many different ways the brain can regulate and alter it. Pain, in this view, is not a single signal to be switched off but a multidimensional construct assembled across multiple networks — and different drugs, or different mechanisms within the same drug, can intervene at different points in that assembly process.

The clinical implications could be far-reaching. If researchers can learn how to activate ketamine’s pain-relief mechanism without simultaneously inducing dissociation, it may become possible to develop future treatments that preserve the drug’s analgesic effectiveness while reducing the sense of detachment and other subjective side effects. Such decoupled therapies could broaden access to ketamine’s benefits, particularly for patients who cannot tolerate its psychological effects, and could reshape how clinicians think about the role of altered consciousness in psychedelic-assisted medicine more broadly. For now, the study stands as a landmark demonstration that, where ketamine and pain are concerned, relief and detachment are separable — and the road to better analgesics may not need to pass through an altered mind at all.

Subject of Research: Dissociation of ketamine's analgesic and consciousness-altering brain mechanisms in humans

Article Title: Ketamine relieves pain without “disconnecting” consciousness

Article References: Ketamine relieves pain without “disconnecting” consciousness. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: ketamine, pain relief, dissociation, consciousness, fMRI, psychedelic research, insula, anterior cingulate cortex, default mode network, Neuropsychopharmacology, Tel Aviv University, analgesia

The central challenge, the editors argue, is not a shortage of data but the character of the data themselves. Neuroscience and psychology datasets are frequently noisy, heterogeneous, and unevenly labeled, which complicates model development, result interpretation, and ultimately clinical translation. A single brain imaging study may combine functional MRI sequences collected under different protocols, self-report questionnaires with inconsistent response scales, and clinical records riddled with missing entries. Measurement noise alone has been shown to limit the accuracy of brain-behavior predictions, a point underscored by a 2024 Nature Communications analysis cited in the editorial. Any AI system hoping to move from the laboratory to the clinic must therefore be robust, interpretable, and generalizable across diverse populations and real-world settings, not merely accurate on a curated benchmark.

One of the most striking contributions comes from Gao and colleagues, who developed a deep learning framework for health risk assessment in adolescents. Their architecture combines a two-dimensional convolutional autoencoder with multi-sequence learning and multi-scale asynchronous correlation information extraction, a mouthful of terminology that describes a genuinely clever engineering solution. Applied to functional MRI data from the Adolescent Risk Behavior dataset, the framework was designed to capture both spatial relationships among brain regions and time-shifted correlations between them, all while contending with limited labeled data, a chronic constraint in psychiatric research. The study demonstrated promising performance in understanding and predicting health-related risks in adolescents, suggesting that deep learning could enhance the precision of health risk assessment and support early identification of individuals who may benefit from timely intervention. The editors caution, however, that more robust external validation and assessment of generalizability across diverse populations are essential before such tools are ready for real-world implementation.

Perhaps the most conceptually ambitious study in the collection, from Chalmers and colleagues, used deep reinforcement learning to simulate the effect of dendritic spine loss as observed in major depressive disorder. Dendritic spines are the tiny protrusions on neurons that receive synaptic inputs, and their loss is a well-documented histopathological correlate of depression. When the researchers simulated this spine loss in an artificially intelligent agent, the agent produced a variety of MDD-like behaviors, including anhedonia, increased temporal discounting, avoidance, and an altered exploration-exploitation balance. Remarkably, simulating alternative and longstanding reward-processing-centric conceptions of MDD, such as dysfunction of the dopamine system, altered reward discounting, context-dependent learning rates, and increased exploration, did not produce the same range of depressive behaviors. This differential result positions AI not merely as a diagnostic or predictive instrument but as a computational microscope for probing pathophysiology, offering a way to test mechanistic hypotheses about depression that would be impossible to isolate experimentally in human patients, and potentially informing therapeutic strategies.

Clinical sensitivity is also the theme of a perspective contributed by Haykal and Kroumpouzos, who examined the opportunities, limitations, and ethical considerations of leveraging AI to detect body dysmorphic disorder. BDD is a frequently overlooked yet clinically significant condition characterized by a persistent and intrusive preoccupation with perceived flaws in appearance, flaws that are typically minor or imperceptible to others. Individuals with BDD often seek repeated cosmetic procedures in hopes of achieving relief, but they typically find little psychological benefit from such interventions, making early recognition a genuine clinical priority. The authors explored how AI approaches could improve the management of BDD through early diagnosis, prompt reflection, highlighting of concerning patterns, and enhanced clinical sensitivity, while insisting on transparency, empathy, and restraint in how such systems are deployed with a vulnerable patient population.

Rehabilitation technology represents another frontier where AI and computational neuroscience converge, particularly for older adults. Xing and colleagues developed a deep learning model to process point-cloud data acquired from laser sensors for non-contact analysis of hand gestures and surface features, with the aim of supporting hand function rehabilitation through human-machine interaction. By analyzing hand surface point clouds with local feature extraction and abstraction and enhancement of dimensional information, the approach effectively captured relevant spatial information and demonstrated promising performance in gesture and hand-surface feature recognition. The significance of non-contact sensing should not be understated: cameras and laser scanners can monitor movement without wearable electrodes or markers, making rehabilitation assessment more natural, accessible, and data-driven, especially for patients who find conventional instrumentation uncomfortable or stigmatizing.

At the scale of entire disease categories, Liu and colleagues reviewed the application of AI-based radiogenomics in managing neurodegenerative diseases, including multiple sclerosis, dementia, Parkinson’s disease, amyotrophic lateral sclerosis, and Creutzfeldt-Jakob disease. Radiogenomics, the fusion of imaging-derived radiomic features with genomic data, allows computational models to link visible brain changes with molecular signatures. The review summarized representative AI models across these conditions and catalogued publicly available data resources supporting radiomics, genomics, and radiogenomics studies. It also highlighted key directions for future research, including strategies to mitigate overfitting in high-dimensional genomic data, robust validation approaches that account for modality-specific variability and incomplete data, and efforts to bridge the gap between AI-enhanced models and their translation into real-world clinical practice, a gap that has swallowed many otherwise impressive algorithms.

Looking forward, the editors identify emerging methodologies that extend beyond what the Research Topic has demonstrated. Large language models may enable more sophisticated analysis and interpretation of complex medical information, supporting personalized risk assessment and tailored intervention, with recent work on knowledge-augmented models using verified retrieval for improved medical reasoning cited as an example. Multimodal learning could integrate complementary data sources to generate more comprehensive representations of individual health, and integrating subjective measurements such as questionnaires with objective measurements such as wearable sensing data may further enhance the understanding of individual health status. Further advances in model optimization, robustness, interpretability, and clinical validation will be essential for translating these innovations into real-world applications, and the editors point to automated architecture search and layer-wise convex network designs for physiological signal classification as evidence that optimization methods are maturing rapidly.

