The Brain Is Never Just in One State

This week

This week’s papers make a persuasive case against treating the brain as an isolated information processor. Its activity is shaped by the room it is in, the body it inhabits, the people around it, and the history it carries. Psychedelics appear to make neural dynamics more responsive to immediate context rather than merely chaotic. Attention can reach down into inflammation. And changes linked to Alzheimer’s may be visible before a defining pathological threshold is crossed. Across these studies, the central question is not simply what the brain does, but when, where, and for whom it does it.

When context enters the brain

Psychedelics are often described as scrambling normal brain networks. The largest neuroimaging study of psilocybin to date suggests something more structured—and more interesting. In 62 participants scanned while resting, meditating, listening to music, or watching films, psilocybin weakened the usual divide between internally oriented and externally responsive networks. But it did not flatten experience into undifferentiated noise. Instead, neural activity traced distinct trajectories that reflected the experimental setting and participants’ reports of self- and boundary-dissolution. The people whose experiences felt positively “embedded” in the surrounding world showed the most coherent context-aligned dynamics—and greater next-day changes in mindset. [1]

That finding is a useful corrective to a simplistic “more connectivity equals better” story. The relevant property may be whether the brain can reorganize around what is actually happening: a piece of music, a meditation instruction, a social environment. It also bears on the increasingly practical question of psychedelic therapy. Set and setting may not merely color an experience after the fact; they may become written into the moment-by-moment organization of brain activity.

A parallel lesson comes from a rather different manipulation: asking people to attend to an acute inflammatory reaction in the skin rather than distract themselves from it. Across three preregistered experiments, internal attention produced inflammatory responses roughly 1.5 times smaller than distraction, with effects appearing in about 90% of participants in two cohorts. The researchers identify both a sensory route—attention scaling the biological impact of identical input—and a top-down route involving parasympathetic vagal activity. [2] Attention, in this account, is not just a spotlight on experience. It is part of the machinery through which the brain regulates the body.

The body, meanwhile, supplies context to the brain in ways we are only beginning to map. In mice, different sensory inputs recruited different layers of cortical arterioles: superficial vessels followed superficial activity, while deeper arterioles integrated more broadly across conditions. [3] This is a reminder that blood-flow signals, including those used in fMRI, are not a generic readout of “how active” a brain area is. Vascular architecture helps decide where energy goes.

The long prehistory of disease

Alzheimer’s research has increasingly organized itself around amyloid: once amyloid burden passes a PET-detectable threshold, the disease process is said to be underway. But longitudinal imaging from three cognitively healthy cohorts suggests that the cortex may change earlier still. People who later became amyloid-positive had relatively thicker cortices and less cortical thinning up to seven years before conversion. Some of these differences remained even after accounting for amyloid quantity. [4]

Thicker is not necessarily healthier. The authors do not establish the cellular cause, and possibilities include inflammation, swelling, altered myelination, or compensatory change. Still, the result expands the window in which an early Alzheimer’s trajectory might be detectable—and complicates the intuition that degeneration always begins with visible shrinkage.

Several studies this week fill in plausible routes by which vulnerability could precede obvious cognitive disease. Among cognitively normal adults in midlife, greater insulin resistance was associated with more frontal white-matter lesions and lower overall cognitive performance. The cognitive association was not explained away by vascular burden, hippocampal volume, or global cerebral blood flow, suggesting that metabolic and vascular injury may operate partly in parallel. [5] It is cross-sectional evidence, not proof that improving insulin sensitivity will protect cognition. But it strengthens the case for metabolism as a dementia-prevention target before neurodegenerative markers become prominent.

Sleep may offer another early clue. In an Alzheimer’s mouse model, amyloid pathology selectively damaged cholinergic neurons in the medial septum, reducing REM sleep alongside cognitive and emotional changes; removing those cells genetically reproduced much of the syndrome. [6] And in a large wearable-data study, stronger, more stable daily activity rhythms tracked with lower odds of accelerated biological aging. [7] Neither result makes a Fitbit a diagnostic device. Together, they underscore that aging brains are embedded in rhythms—sleep, metabolism, movement—that are measurable long before a clinic sees dementia.

Memory needs both flexibility and time

Memory is often framed as storage, but storage is a balancing act. In mice, inhibition in the dentate gyrus determines whether memory prioritizes sensitivity to small differences or stability against interference. Reducing inhibition helped animals distinguish overlapping experiences later, but made their memories more vulnerable to confusion under heavy load. Increasing inhibition stabilized representations, at the cost of making similar events harder to tell apart. [8] The ideal memory system, then, is not maximally detailed or maximally durable. It must adjust its operating point to the problem at hand.

Human recordings from neurosurgical patients offer a complementary view of how goals persist over minutes. When task context was explicitly signaled, the hippocampus encoded it in a fast-changing format, while medial frontal cortex maintained a more stable code that generalized across time. But when participants had to infer the hidden context, the hippocampus shifted into a stable coding mode as well. [9] The division of labor between “memory” and “control” regions is therefore less fixed than textbooks imply; the format of a representation changes with what the task demands.

Even isolated neural tissue seems to retain a remarkable sense of duration. Human brain organoids maintained for five years continued to mature in cell-type-specific ways, with epigenetic age closely tracking time in culture. When progenitors of different ages were mixed, older cells skipped early developmental stages and produced later neuronal fates. [10] These organoids are not miniature minds, nor complete models of a developing child. But they show that human neural progenitors carry an internal developmental history—one that can keep unfolding outside a body for years.

The social and sensory brain is built for action

Social roles, too, are active neural constructions. In a cooperative mouse task, stable leaders and followers emerged spontaneously. The medial prefrontal cortex represented these roles asymmetrically, encoding the partner’s position through role-specific social value maps. Disrupting prefrontal activity, especially in followers, impaired cooperation and changed how animals weighted their own versus their partner’s cues. [11] Leadership here is not simply a trait residing inside one animal; it is a reciprocal computational relationship.

That principle extends to larger groups. Mice formed stable social cliques, and removing oxytocin receptors from olfactory cortex left social motivation intact but disrupted consistent approaches and membership in influential “rich-club” networks. [12] In humans, repeated shared drawing increased brain-to-brain synchrony in same-generation pairs but decreased it in intergenerational pairs, even as particular synchrony measures predicted closeness and loneliness. [13] Synchrony is not a universal meter of connection. Relationships may become coordinated by converging, or by finding complementary roles.

Finally, vision turns out to be less still than it feels. Real-time retinal stabilization experiments in monkeys found that generating microsaccades was itself sufficient to improve peripheral visual sensitivity, even without changes in attentional demands. [14] During natural visual search, foveal attention helped sustain fixation on targets while peripheral attention guided the next eye movement. [15] We do not inspect the world with a stationary camera. Tiny motions of the eye are part of the computation that makes the visual world available.

Looking ahead

The week’s research points toward a brain that is neither sealed off from the body nor governed by immutable wiring. Its dynamics can align with a room, a task, a relationship, a hunger state, or a developmental clock. That flexibility is a source of resilience—but also a route to disease, when metabolism falters, immune cells age, sleep changes, or memory circuits lose their balance. The next challenge is to turn these context-sensitive insights into interventions without losing their nuance: treatments that do not simply alter a molecule or stimulate a region, but restore the conditions under which brains and bodies regulate themselves well.