The research, led by Florian Giesche of Goethe University Frankfurt together with colleagues in Germany and Luxembourg, enrolled 231 community-dwelling adults aged 60 and above, of whom 103 reported at least one fall in the previous twelve months. All participants were functionally independent and living at home, the population in which subtle early warning signs are hardest to detect. The investigators deliberately excluded falls with obvious non-motor causes such as dizziness, fainting, or loss of consciousness, ensuring that the comparison between fallers and non-fallers reflected genuine differences in movement control and cognition rather than unrelated medical events. Participants completed a comprehensive assessment program spanning established motor tests, seated computer-based cognitive tasks, and a novel interactive test battery performed on a device called the SKILLCOURT.

The centerpiece of the new approach was the Random Star Run, a reactive agility assessment in which participants started in the center of a four-by-four or five-by-five meter court and sprinted to eight outer target fields that lit up in random, unpredictable order on a large screen. A LiDAR system sampling at 40 hertz continuously tracked each participant’s position, allowing the device to measure total completion time automatically and objectively. Because the target sequence was never repeated, participants could not plan their movements in advance; every sprint, deceleration, and change of direction had to be initiated in reaction to a visual stimulus. This unplanned, reactive quality is precisely what distinguishes agility from simple change-of-direction speed, and it mirrors the demands of everyday situations such as dodging a pedestrian, stepping around an obstacle, or recovering from an unexpected stumble.

Alongside the agility test, participants performed motor-cognitive stepping tasks while standing on the court. In the simple reaction task they executed rapid sidesteps toward a target field whenever an orange square appeared; in the choice reaction task they had to interpret color-coded stimuli and step left or right accordingly. Three further stepping tasks probed executive functions: a task-switching paradigm assessing cognitive flexibility, a 2-back task taxing working memory, and a Stroop word-color condition measuring interference control. Performance on these tasks was quantified with an inverse efficiency score that combined response time and error rate, and trials with error rates above 30 percent were excluded to guarantee valid task execution. For comparison, the same cognitive functions were also assessed in a conventional seated setup using keyboard responses, while motor function was evaluated with the Timed-Up-and-Go test, the 30-second Sit-to-Stand test, grip strength, and walking speed, the latter three performed under both single-task and dual-task conditions with concurrent backward counting.

The statistical analysis was rigorous and pre-specified. General linear models adjusted for age and study location, with interaction terms to detect site-specific heterogeneity, were used to compare fallers and non-fallers across every outcome, with p-values Holm-adjusted within each domain. Only two measures survived this stringent filtering: the Random Star Run and the simple stepping reaction task. Participants without a history of falls completed the agility course and responded to the simple stepping stimuli significantly faster than those who had fallen, with partial eta-squared effect sizes of roughly 0.04. Strikingly, none of the established motor or cognitive assessments, including the Timed-Up-and-Go, Sit-to-Stand, walking speed, dual-task variants, grip strength, and the seated cognitive battery, showed group differences that met the predefined threshold, with effect sizes ranging only from 0.001 to 0.013.

The discriminative power of the agility test was then quantified with adjusted logistic regression and receiver operating characteristic analysis. Each standard-deviation increase in Random Star Run completion time was associated with an 82 percent increase in the odds of having a fall history, an odds ratio of 1.82 that remained significant after adjustment. The resulting model achieved an area under the curve of 0.73, with 70 percent sensitivity, 71 percent specificity, and roughly 70 percent overall classification accuracy. Crucially, this significantly outperformed the Timed-Up-and-Go test, the most widely used functional mobility screen in fall clinics, which managed an AUC of only 0.63. Adding the agility measure to a model containing age and study location alone improved discrimination from 0.63 to 0.73, a statistically significant increment, and bootstrap validation with 1,000 resamples showed only modest shrinkage to an optimism-corrected AUC of 0.70, indicating limited overfitting. An exploratory clinical cutoff of 28.6 seconds was derived from unadjusted data for practical interpretation.

The simple stepping reaction task told a more nuanced story. It was likewise associated with fall history, with an odds ratio of 1.58 and an adjusted AUC of 0.72 at 63 percent sensitivity and 75 percent specificity. However, when compared directly with the corresponding seated PC-based simple reaction test, which achieved an AUC of 0.63, the stepping version did not demonstrate statistically superior discrimination. The same held for the choice reaction task. In other words, the added value of reactive stepping over conventional cognitive testing could not be confirmed, and the authors’ second hypothesis was not supported. The executive-function stepping tasks, including the 2-back and Stroop conditions, showed no relevant group differences at all, though substantial missing data, particularly for the demanding 2-back task, may have reduced statistical power and complicated interpretation.

One of the most intriguing findings emerged from the interaction between age and fall history on agility performance. The gap between fallers and non-fallers was widest among the younger-old participants but steadily narrowed with advancing age, converging and even reversing at approximately 80 years. This suggests that reactive agility testing may be most informative in relatively high-functioning adults in their sixties and seventies, precisely the group in which conventional clinical thresholds, such as a Timed-Up-and-Go time of 13.5 seconds, often fail because most participants clear them easily. Meta-analytic evidence cited in the paper indicates the Timed-Up-and-Go achieves only about 31 percent sensitivity at that cutoff, leaving a large share of subtly impaired older adults undetected. Ceiling effects in higher-functioning populations appear to blunt the established tests, whereas the physically and cognitively demanding agility task retains the resolution to separate the groups.

The authors are careful to frame these results as a cross-sectional snapshot rather than proof of predictive power. Fall history was assessed retrospectively by self-report, which carries a risk of recall bias, and the study cannot establish whether poor agility precedes falling or reflects its consequences, including heightened fear of falling, which was itself elevated among fallers. Differences in testing organization between the German and Luxembourg sites, and the exclusion of participants with high error rates, introduce further caveats, although sensitivity analyses adjusting for sex, education, and physical functioning largely confirmed the stability of the main findings. The research team calls for prospective longitudinal studies to determine whether reactive agility can predict future first and recurrent falls beyond established measures, whether it adds value over planned change-of-direction tests without a reactive component, and whether specific cognitive elements such as response inhibition enhance stepping-based paradigms.

If those prospective studies succeed, the implications for fall prevention could be substantial. Agility-based assessments could be folded into multicomponent screening batteries for community-dwelling seniors, and similar paradigms could be implemented with more accessible light-sensor systems rather than dedicated laboratory equipment. Beyond diagnosis, the same reactive movement demands that make the test diagnostic may point toward a new generation of training interventions emphasizing stop-and-go movements, rapid directional changes, and decision-making under time pressure, capacities that conventional strength and balance programs rarely challenge. For a field that has long assessed the aging body and the aging mind in separate rooms, the message of this study is clear: the most revealing test of fall risk may be the one that forces both to work together, unpredictably, and at speed.

Subject of Research: Reactive agility and motor-cognitive assessment for distinguishing older adults with and without a history of falls

Article Title: Reactive agility and motor-cognitive assessments for distinguishing community-dwelling older adults with and without a history of falls: a dual-centre cross-sectional study

Article References: Giesche, F., Abobakr, A. H., Banzer, W., Groneberg, D. A., Vogt, L., Hoffmann, M., Albert, I., & Hülsdünker, T. (2026). Reactive agility and motor-cognitive assessments for distinguishing community-dwelling older adults with and without a history of falls: a dual-centre cross-sectional study. GeroScience. https://doi.org/10.1007/s11357-026-02477-4

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02477-4

Keywords: reactive agility, fall prevention, older adults, motor-cognitive assessment, Timed-Up-and-Go, SKILLCOURT, executive function, dual-task, GeroScience, healthy aging, reaction time, functional mobility

Episodic foresight is often described as mental time travel. It allows a person to simulate a possible future event, recognize what it will demand, and organize current behavior accordingly: packing a raincoat because a storm is forecast, or grabbing a permission slip because it must be handed in tomorrow. Psychologists regard it as the most flexible and functionally powerful form of prospective cognition, and a critical prerequisite for independent living. Yet almost everything previously known about episodic foresight in autism came from studies asking participants to describe imagined future events, rating how vivid, detailed or specific those descriptions were. A 2023 meta-analysis of 31 such studies found moderate-to-large autism-related differences across measures of detail, experiential quality and specificity, and a recent longitudinal study showed that while future-scenario imagination developed over three years in non-autistic adolescents, it remained stable in autistic adolescents.

What those paradigms could not answer was whether the reported differences in imaginative richness translate into practical consequences — whether autistic children actually act differently when the future matters. To find out, the team turned to the child version of Virtual Week-Foresight, a computerized board game that is currently the only standardized behavioral measure meeting the stringent experimental criteria laid out by Suddendorf and Corballis for demonstrating genuine episodic foresight. Those criteria are designed to rule out four alternative explanations for actions that happen to benefit the future, ensuring that success truly reflects acting with the future in mind rather than habit, chance or instruction-following.