What makes this editorial more than a routine literature survey is its insistence on interdisciplinary synergy as the decisive factor. The editors argue that progress will depend on collaboration across neuroscience, psychology, clinical medicine, and computational sciences, developing AI systems that are technically capable, clinically meaningful, and generalizable across diverse populations and settings. The studies collected here, from simulated dendritic spines to laser-scanned hand gestures, show that the field has moved well beyond proof-of-concept demonstrations. The remaining distance to responsible clinical translation is not primarily a matter of algorithmic sophistication but of validation, ethics, and deployment, and the editors hope this Research Topic will stimulate further interdisciplinary research and responsible translation of health data science and AI in neuroscience and psychology, fields where the stakes, quite literally, include the human mind itself.

Subject of Research: Health data science and artificial intelligence applications in neuroscience and psychology

Article Title: Editorial: Health data science and AI in neuroscience & psychology

Article References: Lu, L., Huo, Z., Li, J., & Zhao, E. (2026). Editorial: Health data science and AI in neuroscience & psychology. Frontiers in Computational Neuroscience, 20, Article 2002192. https://doi.org/10.3389/fncom.2026.2002192

Image Credits: AI Generated

DOI: 10.3389/fncom.2026.2002192

Keywords: artificial intelligence, health data science, neuroscience, psychology, deep learning, major depressive disorder, body dysmorphic disorder, rehabilitation, radiogenomics, neurodegenerative diseases, fMRI, machine learning

The core problem with conventional smoothing lies in its indifference to brain anatomy. A Gaussian kernel treats the brain as a continuous 3D volume, aggregating signal from any voxel within its radius regardless of whether those voxels belong to the same gyrus, the same cortical layer, or even the same tissue type. When the cortex folds in complex sulcal patterns, two voxels that sit just millimeters apart in Euclidean space may be separated by a deep sulcus and belong to functionally distinct regions. Smoothing across that boundary blurs signal from one gyrus into its neighbor, creating activation clusters that never existed biologically. The white-gray matter interface presents an even more vulnerable boundary, as signal from cortical gray matter can leak into white matter and artificially shift the apparent peak of activation away from the cortex.

Surface-based fMRI analysis has long been considered the gold standard for avoiding these volumetric smoothing artifacts. By projecting data onto an unfolded cortical surface, this approach inherently respects the geodesic distance along the cortex and prevents signal from crossing sulci. But surface-based workflows require complex preprocessing pipelines, specialized file formats like CIFTI, and a small set of tools that may not integrate with researchers’ preferred software. Many established neuroimaging packages still lack surface-based capabilities, leaving most labs to choose between adopting an entirely new analytical paradigm or continuing with volumetric smoothing and accepting its artifacts.

Ellis and Aizenberg built their constrained smoothing tool as a practical middle ground. The method represents each brain voxel as a node in a graph, with edges connecting all 26 neighboring voxels in 3D space. Critically, the researchers then prune any edge that crosses either the pial or white matter surface as reconstructed by FreeSurfer, using ray tracing to rapidly evaluate surface intersections across the more than 40 million nodes in a typical brain graph. After pruning, the graph separates into four independently smoothable components—left and right gray matter and left and right white matter—ensuring that smoothing kernels can never aggregate signal across the cortical boundaries. The researchers implemented the approach in Python, avoiding the licensing costs of MATLAB, and optimized the heat kernel smoothing parameter to approximate a user-specified Gaussian full-width at half-maximum, allowing direct comparison between constrained and unconstrained smoothing at equivalent kernel widths.

To evaluate the method, the researchers applied both smoothing approaches to simulated data, a sensory task fMRI dataset, a precision motor mapping dataset from the Midnight Scan Club, and resting state fMRI data from the Human Connectome Project in Aging. In the simulation experiments, where 68 cortical regions were isolated and white noise added, constrained smoothing accurately recovered the shape of the original regions while Gaussian smoothing created artifacts by smoothing across boundaries. Constrained smoothing demonstrated higher accuracy in terms of both Dice similarity coefficient and Pearson correlation across all tested kernel widths, and more accurately concentrated positive predictions within gray matter while reducing white matter activation.

The real-world task fMRI results told a similarly compelling story. In the sensory task analysis, where participants received brushing stimulation to the left hand, both Gaussian and constrained smoothing increased the reliability of activations across repeated acquisitions. However, Gaussian smoothing significantly increased the percentage of active voxels in white matter relative to constrained smoothing. In the motor task experiments, where each of ten subjects completed 20 runs of finger tapping, toe tapping, and tongue movement, Gaussian smoothing with kernel widths greater than 3 mm decreased the accuracy of individual session maps compared to constrained smoothing. Notably, smoothing did not uniformly improve accuracy across all metrics—Dice scores were actually highest with no smoothing at all, suggesting that the reliability gains from smoothing come at a cost to spatial precision.

The false positive analysis addressed a critical statistical concern. Because constrained smoothing is inherently anisotropic, it produces smoothed data that may violate the Gaussian random field assumptions underlying parametric statistical models in packages like FSL FEAT. Previous research by Eklund and colleagues has shown that such violations can lead to inflated false positive rates in fMRI studies. Ellis and Aizenberg tested this by running task fMRI analysis on resting state data—data acquired with no task being performed—so that any detected activation could be considered a false positive. With cluster significance thresholding, mean false positive voxel percentages remained below 5% for both smoothing methods across all tested kernel widths, providing reassurance that constrained smoothing does not dramatically inflate statistical errors when appropriate thresholding is applied.

Perhaps the most surprising findings emerged from the resting state connectivity analysis. Using the Schaefer/Yeo 1,000-parcel atlas, the researchers fit generalized additive mixed models to examine how smoothing altered the relationship between Euclidean distance and functional connectivity. Both Gaussian and constrained smoothing inflated connectivity between spatially proximal regions, but Gaussian smoothing demonstrated more inflation at kernel widths of 6 mm or greater. Even more striking, both smoothing methods significantly altered graph theory metrics including clustering coefficient, global efficiency, local efficiency, and modularity under all tested kernel widths. Constrained smoothing had less effect on local efficiency, global efficiency, and clustering coefficient at higher kernel widths, though the differences between methods were marginal compared to the overall effect of smoothing itself on these network measures.

The researchers acknowledge an important caveat: constrained smoothing is inherently more conservative than unconstrained smoothing at identical nominal kernel widths. By restricting connections across cortical boundaries, constrained smoothing aggregates fewer data points, effectively lowering the dimensionality of the smoothing operation. The effective level of smoothing also depends on graph connectivity, which in turn depends on voxel size—smaller voxels produce more connections and more data aggregation. This suggests constrained smoothing may be most advantageous in high-resolution fMRI experiments, where the additional anatomical detail justifies the computational cost of graph construction and pruning.

