Attention reaches the immune system — and three tidy claims come apart
The headline result this week is not cognitive at all: across three pre-registered within-subject experiments, where people pointed their attention during an acute skin inflammation changed the size of the inflammatory response, with a vagal route implicated. Alongside it, an active-controlled PET trial gives the first in vivo human evidence that cognitive training shifts cholinergic terminal binding — a Merzenich-lineage claim finally attached to a neurochemical measure, with the usual caveats about vendor authorship and biomarker outcomes. The rest of the week is corrective: uniform 'attentional decline with age' turns out to depend on whether you score accuracy, the reversed gaze-cueing effect may not be social, and a distraction RCT returned a null that is underpowered rather than reassuring.
smaller acute inflammatory response when attention was directed to the body rather than away — consistent in ~90% of participants across two cohorts (Nature Human Behaviour, 2026)
Where the human evidence is
Count by statusEvidence-stage map
Stage × promise (1–10)Attention domain impact
Strength 0–10How attention reaches the immune system
MechanismNature Human Behaviour, 2026. Across three pre-registered within-subject experiments using an identical acute skin-inflammation challenge, where attention was pointed changed how large the inflammatory response became — via a sensory-scaling route and a top-down vagal route.
Where attention actually acts in the brain
Gaze amplifies accumulated-value signals in pre-SMA and dlPFC
Human participants gradually learned the value of two food lotteries while eye position and BOLD were recorded simultaneously, a design that extends the decision over time and dissociates sampled from accumulated evidence. More gaze on an option raised its choice probability, and gaze consistently amplified accumulated-value signals in pre-SMA and partially in dlPFC, above and beyond non-gaze-modulated signals. Correlations in vmPFC, ventral striatum and intraparietal sulcus were inconsistent. Single-lab fMRI with a modest sample — the regional specificity should be treated as provisional.
Read the paperFeature-based attention is not mainly a peripheral phenomenon
Recording simultaneously from foveal and peripheral receptive fields in areas V4 and IT of monkeys during free-gaze search among complex stimuli, the authors found robust feature-based attentional enhancement in foveal units — against the prevailing view that such modulation is predominantly peripheral. Foveal enhancement promoted sustained or repeated fixation on targets, while peripheral attentional signals facilitated detection and guided the next saccade. Non-human primate electrophysiology; the mapping onto human search is an inference, not a demonstration.
Read the paperDorsal-attention network reinstatement predicts what gets recognised
Dynamic large-scale networks were constructed from trial-wise activation maps in the Natural Scenes Dataset, revealing item-specific representations at whole-brain, subsystem and regional scales. Network-level reinstatement — particularly of connections involving higher-order visual and dorsal attention systems — tracked condition-level recognition performance and predicted item-level accuracy, contributing independently of local activation-pattern reinstatement. A re-analysis of an existing open dataset: statistically well powered, but correlational with respect to attention itself.
Read the paperWhat develops, and what actually declines
Vigilance develops in two phases — and lapses change meaning with age
In 809 children, adolescents and young adults, executive vigilance (hits and false alarms modelled with logistic curves) was separated from arousal vigilance (reaction times modelled as ex-Gaussian distributions). Responses became faster, less variable and less prone to extreme reaction times across development, stabilising by late adolescence. The vigilance decrement showed a two-phase pattern: in childhood and early adolescence it appears to reflect limited control over attentional resources, whereas later it increasingly reflects reallocation of resources toward mind-wandering. Cross-sectional, so trajectories are inferred across cohorts rather than followed within individuals.
Read the paper"Attention declines with age" is partly a measurement artefact
Using Combined Attention Systems Test data, the authors show that alerting, orienting and executive control follow dissociable age trajectories rather than declining uniformly, and that spatial orienting (exogenous and endogenous) modulates executive control by reducing flanker conflict. Critically, executive control appears to decline when indexed by reaction time alone but is stable or improved once accuracy is taken into account — consistent with older adults trading speed for accuracy rather than losing control. This is an integrative empirical synthesis, not a new pre-registered experiment, and it argues a measurement point; it does not show that no neurocognitive ageing of attention occurs.
Read the paperClaims under strain
Intermittent digital distraction did not degrade task accuracy in a small RCT
Eighty-seven adults were randomised to standardised audiovisual digital distraction or control while completing dual n-back, Stroop and d2-R tasks with continuous 7-channel EEG and NASA-TLX workload ratings. There was no statistically significant main effect of distraction on error rates in dual n-back or Stroop; subjective workload correlated with errors only during working-memory tasks; and the EEG analysis was compromised by near-collinear frontal-midline theta and alpha bands (ρ > 0.98), limiting interpretation. Read this as an underpowered short-interval null in a single lab, not as evidence that distraction is harmless — brief lab interruptions are not equivalent to sustained notification load.
Read the paperThe reversed gaze-cueing effect may not be "social" at all
In spatial Stroop tasks, gaze stimuli produce a reversed congruency effect while arrows produce a standard one — routinely attributed to a dedicated social-attention mechanism. Two pre-registered experiments manipulating stimulus-onset asynchrony found both effects present at 0 ms SOA and both diminishing as SOA increased, consistent with a non-social account based on transient response activation and inhibition rather than social specialisation. This weakens but does not eliminate the social-attention interpretation: it is one lab's account within a narrow paradigm.
Read the paperOne night without sleep hits motor execution, not the goal representation
Forty-eight healthy adults underwent one night of total sleep deprivation or normal sleep and were tested the next morning on an explicit motor sequence task that dissociates goal-based accuracy from movement-based execution speed, alongside psychomotor vigilance, subjective sleepiness, cortisol, BDNF and corticospinal excitability. Sleep deprivation produced the expected psychophysiological stress signature — raised cortisol, reduced vigilance, increased sleepiness — and selectively reduced movement-time gains during acquisition while sparing spatial accuracy; delayed consolidation at 24 hours and 7 days was preserved. BDNF and corticospinal excitability were unchanged, so the mechanism remains unidentified, and n=48 across two groups is modest for a preserved-consolidation null.
Read the paperWhere a patient directs attention during an acute inflammatory challenge is now a measured physiological variable, not merely an experiential one — but the Nature Human Behaviour model is localised cutaneous inflammation, so do not extrapolate to systemic inflammatory disease.
Cite the INHANCE PET result as evidence of cholinergic plasticity, not as evidence that brain training improves everyday cognition: the endpoint was [18F]FEOBV binding, and vendor-affiliated authors were involved.
When assessing attention in older adults, score accuracy alongside reaction time — the Combined Attention Systems Test analysis shows executive control looks preserved or improved once accuracy is included, and 'declining' when it is not.
After one night of total sleep loss, expect execution speed rather than accuracy or consolidation to suffer: protect speed-dependent skill acquisition, and worry less about precision work or overnight retention.
Do not repurpose the n=87 distraction null as reassurance about notifications — it is a short-interval, single-lab study with a compromised EEG analysis, and absence of an effect there is not evidence of absence.