In the game, children roll a die and move a token around a board representing a virtual day. As they pass green ‘S’ squares, they draw Situation Cards presenting everyday scenarios, such as having breakfast, and choose among options. Embedded among twenty such cards — many of them deliberate distracters — are seven episodic foresight problems and seven contexts in which those problems can be resolved. Along the way, Daily Activity Cards offer chances to pick up one item from a list of five, four of which are useless distracters. The design simulates how foresight problems arise in real life: buried in the flow of ongoing activity, requiring the child to spot a future need, self-generate a solution, acquire the right item, and then remember and choose to deploy it at the right future moment. The task yields two scores: item acquisition, reflecting the ability to take appropriate preparatory steps, and item use, reflecting the ability to follow through when the problem reappears.

Forty autistic children aged 8 to 12, all with IQ scores above 80 and no intellectual impairment, completed the task alongside 55 age- and IQ-matched non-autistic controls. All children were tested at home in distraction-free rooms, and the autistic children’s diagnoses were verified through clinician reports and the Social Communication Questionnaire, on which the autistic group scored substantially higher, confirming poorer social and communication skills. Crucially, the two groups were statistically indistinguishable in age, overall cognitive ability and retrospective memory, although the autistic group did perform more poorly on standardized tests of executive function and were rated by parents as showing lower functional self-direction in daily life.

The results were unexpected and, in their way, more revealing than a simple deficit story. On a mixed-model analysis of variance, the groups diverged not in what they acquired but in what they did with it. Autistic children were just as successful as their non-autistic peers at identifying future problems and acquiring the items needed to resolve them — the preparatory, construction phase of foresight appeared fully intact. But when the moment of resolution arrived, they were significantly less likely to actually use those items, a difference that was statistically reliable and could not be explained by disengagement or misunderstanding of the task, since the same children had demonstrably grasped the problems and their solutions moments earlier. Both groups found item use harder than item acquisition, but the gap was wider for the autistic children.

Could broader cognitive weaknesses explain the follow-through failure? The researchers tested this directly. Children completed the Trail Making Test and the Color-Word Interference Test from the Delis-Kaplan Executive Functioning System, indexing cognitive flexibility and inhibition respectively, along with the List Memory Delayed subscale of the NEPSY-II to measure retrospective memory. Hierarchical regression analyses controlling for age and intelligence found that neither executive function nor retrospective memory uniquely predicted item acquisition or item use in either group. The autism-related difference in applying foresight therefore appears to be a specific challenge in its own right, not a downstream consequence of memory or executive difficulties — a finding the authors describe as raising new questions about what cognitive processes actually underpin the effect.

The study also hints at why previous research painted such a uniformly negative picture. Under the constructive episodic simulation hypothesis, imagining the future involves two phases: a basic construction phase that assembles a hypothetical event from retrieved information, and an elaboration phase that fleshes it out with rich episodic and contextual detail. Prior studies, which asked children to narrate detailed future scenarios, taxed both phases. Virtual Week-Foresight, by contrast, presents relatively obvious problems and solutions, so acquiring an item may demand only enough simulation to recognize a future need. On this reading, autistic children can construct adequate future representations to guide preparation, but falter at the point of elaboration — or at the distinct, self-initiated act of implementing an intention at the right time. Notably, earlier work on prospective memory in autism involved intentions set by experimenters; this study shows the difficulty persists even when the intention is self-generated.

The functional stakes are considerable. In the non-autistic control group, both item acquisition and item use correlated strongly with parents’ ratings of self-direction on the Adaptive Behavior Assessment System, the scale that asks how often a child routinely arrives on time or manages independent routines. Longitudinal research has documented a developmental gap of more than eight years in daily living skills between autistic and non-autistic adolescents by the end of adolescence, persisting into young adulthood, and this gap exists even among those with preserved intelligence. The new findings suggest a possible mechanism: knowing what the future needs is not the bottleneck; converting that knowledge into timely action is. Intriguingly, a parallel study in multiple sclerosis found the same acquisition-intact, use-impaired pattern in depressed patients who also showed poorer daily living skills, hinting that the implementation stage of foresight may be a vulnerable link across conditions.

The authors caution that only one domain of functional capacity was assessed, that parent report alone may miss school-based demands, and that the study was underpowered to test sex differences, though exploratory analyses found no male-female differences in foresight performance among autistic children. Still, the implications are concrete. Support strategies for autistic children, the researchers suggest, should focus not on teaching children to anticipate problems — which they demonstrably can do — but on scaffolding the execution of self-generated intentions: prompts, routines and environmental cues that bridge the gap between recognizing a future need and acting on it. In a field long dominated by narratives of imaginative deficit, the study reframes the story in a more precise and arguably more hopeful way: the future is visible to these children; what is needed is help seizing it at the right moment.

Subject of Research: Episodic foresight and future-directed behavior in autistic children without intellectual impairment

Article Title: The Functional Application of Episodic Foresight in Autistic Children

Article References: Chua, S. J. S., Terrett, G., Coundouris, S. P., Rendell, P. G., Suddendorf, T., & Henry, J. D. (2026). The Functional Application of Episodic Foresight in Autistic Children. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-026-07561-4

Image Credits: AI Generated

DOI: 10.1007/s10803-026-07561-4

Keywords: autism spectrum disorder, episodic foresight, prospective cognition, future thinking, daily living skills, executive function, retrospective memory, Virtual Week-Foresight, developmental psychology, prospective memory, children, functional independence

The study drew on 461 participants from the Wisconsin Sleep Cohort, a community-based longitudinal project that has tracked sleep health in Wisconsin adults since the late 1980s. For this analysis, the researchers used data collected between 2019 and 2023, when participants first completed the Sleep Inertia Questionnaire, a validated 23-item instrument that probes the cognitive, behavioral, emotional, and physiological dimensions of the sleep-to-wake transition. Participants averaged 73.8 years of age, were slightly more likely to be male, and were predominantly non-Hispanic white with at least some college education. Crucially, the cohort’s decades of infrastructure meant the researchers could adjust their statistical models for an unusually rich set of confounders, including body mass index, depression scores, alcohol and caffeine use, smoking history, chronotype, habitual sleep duration, and objectively measured sleep apnea severity from overnight polysomnography.

Cognitive performance was assessed with a standardized battery of six widely used neuropsychological tests, yielding seven outcome measures. The battery included the Rey Auditory Verbal Learning Test for memory, the Grooved Pegboard for fine motor speed and dexterity, Trail Making Test Part B for executive function and cognitive flexibility, the Symbol Digit Modalities Test for processing speed, the Controlled Oral Word Association Test for verbal fluency, and the Digit Cancellation Test for attention. Testing sessions lasted roughly 35 minutes and were deliberately scheduled during typical waking hours—usually late morning, mid-afternoon, or evening—rather than immediately upon awakening, with testing beginning an average of nearly 11 hours after participants reported waking on the day of assessment.

The results were striking in their selectivity. Higher scores on the Sleep Inertia Questionnaire were significantly associated with slower completion times on the Grooved Pegboard and Trail Making Test Part B, and with fewer correct responses on the Symbol Digit Modalities Test in unadjusted analyses. After the researchers statistically controlled for the full panel of demographic, psychosocial, and sleep-related covariates, the associations with the Grooved Pegboard and Trail Making Test Part B remained robust. These two tasks share a common thread: both depend heavily on psychomotor speed and executive functioning, suggesting that sleep inertia may be most closely tied to the brain’s capacity for rapid, coordinated, goal-directed processing rather than to memory or language abilities.

Perhaps the most intriguing finding was what did not predict cognition. The Epworth Sleepiness Scale, the gold-standard self-report measure of daytime sleepiness, showed no significant associations with any cognitive outcome. The Hypersomnia Severity Index, a broader measure of hypersomnia symptoms and impairment, produced only a single weak association—and it ran in a seemingly paradoxical direction, with greater hypersomnia severity linked to slightly better immediate verbal recall. The contrast implies that sleep inertia is not simply another face of general sleepiness. Instead, it appears to be a specific symptom with its own relationship to brain function, one that broader somnolence questionnaires may dilute or miss entirely.

Secondary analyses of the questionnaire’s four subscales sharpened the picture. The physiological, cognitive, and emotional subscales all showed significant associations with cognitive performance, particularly on the Grooved Pegboard and Trail Making Test Part B, with the physiological and cognitive subscales also linked to slower performance on the Digit Cancellation Test and reduced output on the Symbol Digit Modalities Test. The responses subscale, which captures behaviors such as waking to an alarm, showed no significant relationships—a pattern the authors attribute in part to the fact that most participants were retired and less likely to encounter the situations those items describe. Sensitivity analyses confirmed that the findings held when accounting for APOE4 carrier status, a major genetic risk factor for Alzheimer’s disease, and when excluding the small number of night-shift workers in the sample.