As a practical alternative to both unconstrained volumetric smoothing and full surface-based analysis, anatomically constrained volumetric smoothing offers researchers a way to enhance fMRI reliability while keeping data in the familiar voxel space. The method requires reconstructed cortical surfaces but avoids the need to transform data into surface-based representations, preserving compatibility with decades of established volumetric processing pipelines. Future work could extend this graph-based approach to smooth along white matter tracts, integrate with Bayesian inference frameworks that determine optimal smoothing from the data, or apply nonparametric permutation testing for valid statistical inference on constrained-smoothed volumetric data. For now, the study provides compelling evidence that the smoothing step so many researchers take for granted deserves a second look—and that respecting the brain’s own boundaries may be the key to more accurate brain maps.

Subject of Research: Anatomically constrained volumetric smoothing enhances fMRI reliability while avoiding smoothing artifacts

Article Title: Anatomically constrained volumetric smoothing enhances fMRI reliability while avoiding smoothing artifacts

Article References: Ellis, D. G., & Aizenberg, M. R. (2026). Anatomically constrained volumetric smoothing enhances fMRI reliability while avoiding smoothing artifacts. Frontiers in Neuroimaging, 5, Article 1753534. https://doi.org/10.3389/fnimg.2026.1753534

Image Credits: AI Generated

DOI: 10.3389/fnimg.2026.1753534

Keywords: Anatomically, constrained, volumetric, smoothing, enhances, fMRI, reliability, while, avoiding, artifacts, scientific research

The article, authored by Maël Donoso of Ouroboros Neurotechnologies in Lausanne, Switzerland, lays out a framework for what the author calls brain-trained humanoid robots. The core idea builds on the recently proposed concept of brain-trained foundation models, in which neuroimaging data serves as a direct training resource for AI systems rather than a mere source of inspiration. According to this framework, if B(t) represents the brain activity of a human subject over time and A(t) the observable actions of that subject, then neuroimaging data B*(t) can capture elements of brain activity that are not visible in behavior alone. In other words, the brain may contain information about cognition that actions never reveal, and machines could in principle learn from it.

Until now, however, nearly all brain-trained AI research has focused on cognitive processes that are relatively disembodied: vision, audition, language, and reward or error signals recorded while a human passively observes a machine. What has been almost entirely missing, Donoso argues, is the vast territory of embodiment-dependent neural signals, such as those tied to touch, motor control, and physical interaction with the world. The reason is practical. To make sense of neural activity generated by grasping a fragile object or navigating a crowded room, an AI system needs a body that can do roughly the same things. Humanoid robots, though far from equivalent to humans, may now be approaching the level of sensorimotor capability where this becomes possible.

The timing is not coincidental. Robotic arms can now solve Rubik’s cubes and perform fine bimanual manipulations, quadrupedal robots have mastered agile locomotion, and bipedal humanoids can navigate diverse outdoor terrain and even learn sports skills through deep reinforcement learning. Foundation models have added semantic and sensorimotor reasoning on top of these physical abilities, with vision-language-action models combining linguistic knowledge with robotic control. Crucially, researchers have also developed methods for crossing the so-called embodiment gap, transferring skills learned from human demonstrations to robots with entirely different bodies, using both visual and tactile data. If behaviors can cross that gap, the article hypothesizes, perhaps the neural signals underlying those behaviors can cross it too, provided they encode sufficiently high-level representations rather than the low-level singularities of the human body.

The article proposes two computational strategies for putting brain data to work in humanoid robotics. The first, called pattern decoding, aligns the internal representations of an AI model with high-dimensional neural patterns from the human brain. This approach has worked well in vision and audition, where deep networks already show representational similarity to the visual and auditory cortex. The second, called cognitive and computational inference, extracts low-dimensional signals such as reward, error, uncertainty, or surprise, and uses them to modulate reinforcement learning. This is the approach that has dominated brain-trained robotics so far, with EEG-based reward signals successfully guiding robot learning since pioneering work in 2010.

To make the idea concrete, the article presents two worked examples. In the first, a humanoid robot learning complex manipulations, such as handling fragile or deformable objects, could be trained on human demonstrations weighted by neural uncertainty signals. The policy optimization objective would include a term derived from brain activity, mapped through a sigmoid function, that down-weights examples when the human demonstrator was uncertain. Neuroscience offers plausible sources for such signals: the somatosensory cortex can encode uncertainty in sensory processing, motor cortex activity varies with the degree of certainty even when actions are identical, and uncertainty-related patterns have been found in the posterior parietal cortex and cerebellum. In the second example, a robot learning spatial navigation could be guided by neural surprise signals, up-weighting training examples when the human demonstrator was surprised by an outcome, since surprising scenarios are rarer and more informative. The hippocampus and the hippocampal-entorhinal system, which show novelty-related activity during navigation, could provide the raw material.

From these examples, the article expands to a broader map of six humanoid robot functions that could benefit from embodiment-dependent neural signals, ordered by increasing abstraction. Sensorimotor tasks would draw on the somatosensory cortex, motor cortex, posterior parietal cortex, and cerebellum. Spatial navigation would target the hippocampus and entorhinal cortex. Sensorimotor integration, the ability to maintain a functional body schema, would involve the posterior parietal cortex, premotor cortex, and supplementary motor area, with a newly developed motion-compatible PET prototype that can record the brains of walking humans hinting at future possibilities. Embodied alignment would tap valuation networks in the ventral striatum, ventromedial prefrontal cortex, orbitofrontal cortex, and insula to help robots follow human values and preferences in open-ended contexts. Embodied reasoning would engage the dorsolateral prefrontal cortex, anterior cingulate cortex, and frontopolar cortex to improve step-by-step planning under real-world constraints. Finally, social integration would draw on the temporo-parietal junction, medial prefrontal cortex, and superior temporal sulcus to align robots with human social representations.

The practical challenges are substantial. Neuroimaging data is noisy, highly variable across individuals, and only an approximation of true brain activity, and the correspondence between neural signals and cognitive functions must be inferred with care. Portable techniques such as EEG and functional near-infrared spectroscopy would likely be preferred for real-world recording, while fMRI and MEG combined with virtual reality could serve controlled experiments. Invasive techniques, including intracortical arrays and intracranial EEG, offer far higher resolution and signal quality, and have already been used to control robotic arms, deliver tactile sensations, and study navigation-related neural activity, but their restriction to clinical settings limits scalability. The article suggests that invasive recordings might complement non-invasive methods for optimizing a small set of widely shared behaviors, while non-invasive modalities cover the broader diversity of human action. The scarcity of neuroimaging datasets on touch, compared with the abundance available for vision, remains another obstacle, though pre-trained models that generate synthetic neuroimaging data could help fill the gap.

Beyond engineering, the article raises a distinctive ethical concern: excessive anthropomorphism. If robots are trained on human brain data, they may appear more human than they actually are, creating a mismatch between human expectations and the real capabilities of the machines. The author proposes a guiding principle: neural signals should be incorporated only when they improve robot function without undermining transparency about their robotic nature. The stakes are high, but so is the potential payoff. Brain-trained humanoid robots could serve as embodied computational models of human cognition, offering neuroscientists a new experimental platform for testing hypotheses about sensorimotor processing, one of the least understood systems in the brain. In the long term, successful pattern decoding might even allow robots to generate synthetic neuroimaging data at scale. And in a nod to Isaac Asimov’s fictional positronic brains, the article closes with the suggestion that narrowing the distance between humanoid robotics and human neuroscience may one day reveal whether robots themselves dream of electric brains.

Subject of Research: Using human neuroimaging data, especially embodiment-dependent signals like touch, to train humanoid robots

Article Title: Toward brain-trained humanoid robots

Article References: Donoso, M. (2026). Toward brain-trained humanoid robots. Frontiers in Neurorobotics, 20, Article 1903423. https://doi.org/10.3389/fnbot.2026.1903423

Image Credits: AI Generated

DOI: 10.3389/fnbot.2026.1903423

Keywords: brain-trained AI, humanoid robots, neuroimaging, EEG, fMRI, reinforcement learning, embodiment, tactile perception, spatial navigation, alignment, human-robot interaction, Frontiers in Neurorobotics

The review is striking for how little evidence there was to synthesise. Following PRISMA guidelines, the team searched PubMed for studies combining Sudoku with any major neuroimaging technique, screened 289 results through title, abstract and full-text stages, and ended up with just six eligible publications: five using functional magnetic resonance imaging and one using functional near-infrared spectroscopy, or fNIRS. That scarcity is itself a headline. One of the world’s most popular puzzles, played by millions daily since it was popularised in 1980s Japan by the company Nikoli, has almost never been studied with modern brain imaging.

What the six studies do show is remarkably consistent. Across haemodynamic measures of brain activity, Sudoku solving reliably engages the dorsolateral prefrontal cortex and posterior parietal regions, including the intraparietal sulcus, the core components of the frontoparietal control network that underpin working memory and executive function. The dorsal anterior cingulate cortex, a region long implicated in conflict detection and performance monitoring, also appears repeatedly, as does the lateral frontopolar cortex, which the review links to metacognitive control, the ability to adjust decisions based on uncertainty.

The technical details of the included studies reveal how researchers have squeezed the puzzle into the scanner. Qin and colleagues in 2012 asked healthy young adults to solve simplified 4×4 Sudoku grids during fMRI, tracking blood-oxygen-level-dependent responses in predefined regions of interest including the prefrontal cortex, posterior parietal cortex, fusiform gyrus, caudate and dorsal anterior cingulate. All five regions showed task-positive BOLD responses, linking Sudoku to heuristic problem solving. Jin and colleagues in 2012 studied elderly participants performing simple and complex rule-based Sudoku tasks and found something different: deactivation of the posterior cingulate cortex and precuneus, key nodes of the default mode network, with greater suppression during more complex puzzles.

That default mode suppression is a technically important finding. The posterior cingulate and precuneus are most active when the mind wanders internally, and their deactivation during demanding Sudoku solving suggests the puzzle pulls the brain firmly into externally directed, goal-oriented cognition. The extent of deactivation scaled with task complexity, meaning the harder the puzzle, the more the brain suppressed its internally oriented circuitry. This pattern mirrors what is seen in other demanding cognitive tasks and reinforces the review’s argument that Sudoku is a genuine probe of executive engagement rather than a passive pastime.

Two later fMRI studies pushed the paradigm further by examining metacognition. Qiu and colleagues in 2018 used a re-decision paradigm in which participants made an initial Sudoku judgment and then revisited it, implicating the lateral frontopolar cortex in metacognitive control and the dorsal anterior cingulate and anterior insular cortex in metacognitive monitoring. Intriguingly, the ventral striatum was uniquely active during Sudoku, tracking reductions in decision uncertainty, and its functional coupling with the frontopolar cortex predicted how much individuals improved their accuracy. Su and colleagues in 2022 confirmed the dorsal anterior cingulate’s role, showing increased activity when participants re-evaluated their initial answers, alongside the dorsolateral prefrontal cortex and intraparietal sulcus.

The single fNIRS study broadened the ecological picture. Ashlesh and colleagues in 2020 fitted 33 healthy adults with a 16-channel forehead sensor array measuring oxygenated and deoxygenated haemoglobin over the prefrontal cortex, then had them solve paper-and-pencil Sudoku progressing from isolated 3×3 subgrids to a full 9×9 matrix. Both medial and lateral prefrontal regions showed increased activation, scaling with task complexity. Because fNIRS is lightweight and tolerant of movement, it captures Sudoku solving under far more natural conditions than fMRI, and the converging prefrontal findings suggest the scanner results are not artefacts of an artificial laboratory setup.

The review also situates Sudoku against established cognitive training paradigms. The n-back task isolates continuous updating of information under time pressure, and mental arithmetic emphasises numerical manipulation and speeded processing, but Sudoku integrates multiple executive processes within a single task: visuospatial working memory, rule maintenance, hypothesis testing, conflict monitoring and metacognitive evaluation. That composite structure more closely resembles real-world problem solving, where demands shift dynamically. Sudoku also differs from chess, which the authors note in a companion review, because chess places stronger demands on visual-perceptual expertise and occipito-temporal regions, whereas Sudoku offers a more flexible, rule-based probe of working memory and cognitive control, and is typically played individually without competitive pressure.

Translational potential is where the review becomes genuinely provocative. In the only longitudinal study included, Nombela and colleagues in 2011 put patients with Parkinson’s disease through a six-month Sudoku-based cognitive training programme and found enhanced activation in frontal, temporal and parietal regions afterwards, interpreted as a marker of functional reorganisation and neuroplasticity. The authors argue that Sudoku’s accessibility, familiarity and intrinsic motivation could make it a low-cost vehicle for maintaining executive function in older adults, particularly since motivation is a critical determinant of adherence in cognitive training interventions. Portable fNIRS extends this viability to bedside, home-based and rehabilitation contexts.

Yet the authors are careful not to oversell. With only six studies, small samples, heterogeneous designs and just one longitudinal clinical investigation, the evidence base cannot yet establish whether Sudoku training produces benefits that generalise beyond the puzzle itself. Most studies relied on cross-sectional, single-session imaging, and no study employed EEG, leaving the temporal dynamics of Sudoku-related cognition unexplored. The review calls for larger and more diverse samples, harmonised task protocols contrasting grid sizes and decision phases, and multimodal imaging combining fMRI, fNIRS and EEG within unified designs. If those gaps are filled, Sudoku may earn a place not just in the newspaper but in the clinical toolkit, as a window into the neural architecture of executive function and a practical instrument for supporting cognitive health across the lifespan.