Why would the severity of morning grogginess relate to cognitive performance hours later? The biology of sleep inertia remains incompletely understood, but leading hypotheses center on the neural machinery of the sleep-to-wake transition itself. Awakening is not a single switch but a staggered cascade: brainstem arousal systems activate rapidly, while cortical regions—especially prefrontal areas supporting executive function—come online more slowly, and the functional segregation between task-positive and task-negative brain networks is temporarily lost. Proposed contributors include the slow clearance of sleep-promoting substances such as adenosine after waking, disrupted cortical arousal, circadian misalignment, and awakening from slow-wave sleep or during the biological night, all of which can intensify inertia. Pathologically severe sleep inertia, as seen in idiopathic hypersomnia, may reflect dysregulation of these transition mechanisms.

The connection to neurodegeneration is speculative but biologically plausible, and it is where the study becomes genuinely provocative. The locus coeruleus, the brain’s primary noradrenergic nucleus, is a key regulator of transitions between sleep and wakefulness and between REM and non-REM sleep—and it is among the earliest sites of pathological tau accumulation in Alzheimer’s disease, becoming hypoactive during preclinical stages. Impaired monoaminergic neurotransmission upon awakening could, in theory, mediate prolonged episodes of sleep inertia. Adding an intriguing circumstantial link, caffeine—an adenosine receptor antagonist—reduces sleep inertia in sleep-deprivation paradigms, increases locus coeruleus activity, and has been associated in epidemiological studies with diminished Alzheimer’s risk and better cognitive function. None of this establishes causation, but it sketches a coherent framework in which chronic, severe sleep inertia could serve as an early behavioral fingerprint of arousal-circuit vulnerability.

The authors are careful to frame the work as exploratory and preliminary. The cross-sectional design cannot determine whether sleep inertia contributes to cognitive decline, shares an underlying cause with it, or both. The sample, while large and well characterized, was predominantly white and non-Hispanic, limiting generalizability, and most cognitive testing occurred in the evening hours—though the models adjusted for hours since awakening, which the researchers argue makes it unlikely that the measured performance simply reflected acute grogginess during testing. The Sleep Inertia Questionnaire also relies on retrospective self-report, and because the analyses were exploratory, no correction for multiple comparisons was applied, leaving open the possibility that some associations reflect chance. Objective, ambulatory measures of sleep inertia and longitudinal designs in more diverse populations are the clear next steps.

Even with those caveats, the implications are tangible. Sleep inertia is a near-universal experience that typically dissipates within 30 minutes of waking, but its duration and intensity vary dramatically between individuals, and prolonged, debilitating inertia is a hallmark of idiopathic hypersomnia and a frequent complaint in obstructive sleep apnea, insufficient sleep syndrome, circadian rhythm disorders, and mood disorders. If replicated, the current findings suggest that a simple question about morning grogginess—how long it lasts, how severe it feels, how often it occurs—could be folded into clinical assessments of older adults as an inexpensive, readily obtainable signal of cognitive risk. They also raise the possibility that interventions designed to reduce sleep inertia, from optimized wake timing to pharmacological countermeasures, might one day help modify cognitive trajectories in aging. For now, the message is that the fog of waking up may be more than a nuisance: it could be a window into how the aging brain transitions between states of consciousness, and how well it will continue to perform.

Subject of Research: The association between sleep inertia severity and cognitive performance in older adults

Article Title: Association between sleep inertia and cognitive performance among older adults in the Wisconsin Sleep Cohort study

Article References: Love, J. J., Cook, J. D., Hagen, E. W., Rasmunon, A., Ravelo, L. A., Palta, M., Peppard, P. E., & Plante, D. T. (2026). Association between sleep inertia and cognitive performance among older adults in the Wisconsin Sleep Cohort study. Journal of Clinical Sleep Medicine, 22(1), Article 105. https://doi.org/10.1007/s44470-026-00133-4

Image Credits: AI Generated

DOI: 10.1007/s44470-026-00133-4

Keywords: sleep inertia, cognitive performance, older adults, Wisconsin Sleep Cohort, hypersomnolence, executive function, psychomotor speed, neurocognitive testing, locus coeruleus, Alzheimer's disease, daytime sleepiness, sleep health

The stakes are enormous. Dementia currently affects roughly 50 million people worldwide, and projections suggest that number could nearly triple by 2050, reaching around 150 million. Alzheimer’s disease and vascular dementia account for the vast majority of cases, and in 30 to 40 percent of patients the two pathologies coexist. While new pharmacological therapies for early Alzheimer’s disease have generated excitement, they remain insufficient to slow the epidemic at a population level. The 2024 Lancet Commission on dementia prevention estimated that up to 45 percent of cases could be prevented by addressing 14 modifiable risk factors spanning early life, midlife and late life—among them hypertension, obesity, physical inactivity and social isolation.

Executive function is among the cognitive domains most sensitive to aging. Early losses in planning, inhibition and cognitive flexibility interfere with instrumental activities of daily living, reduce social participation and elevate dementia risk. Physical inactivity feeds this process through obesity, diabetes, dyslipidaemia and hypertension, but it also appears to act more directly: activity levels are consistently linked to executive performance over time. Objective markers such as gait speed, handgrip strength and leg muscle power integrate neuromuscular capacity, cardiometabolic health and central motor control into a single measurable phenotype, and previous research has tied them to independence, disability and dementia onset years before diagnosis.

What no study had done before, however, was test whether physical performance moderates—that is, changes the strength of—the relationship between age and executive function, rather than merely correlating with it. The research team, working with community-dwelling adults aged 60 to 85 recruited through town halls, mailings and media campaigns in Donostia/San Sebastián, enrolled participants with a high cardiovascular dementia risk score and subtle cognitive weaknesses, but without dementia or loss of functional independence. Only 6.8 percent of the sample met criteria for mild cognitive impairment, making the cohort a window into the preclinical stage of decline.

Executive function was measured with two well-validated instruments. The Trail Making Test asks participants to connect numbered circles and then to alternate between numbers and letters; subtracting the time for the simpler part from the harder part isolates the executive demands of task-switching. The Stroop test, in which people must name the ink colour of mismatched colour words, probes inhibitory control independently of literacy. Physical performance was assessed with a six-metre walk at usual pace, a handheld dynamometer measuring maximal grip force, and a leg press fitted with a linear encoder that captured peak lower-limb power during explosive contractions. Brain scans graded white matter hyperintensities—the hallmark of cerebral small vessel disease—on the Fazekas scale.

The results were striking in their consistency. After adjusting for sex, socioeconomic status, depressive symptoms, body mass index, anxiety, diabetes, hypertension, dyslipidaemia and smoking, all six age-by-performance interactions remained significant after false-discovery-rate correction. Each year of age added roughly 2.6 to 3.0 seconds to the Trail Making difference score and about half a second to Stroop interference, but higher physical performance blunted these penalties. Standardized interaction coefficients were modest yet reliable across both cognitive tests, and the pattern held whether the outcome was task-switching speed or inhibitory control.

The most novel contribution came from Johnson–Neyman analyses, which pinpoint the exact values of a moderator at which an effect disappears. For the whole sample, the damaging association between age and executive function became statistically undetectable above a gait speed of 1.83 metres per second, a handgrip strength of 48.8 kilograms, or a peak leg power of 438 watts on the Trail Making measure—and above 1.68 metres per second, 43.9 kilograms and 366 watts on the Stroop measure. These thresholds sit at the extreme upper end of the performance distribution: only about 1 to 16 percent of participants reached them, depending on the marker and test. The authors caution that these are sample-specific regions of significance, not clinical cutoffs ready for the doctor’s office.

Sex shaped the picture in intriguing ways. For the Trail Making test, the moderating effect of gait speed was confined to men, and a formal three-way interaction confirmed the specificity: among males, the age-related penalty on executive performance vanished above a gait speed of 1.59 metres per second, a level achieved by 14 percent of the men. For the Stroop task, gait speed moderated the association within the female stratum, but the corresponding three-way interaction was not significant, so the researchers urge against over-reading this as a true sex difference. Notably, when grip strength and leg power were re-expressed relative to body size, the moderation effects attenuated to nonsignificance, suggesting that absolute neuromuscular capacity—not strength adjusted for mass—drove the buffering effect.

Several sensitivity checks strengthened the findings. Excluding participants with mild cognitive impairment left most associations intact, indicating the pattern was not simply an artifact of incipient decline, though the gait effect on task-switching did weaken—hinting that gait speed may be an especially sensitive marker of executive problems in people already showing cognitive impairment. Adjusting for sleep quality or standing height changed little. White matter hyperintensity burden, present at moderate-to-high levels in nearly 29 percent of participants, did not independently predict executive function after full covariate adjustment, though the authors attribute this partly to the coarse visual Fazekas rating and the reduced MRI subsample of 774 people, and they stress this does not overturn the well-established links between small vessel disease and cognition.

Beyond the physical findings, socioeconomic status emerged as the second most powerful independent predictor of executive performance after age itself, with each step up the index corresponding to roughly 11 to 15 seconds of faster task-switching across both sexes—a vivid demonstration of cognitive reserve built through education, occupational complexity and access to healthier lifestyles. The study’s cross-sectional design cannot establish causality, and unmeasured factors such as medication use, habitual activity and APOE genotype may still confound the results. Yet the work offers a reproducible analytic template for the World Wide FINGERS network of multidomain prevention trials and raises a provocative possibility: that exceptionally high neuromuscular performance represents a functional reserve capable of buffering brain aging, complementing vascular-risk control. Whether those extreme performance levels can be reached through exercise interventions begun later in life—and whether doing so translates into real cognitive protection—remains the critical question for longitudinal research now underway.

Subject of Research: The moderating role of physical performance in the relationship between age and executive function in older adults at elevated risk of dementia

Article Title: Physical Performance Moderates the Association Between Age and Executive Function in Older Adults at Elevated Risk of Dementia: Cross‐Sectional Analysis From the CITA GO‐ON Trial

Article References: Reparaz‐Escudero, I., Izquierdo, M., Ecay‐Torres, M., Altuna, M., López, C., Estanga, A., García‐Sebastián, M., de Arriba, M., Ros, N., Saldias, J., Limousin, M., Martínez‐Lage, P., & Sáez de Asteasu, M. L. (2026). Physical Performance Moderates the Association Between Age and Executive Function in Older Adults at Elevated Risk of Dementia: Cross‐Sectional Analysis From the CITA GO‐ON Trial. Journal of Cachexia, Sarcopenia and Muscle, 17(5), Article e70396. https://doi.org/10.1002/jcsm.70396

Image Credits: AI Generated

DOI: 10.1002/jcsm.70396

Keywords: dementia, executive function, gait speed, handgrip strength, muscle power, aging, cognitive reserve, white matter hyperintensities, CITA GO-ON trial, sarcopenia, physical performance, dementia prevention

The team, led by Gregory M. Shaw and Brian C. Clark of the Ohio Musculoskeletal and Neurological Institute, focused on a phenomenon called the multi-finger force deficit, or MFFD. When people grip with all four fingers at once, each finger produces less force than it does when pulling alone, so the combined four-finger force falls short of the sum of the individual finger forces. Because the intrinsic force-generating capacity of a finger’s muscles should not depend on whether its neighbors are working at the same time, this shortfall is interpreted as a signature of central nervous system constraints, a limit on how well the brain can assemble and scale simultaneous commands to multiple digits. In younger adults the deficit is modest; in older adults, the new data show, it grows markedly.

What makes the study methodologically important is the device the researchers built to measure it. Previous investigations of the multi-finger force deficit typically used awkward postures, such as a pronated forearm pressed flat against a table, that bear little resemblance to the way grip strength is actually tested in a clinic. The Ohio group instead constructed a custom multi-sensor dynamometer that mimics the biomechanics of the standard Jamar hydraulic dynamometer, the instrument used in virtually all clinical grip assessments. Four S-type load cells, each coupled to a finger loop, recorded force from the index, middle, ring, and little fingers independently, while a 3D-printed wrist holder and Velcro-secured forearm support kept limb positioning identical to clinical guidelines: elbow flexed at 90 degrees, shoulder slightly abducted, wrist neutral. Grip values from the custom device correlated strongly with the Jamar standard, with Pearson’s r of 0.87 for the dominant hand and 0.90 for the non-dominant hand, confirming that the biomimetic setup was measuring the same thing clinicians measure.

Participants included 12 younger adults with an average age of about 24 years and 10 markedly older adults averaging just over 80 years, a group deliberately enriched with people in the middle-old and oldest-old ranges, many of whom showed clinically meaningful weakness. The results were striking. Older adults exhibited handgrip strength 48.8 percent lower than the young participants. Their multi-finger force deficit averaged 25.8 percent, compared with 16.5 percent in the young group, a statistically significant difference. In other words, when older adults tried to use their whole hand, they lost a substantially larger fraction of their available force to the problem of coordination alone.

The most provocative result emerged when the researchers stratified the older participants by the grip strength thresholds used by the European Working Group on Sarcopenia in Older People, the criteria that define probable sarcopenia in clinical practice. Older adults whose grip fell below the threshold, less than 27 kilograms for men and 16 kilograms for women, showed an average multi-finger force deficit of 33.4 percent, nearly double the 18.0 percent seen in their stronger peers. Because the deficit is a ratio, comparing simultaneous four-finger force to the sum of individually generated finger forces, this gap cannot be explained simply by having weaker muscles. The clinically vulnerable older adults could still generate considerable force with each finger in isolation; what they lost disproportionately was the ability to express those forces together.

The authors propose a compelling mechanistic account for this pattern, which they frame as a parallelization penalty. When the four fingers pull one at a time, the nervous system can concentrate its resources on a single digit command. When all four must pull simultaneously, the corresponding motor commands must be generated in parallel while preserving their relative weighting and stabilizing the wrist and hand. Individual fingers are not controlled through anatomically isolated cortical channels; their representations overlap extensively in the motor cortex, and selective finger output emerges from the pattern of activity across distributed cortical and corticospinal populations. Age-related cortical dedifferentiation, reduced segregation of sensorimotor networks, and altered inhibitory neurochemistry could blur this control system, allowing adequate performance when fingers act alone but creating competition when commands must be expressed concurrently. Reduced motoneuron excitability and diminished persistent inward currents, which amplify descending drive in spinal motor neurons, may further shrink the reserve available for high-force output in weak older adults.

The study added a second, independent line of evidence using cognitive dual-task testing, grounded in the idea that motor and cognitive operations draw on overlapping neural resources. Participants performed maximal grips while simultaneously reading aloud a passage from War and Peace, or while completing a visuospatial go/no-go task that required rapid decisions about numbers flashing on a screen. The effects were task- and age-dependent in an intriguing way. Older adults lost roughly 10 percent of their composite grip strength during the reading task compared with gripping alone, a significant decline, while younger adults lost only about 6 percent and showed no statistically significant drop in that condition. Younger adults, by contrast, were significantly affected by the go/no-go task, whereas the older adults maintained their grip force during it. The finding demonstrates that maximal grip strength is not a fixed property of muscle but a context-sensitive performance that fluctuates with cognitive load, and that the direction of interference depends on the nature of the secondary task.

Correlational analyses reinforced the link between the multi-finger force deficit and broader aging outcomes. Across all participants, a larger deficit in the non-dominant hand was associated with slower performance on the Four-Square Step Test of dynamic balance, slower fast-paced gait speed, poorer Purdue Pegboard dexterity, and longer completion times on the Trail Making Test difference score, a measure of executive function and cognitive flexibility. Notably, the deficit in the dominant hand showed no such associations. The authors interpret these patterns through the common-cause hypothesis, which holds that age-related declines in motor and cognitive function arise from shared neurobiological substrates, including white matter deterioration and changes in dopaminergic signaling, while cautioning that the cross-sectional, unadjusted correlations are hypothesis-generating rather than proof of mechanism.

The researchers are careful about what the multi-finger force deficit is and is not. It is not a direct measure of cortical connectivity, corticospinal excitability, or motoneuron gain, and the study’s small sample, fixed testing order, and exploratory subgroup analysis leave the sarcopenia comparison in need of replication. Grip strength below the EWGSOP2 threshold indicates probable rather than confirmed sarcopenia, and the behavioral design cannot isolate the precise neural cause. Still, the authors argue that the measure could eventually serve as a behavioral stress test of multi-digit neuromotor reserve, one that could be incorporated into routine grip assessment simply by adding single-digit trials to a conventional dynamometer test. Before that can happen, future work must establish test-retest reliability, normative values, diagnostic thresholds, and whether the deficit adds predictive value beyond grip strength alone, ideally in longitudinal cohorts that combine it with direct neurophysiological measurement.