Subject of Research: Neuroscience

Article Title: Neural correlates of Sudoku play: a systematic review of brain imaging studies

Article References: Williams, M. J., Williamson, E. J., & Brooks, S. J. (2026). Neural correlates of Sudoku play: a systematic review of brain imaging studies. Frontiers in Neuroimaging, 5, Article 1756394. https://doi.org/10.3389/fnimg.2026.1756394

Image Credits: AI Generated

DOI: 10.3389/fnimg.2026.1756394

Keywords: Sudoku, neuroimaging, fMRI, fNIRS, executive function, working memory, cognitive training, systematic review, Neural, correlates, play, systematic

The research team, led by Yu Yang, Huixia Jia and He Liu of Beijing Normal University’s School of Systems Science, used a chronic unpredictable mild stress (CUMS) protocol in adult male C57BL/6J mice and captured resting-state functional MRI at three time points: baseline, day five, and day ten of the stress regimen. By integrating four complementary analytical approaches—fractional amplitude of low-frequency fluctuations (fALFF), static functional connectivity, graph theory, and dynamic connectivity analyses—the study maps what the authors describe as a hierarchical spatiotemporal dissolution of brain function.

The CUMS protocol itself was deliberately punishing in its unpredictability. Mice endured a rotating cast of stressors including two-hour restraint, water and food deprivation, forced swimming, tail clipping, cold exposure at 4 degrees Celsius, heat exposure at 45 degrees, day-night reversal, cage tilting, and random electric foot shocks, all pseudo-randomized across the ten-day schedule to prevent habituation. Animals were housed singly to prevent social buffering, and experimenters remained blinded during behavioral and imaging procedures. Ten mice began the study; three died during imaging sessions, and the final balanced longitudinal cohort comprised seven animals, each serving as its own baseline control.

Imaging was performed on a 9.4-tesla Bruker BioSpec small-animal MRI scanner under carefully controlled anesthesia with continuous physiological monitoring. The preprocessing pipeline was rigorous: slice-timing correction, rigid-body motion realignment, manual skull stripping, nonlinear normalization to a standard mouse brain template, nuisance regression of motion parameters and white matter and cerebrospinal fluid signals, and band-pass filtering between 0.01 and 0.08 hertz. Functional networks were built from 391 anatomical regions of interest defined by the Allen Mouse Brain Common Coordinate Framework.

The first layer of the story emerges at the voxel level. Sixteen distinct brain clusters showed significant longitudinal changes in spontaneous activity, converging on three critical systems: the caudoputamen, the medial prefrontal cortex, and the hippocampal formation. The caudoputamen, a striatal region vital for sensorimotor and reward integration, showed a progressive decline in fALFF from baseline to day ten, suggesting disruption of the motor-motivation interface. The medial prefrontal cortex followed a striking biphasic trajectory: an initial increase from day zero to day five, consistent with compensatory hyper-reactivity, followed by a significant decline by day ten, pointing to failing top-down emotional regulation. The hippocampal formation, long known to be vulnerable to glucocorticoid-mediated damage, showed significant decreases in spontaneous activity in the dentate gyrus and CA3 field during the later stress phase.

As local signals faltered, the disruption radiated into circuit-level disconnection. Seed-based analyses revealed a loss of bilateral synchronization within the striatum itself, an early breakdown of the motor-motivation interface. More intriguingly, the caudoputamen showed progressive uncoupling from the posteromedial visual area, a sensory-limbic disconnection the authors link to the sensory blunting often observed in chronic mood disorders. The prefrontal cortex showed altered connectivity with the anterior olfactory nucleus, highlighting a disruption in the prefrontal-olfactory axis, while the hippocampal seed revealed dysconnectivity along the hippocampal-prefrontal pathway, which integrates contextual and emotional information and is known to be vulnerable to chronic stress. Together, these findings suggest that stress engages a broader prefrontal-hippocampal-olfactory network rather than isolated regions.

At the level of whole-brain topology, the picture became subtler. Graph theoretical analysis of the 391-node connectome showed that global node efficiency was largely preserved, but local measures told a different story: seven regions showed significant losses of node local efficiency and two regions showed reduced clustering coefficients. In other words, chronic stress primarily erodes the brain’s modular, cliquish structure—the specialized local processing that supports resilience—while temporarily sparing the global communication backbone. The authors interpret this as an adaptive triage: the brain protects its long-range integration pathways at the expense of local processing capacity, a pattern consistent with the concept of cognitive reserve and with the idea that functional decline manifests first as a loss of modular complexity rather than total communication breakdown.

Perhaps the most consequential finding lies in the temporal domain. Using a sliding-window approach with K-means clustering, the team identified five recurring dynamic states of regional activity and connectivity. As stress progressed, the fractional occupancy of high-activity states declined significantly while occupancy of low-activity states increased, accompanied by a marked drop in state transition frequency. By day ten, the dynamic connectivity landscape had become strikingly uniform, dominated by a single low-connectivity state. The healthy brain, normally a metastable system flickering fluidly between functional configurations, had become dynamically rigid—trapped in a constrained, low-entropy configuration that the authors liken to critical slowing down, a phenomenon observed in complex systems on the verge of irreversible state transitions.

This dynamic rigidity offers a mechanistic framework for understanding why chronic depression becomes increasingly resistant to treatment: a brain that has lost the flexibility to reconfigure its networks may also have lost the capacity to respond to pharmacological or psychological intervention. The authors frame the ten-day progression as an example of allostatic overload—cumulative wear and tear on the brain’s regulatory systems—and note that the sensory decoupling they observed may initially serve a protective function by reducing allostatic load, only to ultimately isolate the brain and reinforce depressive rumination and behavioral despair.

The study’s limitations are candidly acknowledged. With no separate non-stressed control group, baseline day-zero scans served as within-subject controls, which cannot fully disentangle stress-specific effects from repeated anesthesia, repeated MRI exposure, single housing, or aging over the brief protocol. The small cohort of seven mice, the exclusive use of males, and the inability of fALFF to reveal the metabolic or cellular underpinnings of the observed changes all warrant caution. The authors call for future work incorporating diffusion tensor imaging and post-mortem histology to determine whether dynamic rigidity is purely functional or underpinned by cellular and white matter alterations, and for validation in independent rodent cohorts, both sexes, and longitudinal human resting-state fMRI studies of first-episode and high-risk patients. If the prefrontal-olfactory dysconnectivity and the onset of dynamic rigidity can be replicated, they may become candidate imaging biomarkers for staging depression—opening a window for stage-specific therapeutic intervention before the brain’s temporal flexibility is permanently lost.

Subject of Research: Longitudinal fMRI mapping of progressive brain network dysfunction in a chronic stress mouse model of depression

Article Title: Neuroprogression of chronic stress-induced depression via longitudinal fMRI

Article References: Yang, Y., Jia, H., & Liu, H. (2026). Neuroprogression of chronic stress-induced depression via longitudinal fMRI. Frontiers in Psychiatry, 17, Article 1898694. https://doi.org/10.3389/fpsyt.2026.1898694

Image Credits: AI Generated

DOI: 10.3389/fpsyt.2026.1898694

Keywords: depression, chronic stress, fMRI, neuroprogression, functional connectivity, graph theory, brain networks, CUMS model, dynamic rigidity, hippocampus, prefrontal cortex, neuroimaging

The research, published in Advances in Drug and Alcohol Research, followed 115 drug-naive adolescents aged 11 to 12 years, all carrying clinician-diagnosed externalizing psychopathology. Seventy-eight participants were male and 37 were female. Because prior longitudinal work on this population has either centered exclusively on male samples or ignored sex altogether, the study addresses a genuine blind spot in addiction science: how the neural machinery of risky decision-making relates to the emergence of problematic substance use differently in females and males with the same diagnoses.

During a magnetic resonance imaging session on a research-dedicated 3.0-Tesla Siemens Prisma scanner, participants performed the Balloon Analogue Risk Task, a well-validated behavioral paradigm that models real-world risk-taking. In the task, adolescents repeatedly chose between inflating a virtual balloon—earning escalating cash rewards but risking an explosion that wipes out the accumulated money—or banking the winnings safely. The probability of explosion rises parametrically with each inflation, from roughly 2 percent at the first pump to nearly 90 percent at the twelfth, allowing researchers to separate brain responses that track escalating risk from those that occur regardless of risk level.

Functional imaging data were preprocessed with distortion correction, motion correction, spatial smoothing, and independent components analysis-based denoising to strip out noise artifacts. The team then isolated two a priori regions of interest: the nucleus accumbens, a core node of reward processing that ascribes emotional and motivational value to choices, and the subgenual anterior cingulate cortex, a region involved in loss avoidance, uncertainty, and the affective evaluation of potential negative outcomes. For each participant, the researchers computed contrasts comparing risky choices against safe ones, both parametrically modulated by explosion probability and unmodulated, and then related standardized activation in these regions to subsequent substance outcomes.

Following the baseline scan, participants and their guardians completed questionnaires every six months for up to 84 months, assessing substance use through a revised Drug Use Screening Inventory covering 25 substance classes. Problematic substance use was operationalized as the endorsement of two or more DSM-5-based consequences for any single substance, reported by either the adolescent or the guardian. Cox proportional hazards models, adjusted for family history of substance use disorder, violence exposure, and parental monitoring, then tested whether brain activation predicted the hazard of reaching that outcome over time.

The central finding concerned the right nucleus accumbens. Higher activation during risky choices as explosion probability increased—the modulated Choose Inflate minus Choose Win contrast—was associated with a lower hazard of problematic substance use across the whole sample, with a hazard ratio of 0.68. Critically, this protective association was significant only in females, who showed a hazard ratio of 0.64, while males showed no significant relationship, and the difference between the sexes’ hazard ratios was itself statistically significant. In other words, girls whose accumbens activation scaled appropriately with rising risk were substantially less likely to develop problematic substance use; boys’ risk-modulated accumbens signaling carried no such protection.

The pattern reversed for unmodulated activation, which ignores the balloon’s risk level. In males, higher overall accumbens activation when choosing to inflate rather than bank was linked to a lower hazard of problematic substance use, with a hazard ratio of 0.60, whereas the same measure was unrelated to outcomes in females. Meanwhile, in the right subgenual anterior cingulate cortex, higher unmodulated activation during risky versus safe choices predicted a lower hazard in females—hazard ratio 0.49—while males trended in the opposite direction, and the sex difference in hazard ratios was significant. Notably, had the researchers analyzed the sample without splitting by sex, the subgenual cingulate finding would have been completely obscured.

The authors interpret these results through the lens of sex-specific adolescent brain maturation. Neuroimaging work has shown that ventral striatal volume rises and then declines in males across adolescence but declines steadily in females, and prior studies have linked larger accumbens volume to sensation seeking and later drinking in alcohol-naive boys but not girls. The new findings suggest that in females with externalizing disorders, appropriately ascribing emotional and motivational valence to choices as risk escalates may constitute a biologically protective system, whereas in males, general engagement of the accumbens during any risky choice—regardless of risk magnitude—appears protective. For the subgenual cingulate, the female-specific result points to intact loss-avoidance signaling during decision-making as a safeguard against later substance problems, a function that may be absent or dysfunctional in boys with the same diagnoses.

Secondary behavioral analyses reinforced the sex divide. Measures of reward sensitivity, derived from the average number of inflations on unexploded balloons, and of reduced loss aversion, derived from continued inflations after explosions, each predicted higher hazard of problematic substance use in the overall sample—but these relationships were driven entirely by males, in whom greater reward sensitivity more than doubled the hazard. The findings align with established evidence that males typically display higher reward sensitivity and lower loss aversion than females, tendencies shaped by a combination of evolutionary pressures and cultural norms around risk.

The study’s implications reach directly into prevention. The authors propose that behavioral interventions targeting decision-making deficits—such as the Impulsive Decision Reduction Training for Youth program developed by their group—could be tailored by sex: for females, therapy might focus on strengthening the assignment of emotional and motivational valence to choices as risk increases, while for males, it might emphasize overall risk calculation and recognition of varying risk signals across different choices. The researchers caution that none of the findings survived correction for multiple comparisons, that the sample contained fewer females than males reflecting the real-world sex distribution of externalizing disorders, and that substance outcomes relied on self- and guardian-report rather than objective toxicology. Even so, the work offers a striking demonstration that the neural signatures of risky decision-making are not one-size-fits-all, and that effective substance use prevention for high-risk adolescents may need to be built differently for girls and boys from the very start.

Subject of Research: Sex differences in reward network activation are linked to problematic substance use among high-risk adolescents

Article Title: Sex differences in reward network activation are linked to problematic substance use among high-risk adolescents

Article References: Murray, O. K., Mattey-Mora, P. P., Aloi, J., Lovins, S., Smoker, M. P., & Hulvershorn, L. A. (2025). Sex differences in reward network activation are linked to problematic substance use among high-risk adolescents. Advances in Drug and Alcohol Research, 5, Article 14591. https://doi.org/10.3389/adar.2025.14591

Image Credits: AI Generated

DOI: 10.3389/adar.2025.14591

Keywords: adolescent substance use, risky decision-making, sex differences, nucleus accumbens, subgenual anterior cingulate cortex, externalizing disorders, fMRI, longitudinal study, differences, reward, network, activation

The research team, led by Marlene Kundlacz and Sarah Gerhardt at the Central Institute of Mental Health in Mannheim together with colleagues at Heidelberg University and the University of Mannheim, recruited fifty men who regularly engaged in heavy episodic drinking, defined by the World Health Organization as consuming at least sixty grams of pure alcohol on a single occasion at least monthly. The scale of the problem they were targeting is considerable: in Europe, 56.5 percent of men over fifteen report such heavy drinking episodes, compared with 24.5 percent of women, and roughly 42 percent of people with heavy episodic drinking eventually develop alcohol use disorder.