If validated, the implications extend well beyond the laboratory. Handgrip strength predicts falls, hospitalization, and all-cause mortality, and in some cohorts it outperforms systolic blood pressure as a mortality predictor, yet it has long been treated as a gross index of muscle force. This study suggests that part of what the dynamometer is really probing is brain health, the capacity of an aging nervous system to orchestrate dozens of muscles, as many as 39 spanning the forearm and hand, into a coordinated burst of power. Weakness in an 80-year-old hand, the work implies, is partly a story about cortical organization, neural reserve, and the mounting cost of doing several things at once. For a rapidly aging population, that reframing could open entirely new avenues for detecting vulnerability early, and perhaps for training the nervous system, not just the muscle, to hold on to its strength.

Subject of Research: Neural mechanisms underlying age-related handgrip weakness assessed by multi-finger force deficit and cognitive dual-task testing

Article Title: Neural contributions to age-related handgrip weakness revealed by multi-finger force deficit and dual-task testing

Article References: Shaw, G. M., Clark, L. A., Grooms, D. R., Carson, R. G., & Clark, B. C. (2026). Neural contributions to age-related handgrip weakness revealed by multi-finger force deficit and dual-task testing. GeroScience. https://doi.org/10.1007/s11357-026-02517-z

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02517-z

Keywords: sarcopenia, handgrip strength, multi-finger force deficit, aging, motor cortex, neural coordination, dual-task, GeroScience, dynamometry, corticospinal, executive function, older adults

The research team recruited participants within three months of a first clinical demyelinating event, often called a clinically isolated syndrome, at The National Hospital for Neurology and Neurosurgery and Moorfields Eye Hospital in London. Forty participants, averaging about 32 years of age at baseline and roughly two-thirds female, completed detailed clinical, cognitive, and magnetic resonance imaging assessments. Five years later, the same individuals returned for a comprehensive re-evaluation, giving researchers an unusually clean window into the earliest stages of the disease, before years of cumulative damage could muddy the picture.

Cognitive function was assessed with the Minimal Assessment of Cognitive Function in MS, a battery widely regarded as the gold standard for this population because it covers the domains most commonly affected, including processing speed, learning, memory, and executive function. Each participant’s raw scores were converted into standardized Z-scores using regression-based normative data, which already account for age, sex, and education. Using the conventional impairment threshold of Z below minus 1.50 on at least one subtest, 43 percent of the cohort met criteria for cognitive impairment at five years. When the researchers applied a more lenient cutoff of Z below minus 1.00, designed to capture subtler dysfunction on the continuum between normal cognition and frank impairment, the figure rose to 53 percent.

The pattern of deficits was revealing. The most frequently failed tests were the Controlled Oral Word Association Test, a measure of phonemic verbal fluency, and the Delis-Kaplan Executive Function System Sorting Test, specifically its Description Score, which demands abstraction, concept formation, and cognitive flexibility. By contrast, failures on the Symbol Digit Modalities Test, the workhorse screening measure of processing speed, were almost absent. The authors argue that larger MS cohorts may systematically underestimate executive dysfunction because brief screening batteries lack dedicated executive measures, whereas the fuller MACFIMS battery is far more sensitive to these higher-order deficits, which can emerge even when processing speed appears intact.

On the imaging side, the team acquired high-resolution three-dimensional T1-weighted and FLAIR sequences on a 3 Tesla scanner and used an automated lesion segmentation pipeline, manually quality-checked, to quantify lesion counts and lesion volumes. Brain tissue volumes, including gray matter, white matter, and total intracranial volume, were derived after lesion filling, a computational step that prevents lesions from distorting tissue segmentation. All volumetric analyses were adjusted for head size. The central question was simple but important: which MRI measures, five years after the first demyelinating event, predicted who had developed cognitive impairment, and did age modify those relationships?

The answer centered on gray matter. In models adjusted for total intracranial volume alone, each cubic centimeter decrease in gray matter volume was associated with roughly a 6 percent increase in the odds of cognitive impairment, an effect that held at both the conventional and lenient Z-score thresholds. Cortical gray matter showed a similar association, while white matter volume, lesion counts, and lesion volumes did not independently predict cognitive outcomes. Notably, when age and sex were added to the models, the MRI associations lost statistical significance, a hint that age was entangled with the gray matter story in a way that simple additive models could not capture.

That entanglement became explicit when the researchers tested interaction effects. Among participants classified as cognitively impaired at five years, each additional year of age was associated with roughly one cubic centimeter less gray matter, a significant negative slope. Among those who remained cognitively preserved, age had no measurable relationship with gray matter volume at all. In other words, the brains of cognitively impaired participants appeared to be aging faster, or at least shrinking faster, than the brains of their cognitively intact peers, even though the cognitive scores themselves had already been adjusted for age through the normative data.

Marginal contrast analyses sharpened the picture further. Below the age of 39, gray matter volumes did not differ meaningfully between cognitively impaired and cognitively preserved participants. Above that threshold, the groups diverged, with statistically significant differences emerging at age 39 and widening thereafter. Comparing modeled gray matter volume at age 25 versus age 45 within the cognitively impaired group revealed a difference of roughly 16 cubic centimeters, while no comparable age-related difference existed among those who stayed cognitively intact. The authors interpret this as a possible synergistic effect between brain aging and disease-related gray matter loss, speculating that diminished gray matter reserve may modulate MS-related cognitive outcomes as people get older. This resonates with a growing literature on accelerated or premature brain aging in MS, often quantified through brain-age modeling, in which MRI-derived estimates of brain age exceed chronological age in people with the disease.

Importantly, the cognitive outcomes were not explained by the usual clinical suspects. Post hoc analyses found no associations between cognitive impairment and disease-modifying therapy use or type, steroid treatment, relapse frequency, fatigue, or anxiety and depression. This absence of association underscores that the gray matter and age effects were not simply proxies for more relapses or more aggressive inflammatory treatment histories, and it reinforces the idea that neurodegenerative processes, unfolding quietly and independently of visible relapse activity, may drive early cognitive change. That aligns with accumulating evidence that progression independent of relapse activity can begin insidiously, even before the first clinical event.

The study has honest limitations. Forty participants is a small sample, which may explain why only gray matter measures reached significance, and the authors caution that the absence of associations with deep gray matter or lesion measures should not be read as evidence of absence. The cohort was also dominated by optic neuritis as the presenting event, and there was no healthy control group. The magnitude of the gray matter differences, while statistically robust, raises questions about biological meaningfulness that only longer follow-up can answer. Still, the core message stands and carries practical weight: cognitive assessment, particularly of executive function, deserves a place in routine early MS care, and age may be a critical variable in identifying who is most vulnerable. Future longitudinal studies, ideally beginning even before the first demyelinating event in people with radiologically isolated syndromes, could clarify whether protecting gray matter reserve in early adulthood offers a genuine buffer against the cognitive toll of multiple sclerosis decades later.

Subject of Research: Age-related modulation of the association between cognitive impairment and gray matter brain volumes in the five years following a first demyelinating event in multiple sclerosis

Article Title: Advancing Age Modulates Associations Between Cognitive Impairment and Brain Volumes in Early MS

Article References: Ananthavarathan, P., Pitteri, M., Foster, M., Collorone, S., Salama, S., Colato, E., Prados, F., Kanber, B., Yiannakas, M., Gandini Wheeler‐Kingshott, C. A. M., Davagnanam, I., Barkhof, F., Chard, D., Ciccarelli, O., & Toosy, A. (2026). Advancing Age Modulates Associations Between Cognitive Impairment and Brain Volumes in Early MS. Annals of Clinical and Translational Neurology, 13(10), 2124-2133. https://doi.org/10.1002/acn3.70385

Image Credits: AI Generated

DOI: 10.1002/acn3.70385

Keywords: multiple sclerosis, cognitive impairment, gray matter volume, brain aging, MRI, clinically isolated syndrome, executive function, MACFIMS, neurodegeneration, brain-age paradigm, demyelinating event, longitudinal study

The research focused on the left dorsolateral prefrontal cortex, a region tucked into the upper outer portion of the frontal lobes that serves as a hub for cognitive control. This area is heavily involved in planning, working memory, attention regulation, and the flexible redirection of thought when circumstances change. It is also the most common target for repetitive transcranial magnetic stimulation, an approved treatment for treatment-resistant depression. Despite its clinical importance, researchers have lacked objective, neurophysiological markers that connect dysfunction in this region to the specific cognitive impairments seen in patients. The new study set out to fill that gap by asking a deceptively simple question: when you stimulate the left dorsolateral prefrontal cortex and record the brain’s immediate electrical reply, does that reply differ in people with depression, and does it relate to how well they think?

To answer it, the team recruited seventy-nine participants: forty-nine individuals diagnosed with major depressive disorder and thirty healthy controls. Each participant completed a comprehensive clinical assessment along with a battery of executive-function tests spanning multiple cognitive domains, including processing speed, sustained attention, cognitive flexibility, task switching, and the ability to inhibit responses during emotional conflict. The researchers then applied single-pulse transcranial magnetic stimulation over the left dorsolateral prefrontal cortex while simultaneously recording brain activity with electroencephalography, a paired technique known as TMS-EEG. The stimulation was delivered at the F3 scalp position, and the resulting TMS-evoked potentials were measured directly at that electrode, yielding a series of characteristic deflections labeled P30, N45, P60, N100, and P180 according to their polarity and timing in milliseconds after the pulse.

The behavioral results confirmed what prior clinical literature has suggested. Compared with healthy controls, patients with depression performed significantly worse across several domains: they were slower on processing-speed measures, showed reduced sustained attention, struggled more with cognitive flexibility and task switching, and had greater difficulty exerting inhibitory control when faced with emotionally conflicting information. These deficits were not confined to a single test but formed a coherent pattern of multidomain executive dysfunction, reinforcing the view that depression’s cognitive burden is broad rather than narrow, and that it touches the very machinery of self-regulation that people rely on in work, study, and daily decision-making.

The neurophysiological findings were more selective, and that selectivity is what makes them compelling. Of the five TMS-evoked components examined, only one distinguished the groups: the N100, a negative-going deflection appearing roughly one hundred milliseconds after stimulation. Patients with depression showed a significantly larger N100 magnitude at the F3 electrode than healthy controls, while the earlier and later components, P30, N45, P60, and P180, did not differ between the groups. This pattern suggests that the alteration in depression is not a blanket change in cortical responsiveness but a specific modification of the mid-latency inhibitory processing that the N100 is thought to reflect, a component widely associated with GABAergic inhibitory mechanisms in the stimulated cortex.

The most striking result emerged when the researchers looked within the patient group. A more negative N100 amplitude was associated with poorer sustained-attention performance, measured with the Continuous Performance Test, and this correlation survived correction for multiple comparisons using the false discovery rate procedure. In other words, the strength of the brain’s electrical response to direct prefrontal stimulation predicted how well a patient could maintain focus over time. The association was specific enough that the other TEP components did not show the same relationship, hinting that the N100 carries information about a particular cognitive function rather than general brain health.

Skeptics might wonder whether the link simply reflects mood severity, anxiety, or demographic differences rather than a genuine neurocognitive relationship. The research team anticipated this concern and tested it directly with hierarchical multiple linear regression. After statistically adjusting for depressive symptom severity, anxiety symptoms, age, sex, and years of education, the association between N100 amplitude and sustained-attention performance remained significant. This robustness matters because it suggests the N100 is not merely a proxy for how depressed someone feels on the day of testing; it appears to index something about prefrontal cortical function that is tied to attentional capacity in its own right, independent of the clinical and demographic factors that often confound such studies.

The authors also took care to address the possible influence of medication. Supplementary analyses compared patients who were medication-free with those taking antidepressants, and separately with those treated with benzodiazepines, examining both clinical characteristics and TEP amplitudes across these subgroups. Quality metrics for the TMS-EEG preprocessing, resting motor threshold, and stimulation intensity were likewise compared between groups to ensure that the group difference in N100 was not an artifact of differing recording conditions or stimulation parameters. These controls strengthen the case that the observed alteration reflects genuine neurophysiological differences associated with the disorder rather than technical or pharmacological confounds, although the authors are careful to note that the findings require longitudinal validation before the N100 can be considered a clinically useful marker.

Why should this matter beyond the laboratory? First, it provides an objective, physiology-based correlate of cognitive dysfunction in depression, a field that has long relied on subjective reports and behavioral testing alone. A measurable electrical signature recorded in a single session could, with further validation, help clinicians identify which patients carry the greatest cognitive burden, track changes over the course of illness or treatment, and potentially guide neuromodulation therapy by revealing how responsive a patient’s prefrontal cortex is before a treatment course begins. Second, the specificity of the N100-sustained attention link offers a mechanistic clue: it points toward altered inhibitory processing in the left dorsolateral prefrontal cortex as a candidate contributor to the attentional failures that so many patients describe, connecting a cellular-level hypothesis about GABAergic function to a real-world symptom.

The study’s limitations are acknowledged by its authors and deserve emphasis. The design was cross-sectional, meaning it captured a single moment in time and cannot establish whether the altered N100 causes attentional impairment, results from it, or both reflect a shared underlying process. The patient group was heterogeneous in medication status, and while the supplementary analyses addressed this, larger samples will be needed to disentangle drug effects definitively. The authors themselves state that the functional specificity and clinical utility of the F3-recorded N100 require longitudinal validation. Even so, the work represents a meaningful step toward grounding the cognitive symptoms of depression in measurable brain physiology. For the millions of people whose depression is defined as much by foggy thinking and wandering focus as by low mood, the prospect of an objective marker, and eventually a targeted way to monitor and treat cognitive dysfunction, is a development worth watching closely.

Subject of Research: Neurophysiological markers of executive dysfunction in major depressive disorder using TMS-evoked potentials from the left dorsolateral prefrontal cortex

Article Title: Association between executive function and left dorsolateral prefrontal TMS-evoked potentials in major depressive disorder

Article References: Zhang, J.-Q., Zhang, H.-G., An, Q., Cao, M.-N., Deng, W.-Y., Jing, W.-T., Zhao, J.-H., Xue, C., Sun, J.-J., & Deng, W. (2026). Association between executive function and left dorsolateral prefrontal TMS-evoked potentials in major depressive disorder. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08646-1

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08646-1

Keywords: major depressive disorder, executive function, TMS-EEG, TMS-evoked potentials, dorsolateral prefrontal cortex, sustained attention, N100, neurophysiology, cognitive dysfunction, neuromodulation, psychiatry, prefrontal cortex

The core problem with traditional cognitive assessment is statistical. Standard batteries such as the Corsi Block-Tapping task for working memory or the Stroop test for inhibitory control treat each task as an isolated measurement. Researchers typically average across repeated trials — mean response times on the Stroop, for example — and treat trial-to-trial variability as mere noise. Structural equation models can link tasks together, but only under assumptions of linearity that the authors argue oversimplify the true architecture of cognition. The result is a testing paradigm that requires many observations per task before estimates stabilize, and one that fails entirely when a task is skipped or data are missing.

The Washington University team had previously introduced an alternative: the distributional latent variable model, or DLVM. Instead of summarizing performance with a single average, DLVM models the full distribution of an individual’s behavior on each task — capturing both central tendency and variability — and compresses that information into a low-dimensional latent space learned by a neural network. Crucially, every single observation contributes to the estimation of multiple constructs at once, because the model exploits the nonlinear dependencies that link performance across tasks. A person’s position in this learned latent space constitutes their cognitive profile, and the trained model can even generate plausible behavioral data for hypothetical profiles, functioning as a generative oracle for simulation.

Paired with DLVM is a second innovation: distributional active learning, or DALE, a Bayesian algorithm that decides which test item to administer next. DALE frames cognitive testing as sequential Bayesian inference. After each response, it updates a posterior distribution over the individual’s latent position and then selects the next trial by maximizing expected mutual information — in plain terms, it always asks the question that will do the most to shrink its own uncertainty. The algorithm can be primed with a small batch of observations spanning all tasks, in this case two samples per task, before active selection kicks in. This lineage descends from Bayesian active learning methods that have already transformed audiology and vision testing, where adaptive stimulus selection dramatically reduced the trials needed to map perceptual thresholds.

What has been missing until now is ground truth. In the team’s earlier human study, real participants produced real data, but the true underlying cognitive parameters were unknown, making it impossible to measure estimation accuracy precisely. The new paper solves this with an elegant simulation strategy. The researchers trained DLVM models on a retrospective dataset — 88 valid testing sessions in which 18 participants completed up to ten sessions of an eight-task battery over ten days via a mobile app, covering tasks such as the Paced Auditory Serial Addition Test, Countermanding, Running Span, Numerical Stroop, Magnitude Comparison, Corsi Simple and Complex Span, and Cancellation. They then sampled 88 points systematically across the learned latent space and used the model to generate the corresponding ground-truth distributional parameters, from which 240 trial-level observations per task were simulated. Every estimate could now be checked against a known answer.

The first set of analyses pitted DLVM against independent maximum likelihood estimation, or IMLE, the optimal approach if tasks truly were independent. Both models received identical data under equal allocation. The verdict was striking. With only two observations per task, DLVM with two latent dimensions achieved Kullback–Leibler divergence values below 0.200 across all tasks, consistently beating IMLE, with the biggest gains on the sigmoid-shaped span tasks that are notoriously data-hungry. DLVM held its advantage until roughly seven observations per task. More dramatically, in validation analyses DLVM needed only about 20 observations per task — 160 total — to reach near-perfect accuracy in recovering marginal distributions, whereas IMLE required about 100 per task, or 800 total. DLVM could even estimate parameters for tasks that were never administered, something IMLE fundamentally cannot do.