The scientific logic behind the experiment rests on a chain of prior observations. Higher testosterone levels have been linked to greater alcohol consumption, and patients with alcohol use disorder show elevated testosterone compared with healthy controls. Dihydrotestosterone, which is produced from testosterone by the enzyme 5-alpha reductase II, binds far more strongly to the androgen receptor and is up to six times more potent than testosterone itself. Critically, a rise in DHT during acute withdrawal treatment has been shown to predict worse twelve-month outcomes in patients with alcohol use disorder, including earlier and more frequent alcohol-related hospital readmissions. A separate clinical trial also found that dutasteride, another 5-alpha reductase inhibitor, reduced alcohol intake in heavy-drinking men, making the pharmacological suppression of DHT a promising but poorly understood therapeutic avenue.

To probe the underlying neural machinery, the researchers designed a rigorous crossover trial registered as DRKS00020569. Each participant attended four study visits. At the first and third visits they received either five milligrams of finasteride or a placebo, and the day after each dose they completed a stop-signal task inside a 3 Tesla MRI scanner at the Central Institute of Mental Health. In this classic test of behavioral control, participants pressed buttons in response to left- or right-pointing arrows, but on twenty percent of trials an upward-pointing arrow appeared shortly afterward, signaling them to abort the already initiated movement. The task lasted just under eleven minutes and comprised 320 go trials and 80 stop trials, with successful inhibition rates held between forty and sixty percent, a range that ensures the task is challenging but achievable.

Blood samples drawn at every visit allowed the team to quantify serum DHT and testosterone using enzyme-linked immunosorbent assays. The single finasteride dose reduced DHT levels by 9.6 percent on average, from 353 to 318 picograms per milliliter, a statistically significant drop, while testosterone levels remained unchanged. That reduction was far smaller than the roughly eighty percent suppression documented with long-term finasteride treatment, a discrepancy the authors attribute to the single-dose design and the timing of blood sampling, which occurred on average nearly twenty-two hours after ingestion, potentially past the point of maximal DHT suppression.

The imaging analysis focused on forty-seven participants with usable data, applying whole-brain flexible factorial models that accounted for the crossover structure of the design, with medication, visit sequence, and time point as factors and individual hormone concentrations as covariates of interest. The results were striking in their specificity. Lower DHT levels were associated with stronger activation in the left middle temporal gyrus when participants failed to inhibit a response, while higher testosterone levels predicted stronger activation in the right middle temporal gyrus during successful inhibitions. Both hormones also showed positive associations with temporal activation when successful stops were contrasted directly against failed ones.

These hemispheric patterns fit neatly with existing models of how the stopping brain is organized. Prior work suggests the right hemispheric network is more involved in attending to the stop signal, whereas the left hemispheric network supports the inhibition of the response itself. The authors therefore propose that testosterone may act primarily on the attentional component of stopping, while DHT relates to the inhibitory process proper. Supporting this interpretation, testosterone has previously been found to correlate positively with brain activation during spatial tasks and mental rotation, which also demand attention, and fetal testosterone exposure has been shown to predict gray matter volumes in the left middle temporal gyrus, the very region where the DHT effect emerged.

Yet the study also delivered a sobering null result: neither the finasteride challenge nor the hormone levels produced any significant change in behavioral performance on the stop-signal task, whether measured as stop-signal reaction time, go reaction time, or stop-signal delay. The authors note that changes in neural activation do not necessarily translate into measurable behavioral differences, as subtle shifts in the recruitment of task-related networks can be detectable with functional MRI while remaining too weak to alter performance. The modest 9.6 percent DHT reduction may simply have been too small to push behavior, and the test-retest reliability of the stop-signal task varies across contrasts and brain regions, which can obscure small medication effects.

The researchers are careful to acknowledge that finasteride’s effects may extend beyond simple androgen suppression. The drug also acts on dopaminergic neurotransmission in the mesolimbic system, influences GABAergic signaling, and has been suggested to interfere with the hypothalamus-pituitary-adrenal axis, potentially reducing stress reactivity and impulsive behavior. Chronic alcohol consumption can additionally promote the aromatization of androgens into estradiol, which has independently been implicated in male problem drinking, meaning the observed testosterone effects could partly reflect estrogenic pathways. Speculatively, the authors also note that testosterone shapes social vigilance and trust, and since alcohol is often used to facilitate social interaction, androgen modulation might have influenced neural activation through social-behavioral routes in the unfamiliar experimental setting.

Despite these caveats, the study stands as the first experimental examination of DHT and the interplay between androgens and inhibitory control in heavy-drinking individuals. The clinical implications are tantalizing: if pharmacological modulation of DHT can shift the neural correlates of response inhibition, and if 5-alpha reductase inhibition has already been shown to improve abstinence and drinking outcomes in some heavy-drinking men, then androgen-targeted therapies could eventually join the arsenal against alcohol use disorder. The authors caution that such applications remain premature. Finasteride is generally well tolerated but has been associated with decreased libido, erectile dysfunction, and, less commonly, mood-related symptoms, and persistent sexual dysfunction has been reported in a subset of patients after discontinuation, though the evidence remains heterogeneous. Larger samples, female participants, clinical populations, and longer treatment durations will all be needed before hormone modulation can be seriously considered as a treatment for problem drinking. For now, the study provides a compelling proof of concept that the brain’s braking system is chemically entangled with the body’s most potent male hormones.

Subject of Research: Effects of pharmacologically reduced dihydrotestosterone on brain activation and response inhibition in male heavy drinkers

Article Title: Pharmacologically altered androgens, brain activation and response inhibition in a stop-signal task in male heavy drinkers

Article References: Kundlacz, M., Boroumand-Jazi, R., Hoffmann, S., Mühle, C., Reinhard, I., Reichl, M., Wenger, L., Tan, H., Bach, P., Vollstädt-Klein, S., Alpers, G. W., Kiefer, F., Lenz, B., & Gerhardt, S. (2026). Pharmacologically altered androgens, brain activation and response inhibition in a stop-signal task in male heavy drinkers. Frontiers in Systems Neuroscience, 20, Article 1907416. https://doi.org/10.3389/fnsys.2026.1907416

Image Credits: AI Generated

DOI: 10.3389/fnsys.2026.1907416

Keywords: dihydrotestosterone, testosterone, finasteride, response inhibition, stop-signal task, fMRI, heavy episodic drinking, alcohol use disorder, impulsivity, androgens, temporal cortex, clinical trial

The core problem Kafkas identifies is one that has frustrated cognitive neuroscientists for years. Animal studies have delivered exquisitely detailed accounts of how dopamine and norepinephrine shape hippocampal plasticity, influencing long-term potentiation, memory persistence and the prioritisation of behaviourally significant events. Work on reward, novelty and surprise suggests that catecholaminergic modulation is not one unitary arousal response but a family of state-dependent mechanisms through which the brain selects experiences for long-term storage. Yet translating those mechanistic insights into human research remains strikingly difficult, because direct measures of endogenous catecholaminergic activity in a living human brain are nearly impossible to obtain during the fleeting seconds in which memories are actually formed.