The second stage asked how the way data are collected changes the picture. The team compared six configurations: DLVM or IMLE, each fed by DALE’s adaptive sampling, uniform random sampling, or a traditional fixed test battery delivered in sequential blocks. DALE combined with DLVM was the clear winner in the sparse-data regime, driving KLD below 0.05 by roughly 80 observations. The adaptive algorithm concentrated trials on the complex distributional tasks that carried the most information while allocating fewer trials to simpler accuracy-based tasks, and each simulated session received its own unique, personalized battery. The contrast with conventional practice was stark: at 80 total observations, when a fixed battery had covered only three tasks, DLVM with the battery achieved a KLD of 0.148 while IMLE with the same battery sat at a catastrophic 39.8, simply because it could not infer anything about tasks it had not yet reached.

The authors were careful to probe their own assumptions. Because the simulated data were generated from a DLVM-learned latent space, DLVM might enjoy a structural home-field advantage. So they repeated the analysis using an IMLE-based generative process instead. The qualitative pattern held: IMLE performed best when recovering parameters from data generated under its own specification with abundant data, but DLVM and DALE retained their advantage whenever data were sparse. The team also examined DALE’s trajectories through latent space, finding that the algorithm made large corrections in the first trials, converged to a localized region after about 30 observations, and reliably landed in regions of high probability — even when those regions did not coincide exactly with the true latent position, a consequence of the nonlinear latent space admitting multiple equally plausible solutions. Mean root mean squared error across all 88 sessions was 1.02, and only seven sessions converged to positions with normalized negative log probability above 0.05.

Perhaps the most counterintuitive insight concerns the trade-off between model flexibility and data hunger. In most of machine learning, highly flexible models like deep neural networks need enormous datasets to converge. Here the logic inverts: IMLE, the more flexible estimator, only overtakes DLVM once more than 800 observations are available under these testing conditions, while DLVM’s deliberately constrained low-dimensional embedding extracts meaningful inference from a handful of trials. The authors even observed that random sampling eventually surpassed active learning at very large sample counts, suggesting that switching to random sampling once DALE plateaus — or refining its acquisition function — could reveal additional structure in the data. Both of the study’s pre-registered hypotheses were supported by the results.

The practical implications are considerable. DALE could support adaptive assessment in clinical screening, longitudinal monitoring of cognitive change, and large-scale educational studies where testing time is limited and participant burden matters. One examinee might receive extra Stroop trials while another gets more span or countermanding items, depending on where their performance remains most uncertain, and the output is an uncertainty-aware summary of observable performance rather than a brittle point estimate. The authors argue that future behavioral tasks should be designed to be multidimensional and fully featured so that adaptive algorithms never need to repeat an item, since unsampled regions of a feature space are almost always more informative than sampled ones. For a field long anchored to rigid, hour-long batteries, the message is clear: cognition can be measured faster, smarter, and more personally than ever before.

Subject of Research: Bayesian distributional latent variable modeling and adaptive active learning for efficient assessment of executive functioning

Article Title: Bayesian distributional models of executive functioning

Article References: Bayesian distributional models of executive functioning. (n.d.). https://doi.org/10.3758/s13428-026-03191-x

Image Credits: AI Generated

DOI: 10.3758/s13428-026-03191-x

Keywords: executive function, Bayesian inference, active learning, latent variable models, machine learning, cognitive assessment, working memory, inhibitory control, cognitive flexibility, psychometrics, neural networks, adaptive testing

A new study published in the Journal of Neurology by Electra Chatzidimitriou of the University of California, San Francisco and Aristotle University of Thessaloniki, together with colleagues including Howard Rosen, Maria Luisa Gorno-Tempini, William Seeley, Bruce Miller, and Katherine Rankin, set out to answer that question with unusual rigor. The team had previously developed an integrative model of functional decline in a small Greek cohort of 26 patients with bvFTD, combining cognitive, behavioral, personality, and brain imaging variables. Now they have tested whether that model holds up in a much larger and completely independent sample: 146 individuals with mild-stage bvFTD evaluated at the UCSF Memory and Aging Center. This kind of external validation is a critical but underused step in dementia research, because it determines whether relationships observed in one cohort are robust enough to generalize across clinical settings, demographics, and cultures.

The participants all met the international consortium criteria for at least possible bvFTD, and only patients scoring above 18 on the Mini-Mental State Examination were included, ensuring the analyses focused on early disease rather than advanced neurodegeneration. Each patient underwent structural magnetic resonance imaging on 3 Tesla scanners, a comprehensive neuropsychological battery, and informant-based assessments of behavior, personality, and daily functioning. Because patients with bvFTD frequently lack self-awareness, the researchers relied on knowledgeable informants, such as close relatives, to rate personality with the Big Five Inventory and behavior with the Neuropsychiatric Inventory. The primary outcome was the Functional Activities Questionnaire, an informant report covering ten domains of instrumental daily living, from balancing a checkbook to preparing meals and keeping track of current events.

On the analytical side, the study combined two complementary statistical approaches. First, the team used penalized Least Absolute Shrinkage and Selection Operator regression, a machine learning technique that shrinks the coefficients of uninformative predictors to zero, allowing the strongest correlates of functional status to emerge even when candidate variables are highly correlated. The regularization parameter was tuned through ten-fold cross-validation, and the data were split into training and test sets to evaluate predictive performance. Second, structural equation modeling was used to map the directional pathways among the surviving variables, testing how brain structure, cognition, behavior, and personality interconnect to produce functional disability.

The LASSO analysis delivered a striking result: the two strongest predictors of functional impairment were not cognitive measures at all, but personality and behavior. Reduced conscientiousness, as rated by informants, carried the largest coefficient, followed closely by apathy. Only then came global cognition on the MMSE, executive performance on a modified Trail Making Test, the ability to detect sarcasm on the Awareness of Social Inference Test, and semantic verbal fluency. Together these six variables explained roughly 39 percent of the variance in functional scores. Variables that often preoccupy clinicians, including inhibitory control on the Stroop test, visuoconstructional skills, disinhibition, and even the volume of the right superior frontal gyrus, were shrunk out of the model, indicating that they contributed no independent predictive information once the stronger correlates were accounted for.

The path model then revealed how these predictors relate to one another, and it fit the data exceptionally well, with a nonsignificant chi-square, a Comparative Fit Index of 1.000, a Root Mean Square Error of Approximation of zero, and a standardized residual below 0.05. Three brain-to-cognition pathways emerged from the volume of the right superior frontal gyrus, a prefrontal region involved in executive control, attentional regulation, and the initiation of goal-directed behavior. Smaller gray matter volume in this region was associated with slower executive performance, which in turn predicted worse daily functioning. A parallel pathway linked the same region to global cognition and, through it, to disability. A third, more hierarchical pathway showed that executive dysfunction feeds into broader global cognitive decline, consistent with the characteristic trajectory of bvFTD, in which early frontal network disruption progressively erodes multiple cognitive domains.

Alongside this neural-cognitive cascade, the model identified a partially independent behavioral-personality pathway. Apathy was associated with functional impairment both directly and indirectly, through its relationship with reduced conscientiousness. In other words, the motivational collapse that defines bvFTD appears to undermine daily life in two ways: patients stop initiating activities altogether, and the erosion of goal-directed behavior manifests as disorganization, unreliability, and an inability to plan and complete multistep tasks. Notably, this pathway appeared relatively independent of the neural-cognitive pathways, hinting that motivational and personality changes in bvFTD are mediated by partially distinct neural systems, such as medial frontal and limbic circuits, rather than being mere downstream consequences of executive dysfunction.

One intriguing nuance concerned semantic verbal fluency. In the larger American cohort, fluency was strongly connected to the superior frontal gyrus, executive performance, and global cognition, but it no longer showed a direct path to functional status, unlike in the original smaller Greek study. The authors interpret this through the framework of controlled semantic cognition, which holds that retrieving and deploying conceptual knowledge depends on both a semantic representational system and an executive control system. In a larger sample, the shared variance that fluency once captured across executive and global domains was more precisely disentangled, positioning semantic fluency as an intermediate cognitive marker rather than an independent driver of disability.

The clinical implications are substantial. Because behavioral and personality variables outperformed cognitive tests as predictors of everyday function, the authors argue that assessing bvFTD with cognitive screening alone fundamentally misses the point. Comprehensive evaluation should integrate neurocognitive, behavioral, socioemotional, and personality domains, and care planning should be tailored to each patient’s profile. Patients dominated by apathy and reduced conscientiousness may benefit most from structured environmental support, external cueing, routine-based interventions, and caregiver-mediated behavioral activation, whereas those with prominent executive inefficiency may need targeted training in task sequencing, planning, and goal management. Many patients will require combined approaches.

The study is not without limitations. Variable selection was partly guided by the earlier Greek findings, the neuroimaging analysis focused on a single prefrontal region, and some assessment instruments differed between cohorts, requiring conceptual rather than exact measurement equivalence. The cross-sectional design also precludes causal inference. Still, the convergence of findings across two countries, two healthcare systems, and a fivefold larger sample provides compelling evidence that disability in early bvFTD emerges from the dynamic interplay of two complementary processes: frontal neurodegeneration eroding the cognitive machinery of independence, and motivational and personality changes directly dismantling the will and organization needed to engage with daily life. Recognizing both streams, the authors conclude, is essential for optimizing clinical care and preserving quality of life for patients and families facing this devastating disease.

Subject of Research: Functional impairment and its neural, cognitive, behavioral, and personality determinants in early-stage behavioral variant frontotemporal dementia

Article Title: External validation of an integrative model of functional impairment in early-stage behavioral variant frontotemporal dementia (bvFTD)

Article References: Chatzidimitriou, E., Moraitou, D., Ioannidis, P., Aretouli, E., Chen, Y., Rosen, H. J., Gorno-Tempini, M. L., Seeley, W. W., Miller, B. L., & Rankin, K. P. (2026). External validation of an integrative model of functional impairment in early-stage behavioral variant frontotemporal dementia (bvFTD). Journal of Neurology, 273(10), Article 642. https://doi.org/10.1007/s00415-026-14182-5

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14182-5

Keywords: bvFTD, frontotemporal dementia, functional impairment, apathy, conscientiousness, executive function, superior frontal gyrus, LASSO regression, structural equation modeling, neuropsychology, neuroimaging, dementia

The study stands out in a crowded field for its methodological rigor. Much of the existing evidence linking sleep to cognitive ageing relies on cross-sectional snapshots, single crude measures of sleep, and broad global cognitive scores that lump together very different mental abilities. Marent and colleagues—working with Mitch J. Duncan of the University of Newcastle, Greet Cardon of Ghent University, Genevieve Albouy of the University of Utah, and Jannique van Uffelen of KU Leuven—deliberately broke from that template. They measured sleep both objectively and subjectively, assessed cognition with a validated computerized battery that yields separate domain scores, and used statistical models designed for repeated observations over time, allowing them to ask not only whether sleep and cognition were related but whether those relationships changed as the study unfolded.

The objective side of the sleep measurement came from seven days of wrist-worn accelerometry, processed with the open-source GGIR software pipeline that has become a standard in large-scale sleep research. From the raw acceleration data, the team derived two key parameters: device-measured sleep duration and sleep efficiency, the latter defined as the share of the sleep period spent genuinely asleep rather than awake. Sleep efficiency is a particularly interesting metric because it captures sleep continuity—how fragmented a night of sleep is—rather than simply how long a person remains in bed. Two people can each log eight hours of sleep opportunity, yet one may sleep soundly through the night while the other wakes repeatedly, and accelerometry can distinguish between them in a way that morning questionnaires often cannot.

Subjective sleep quality was captured with the Pittsburgh Sleep Quality Index, one of the most widely used self-report instruments in sleep science. The researchers also constructed a multidimensional sleep health score, an approach that reflects a growing consensus in the field, formalized in a 2025 American Heart Association scientific statement, that sleep should be evaluated as a constellation of dimensions—duration, continuity, timing, quality, and regularity—rather than as a single number. This composite combined device-measured and self-reported sleep dimensions into one index of overall sleep health, testing the idea that the whole sleep profile might matter more for the brain than any individual component.

Cognitive function, meanwhile, was assessed with the Cambridge Neuropsychological Test Automated Battery, or CANTAB, a computerized assessment platform administered at each of the three time points. Rather than reporting a single global cognition score, the researchers computed composite z-scores for four distinct domains: short-term memory, long-term memory, executive function, and processing speed. This domain-specific approach matters because cognitive ageing is not uniform. Different neural systems decline at different rates and respond differently to lifestyle factors, and a global score can mask a real association in one domain by diluting it across others. Executive function, in particular, is known to be among the earliest and most consequential casualties of unhealthy ageing, underpinning the daily planning and self-regulation that keep older adults independent.

The statistical core of the study was a series of linear mixed-effects models, which are well suited to longitudinal data because they can handle repeated measurements within the same individuals while treating sleep characteristics as time-varying predictors. In practical terms, the models asked whether, when a participant’s sleep efficiency was higher than their own typical level, their executive function was also better than their own typical level. The models were adjusted for a careful set of potential confounders: age, sex, educational level, living situation, and total physical activity—the last of which is important because physical activity influences both sleep and cognition and could otherwise masquerade as a sleep effect. The team also tested sleep-by-time interactions to determine whether any association strengthened or weakened across the two-year follow-up.

The results were, in a word, selective. Sleep duration showed no significant association with any of the four cognitive domains. Subjective sleep quality, as measured by the Pittsburgh index, was likewise unrelated to cognition. Even the multidimensional sleep health score—the most sophisticated sleep measure in the analysis—failed to predict performance in any domain. Only sleep efficiency rose to significance, and only for executive function: higher efficiency was associated with better executive performance. The researchers ran sensitivity analyses using wake after sleep onset, essentially the inverse of sleep efficiency, and observed a comparable pattern, lending confidence that the finding was not an artifact of how the variable was defined. Crucially, the sleep-by-time interactions were not significant, meaning the association between sleep efficiency and executive function remained stable across all three measurement waves rather than emerging or fading over time.

Why would sleep continuity matter for executive function when sleep duration and subjective quality do not? The authors’ finding aligns with a broader literature suggesting that fragmented sleep may be more disruptive to prefrontal-dependent processes than shortened sleep. Executive functions are exquisitely sensitive to the quality of prior-night sleep, and sleep fragmentation is known to blunt the slow-wave and spindle activity that supports overnight neural restoration. It is also plausible that the association is bidirectional: emerging evidence, including studies cited in the paper, indicates that preclinical cognitive decline can itself disturb sleep, so reduced sleep efficiency might be an early signal of brain change rather than a cause of it. The null findings for duration and subjective quality echo recent meta-analytic work showing weak and inconsistent links between those measures and cognition in healthy older adults, and they caution against overinterpreting the popular narrative that sleeping longer automatically protects the ageing brain.

The study’s limitations are worth keeping in view. The sample of 233 adults, while well characterized, was relatively healthy and community-dwelling, mean age 68.3 years, and two years of follow-up may be too short to capture the slow accumulation of sleep-related risk for dementia. Accelerometry, for all its objectivity, estimates sleep from movement and cannot stage sleep or detect the micro-architecture that polysomnography reveals. And as the authors note, observational associations—however carefully adjusted—cannot establish causation. Still, the stability of the sleep efficiency–executive function link across repeated observations, in a design that combined objective and subjective measurement with domain-specific cognitive testing, makes this one of the more informative contributions to a debate that matters to millions of ageing adults.

For the public, the takeaway is refreshingly modest and actionable. Rather than fixating on hitting a magic number of hours, the evidence points toward protecting the continuity of sleep—minimizing the night-time awakenings that erode efficiency—as the sleep dimension most plausibly connected to the executive brain in later life. For researchers, the study is a call to move beyond global cognition scores and single sleep metrics toward multidimensional, longitudinal designs that can isolate which features of sleep, if any, genuinely shape how we think as we age. The full dataset’s GGIR processing code has been made publicly available, and the authors report no competing interests, with the work supported in part by the Research Foundation – Flanders and Australian National Health and Medical Research Council funding.

Subject of Research: Longitudinal associations between multidimensional sleep measures and domain-specific cognitive function in older adults

Article Title: Associations between multidimensional sleep measures and domain-specific cognitive function across repeated observations in adults aged 55 years and older

Article References: Marent, P.-J., Duncan, M. J., Cardon, G., Albouy, G., & van Uffelen, J. (2026). Associations between multidimensional sleep measures and domain-specific cognitive function across repeated observations in adults aged 55 years and older. GeroScience. https://doi.org/10.1007/s11357-026-02518-y

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02518-y

Keywords: sleep efficiency, executive function, cognitive ageing, actigraphy, sleep health, older adults, CANTAB, sleep duration, sleep quality, longitudinal study, GeroScience, accelerometry