Existing tools each capture only a partial slice of the underlying biology. Positron emission tomography offers receptor- and transporter-specific measurements, but its temporal resolution is poor, it is costly and it involves radiation exposure. Neuromelanin-sensitive magnetic resonance imaging reveals the structural integrity of catecholaminergic nuclei, most famously the locus coeruleus, the brain’s principal norepinephrine factory, but says nothing about their moment-to-moment functional dynamics. Pharmacological manipulations can establish causal links between neurotransmitters and cognition, but they lack anatomical specificity and are confounded by systemic effects scattered throughout the body. What the field needs, Kafkas argues, is a biomarker that is non-invasive, inexpensive, temporally sensitive, easy to combine with imaging and modelling, and capable of tracking trial-by-trial fluctuations in neuromodulatory state.

Pupillometry ticks every one of those boxes, and an expanding literature shows that pupil dynamics genuinely matter for memory. Encoding-related pupil responses predict later recognition, memory strength, associative learning and mnemonic discrimination across a wide range of paradigms. They even distinguish qualitative features of remembering, separating recollection from familiarity, novelty from known, and in one striking study true memories from false ones. These are not trivial byproducts of effort or alertness, Kafkas contends, but physiological signatures associated with whether an experience gets encoded at all. Crucially, pupil responses also track the computational variables at the heart of hippocampal learning: novelty, expectation and prediction error all modulate pupil behaviour during encoding, and those modulations predict different subsequent memory outcomes.

The provocative part of the new framework is its refusal to treat the pupil as a narrow proxy for locus coeruleus activity. Pupillary responses emerge through the coordinated action of sympathetic and parasympathetic pathways and are shaped by distributed cortical, subcortical and brainstem circuits. Recent simultaneous pupillometry-fMRI evidence supports that broader view. During incidental encoding, pupil responses covaried with hippocampal and parahippocampal activity and predicted later recognition, and successful encoding was associated with pupil-related functional interactions between the hippocampus, the Edinger-Westphal nucleus, the ventral striatum and the orbitofrontal cortex. Pupil-linked memory effects, in other words, ride on a network spanning hippocampus, cortex and brainstem, not on a single neuromodulatory nucleus. Kafkas is careful about causality: these associations do not establish the direction of influence, and they may reflect direct interactions or common upstream drives, or both.

The framework’s boldest proposal tackles an apparent contradiction in the literature. Many studies report that larger pupil dilations during encoding predict successful memory. Others have found that encoding-related pupil constriction can predict memory too, under specific conditions. Rather than dismissing one set of findings, Kafkas suggests that constriction and dilation may index genuinely different physiological states recruited by different encoding demands. In an expectation-learning experiment, unexpected novel stimuli produced steadily increasing dilation and preferentially boosted later recollection, while expected novel stimuli elicited constriction yet still predicted successful recognition. Both routes led to memory, but through different physiological configurations.

Those two configurations map neatly onto two computational modes of hippocampal learning. When new information broadly fits existing knowledge, successful encoding may depend on integrating it into established representations, a state that plausibly leans more on parasympathetic and autonomic brainstem processes, consistent with the link between expected novelty, constriction and hippocampal connectivity with the Edinger-Westphal nucleus, the parasympathetic hub controlling the pupil’s constrictor muscle. When events violate predictions and demand revision of internal models, encoding may instead rely more heavily on catecholaminergic hippocampal-midbrain interactions. Here dilation during surprise reflects rapid memory updating rather than mere arousal, an idea compatible with the GANE model, in which norepinephrine amplifies high-priority neural representations under salient conditions, but extending it to hippocampal-midbrain network states. Animal work adds a further nuance: rapid pupil fluctuations track noradrenergic activity closely, whereas sustained dilation during locomotion tracks cholinergic activity instead. One pupil response, multiple possible neuromodulatory backdrops.

Turning the framework into hard science will require multimodal and, ultimately, causal tests, and Kafkas lays out a concrete experimental agenda. High-resolution fMRI can relate trial-by-trial pupil fluctuations to hippocampal subfields and network connectivity. Neuromelanin-sensitive MRI can index individual differences in catecholaminergic nuclei, despite ongoing challenges in accurately localising the locus coeruleus. Pharmacological studies can test whether cholinergic, dopaminergic and noradrenergic manipulations differentially alter constriction- and dilation-linked memory states. Magnetic resonance spectroscopy can probe the excitatory-inhibitory balance of the hippocampus, and autonomic measures such as heart-rate variability and skin conductance can help tease apart sympathetic and parasympathetic contributions. Computational modelling supplies the final ingredient: instead of simple remembered-versus-forgotten contrasts, future studies should estimate latent variables like surprise, uncertainty and belief updating on each trial, then relate them simultaneously to pupils, hippocampal activity and later memory quality.

The falsifiable predictions are explicit. Manipulating expectation while holding stimulus novelty constant should shift encoding-related pupil responses between constriction-dominated and dilation-dominated profiles, and those shifts should be accompanied by changes in hippocampal-midbrain connectivity rather than locus coeruleus activation alone. Drug studies should perturb the two profiles differently if they truly reflect distinct neuromodulatory states. The framework would falter, by contrast, if a single arousal process could explain everything, or if the dissociations failed to replicate beyond the small set of paradigms that currently motivate them. Kafkas acknowledges this limitation candidly, calling for independent replication and cross-species work in which circuit-level manipulations could determine whether pupil profiles are causally tied to specific neuromodulatory mechanisms or merely parallel consequences of broader physiological states.

The stakes extend well beyond the laboratory. Altered catecholaminergic signalling and hippocampal dysfunction are central features of healthy ageing, Alzheimer’s disease and several neuropsychiatric conditions, and recent evidence links pupil dynamics to functional network integrity in ageing and to Alzheimer’s disease biomarkers. Because pupillometry is cheap, non-invasive and endlessly repeatable, it could help characterise individual differences in neuromodulatory function during memory formation, not as a standalone diagnostic but as one layer in multimodal profiles combining memory performance, network integrity and disease progression. Whether every specific hypothesis survives testing or not, the reframing itself may prove durable: by treating the pupil as a systems-level marker of integrated neuromodulatory, autonomic and hippocampal processes, the field gains a scalable route for connecting molecular mechanisms of dopamine and norepinephrine to the human memories that depend on them, across health, ageing and disease.

Subject of Research: Pupillometry as a systems-level biomarker of neuromodulatory influences on human hippocampal memory

Article Title: Beyond the locus coeruleus: pupil dynamics as a systems-level marker of neuromodulatory influences on hippocampal memory

Article References: Kafkas, A. (2026). Beyond the locus coeruleus: pupil dynamics as a systems-level marker of neuromodulatory influences on hippocampal memory. Frontiers in Molecular Neuroscience, 19, Article 1976085. https://doi.org/10.3389/fnmol.2026.1976085

Image Credits: AI Generated

DOI: 10.3389/fnmol.2026.1976085

Keywords: pupillometry, hippocampus, dopamine, norepinephrine, locus coeruleus, episodic memory, neuromodulation, novelty, prediction error, memory encoding, autonomic nervous system, fMRI