The Lyceum: Brain & Mind Weekly — Mar 18, 2026

Three completely different anesthesia drugs turn out to break consciousness the same way — a finding that's been hiding in plain sight for decades and could change how surgeons keep you safe on the table. Meanwhile, a 25-year study traced the roots of psychosis back to childhood, when the brain's waste-disposal plumbing quietly fails years before hallucinations ever start. And the largest-ever clinical review of cannabis for mental health landed in The Lancet Psychiatry with a verdict that will make a lot of people uncomfortable: for anxiety, depression, and PTSD, the evidence simply isn't there.

Every time you go under for surgery, your anesthesiologist picks from a pharmacological toolkit where nothing quite matches: propofol boosts GABA receptors (the brain's main "off switch"), ketamine blocks NMDA receptors (suppressing excitatory signaling), and dexmedetomidine cuts off norepinephrine release entirely. Three different molecular targets, three different mechanisms, same goal. But how any of them actually switches off consciousness has been one of medicine's most durable mysteries.

A new MIT study published in Cell Reports found the answer was the same every time. Although each drug acts on neurons differently, all three produce an identical result: a disruption of the brain's balance between stability and excitability. Think of the awake brain as a tuning fork — constantly nudged by sensory input, constantly snapping back to baseline. Under anesthesia, neural networks take progressively longer to recover after processing a stimulus, until the system destabilizes entirely and consciousness is lost. The signature is characteristic: rapid, synchronized bursts followed by prolonged quiet lulls — a flickering pattern that predicts the moment a patient stops responding.

The practical payoff is significant. Clinicians have historically relied on crude proxies like heart rate and blood pressure to gauge anesthetic depth, which can miss overdose or excessive sedation. A universal brain-state metric — measurable with scalp EEG regardless of which drug is being used — could change that. The MIT team is now planning a small clinical trial of an EEG-based monitoring device with surgical patients at Brown University. For vulnerable populations — young children, the elderly, patients with dementia — where excessively deep anesthesia carries serious risk of post-operative cognitive complications, this could reduce the incidence and severity of those complications.

The signature also appears consistent across species, showing up in controlled animal recordings and human EEG alike — which suggests this is a fundamental property of how brains lose consciousness, not a drug-specific quirk. GABA, norepinephrine, NMDA — three completely different systems, one circuit breaker.

Schizophrenia doesn't appear out of nowhere. There are almost always years of silent biological changes before the hallucinations start. For the first time, researchers have traced one of those changes back to childhood — in a system most people have never heard of.

The glymphatic system is your brain's overnight cleaning crew: while you sleep, cerebrospinal fluid flushes through channels around blood vessels, sweeping out metabolic debris, inflammatory proteins, and excess neurotransmitters. A team at the University of Geneva, publishing in Biological Psychiatry: Global Open Science, found that when this drainage system fails to mature during development, it sets the stage for psychosis decades later.

The researchers used a clever natural experiment. They studied individuals with 22q11.2 deletion syndrome — a genetic condition carrying a 30–40% lifetime risk of developing psychotic symptoms, whose deleted genes include some involved in glymphatic integrity. Longitudinal imaging data collected more than 25 years ago were reanalyzed using modern automated techniques. The key finding: in healthy individuals, glymphatic efficiency increases with age. In those who later developed psychosis, this maturation never happened. The downstream chemistry tells the story — drainage failure leads to an imbalance of excitatory glutamate and inhibitory GABA signals in the hippocampus, associated with excessive neuronal excitation and neurotoxicity that precede psychosis.

This isn't an isolated finding. A separate study from the University of Hong Kong, published in Schizophrenia Research in February, found reduced glymphatic function — measured using the same DTI-ALPS imaging index — in people with first-episode schizophrenia and in individuals at clinical high risk. A 2025 Molecular Psychiatry paper reported weaker and more delayed glymphatic clearance in early psychosis patients compared to healthy controls. Two independent teams, two different high-risk populations, converging on the same system.

The clinical implication is tantalizing: if sleep quality modulates glymphatic function, and glymphatic dysfunction precedes psychosis, then sleep intervention might one day be part of preventing a first psychotic episode. The Geneva team's next steps are to analyze links between peripheral inflammation, sleep quality, and psychosis onset.

Millions of people use medical cannabis for anxiety, depression, or PTSD. Most believe it's helping. This week, the most rigorous accounting of the evidence to date arrived in The Lancet Psychiatry — and the math doesn't support the belief.

Researchers led by the University of Sydney synthesized data from roughly 54 randomized controlled trials spanning multiple decades — the largest and most comprehensive systematic review of cannabinoids for mental disorders ever conducted. The result: no benefit for opioid use disorder, tobacco use disorder, cocaine use disorder, bipolar disorder, anxiety, ADHD, psychotic disorders, PTSD, OCD, or anorexia nervosa. Side effects were 1.75 times more common in cannabis groups than placebo, though serious adverse events were not significantly elevated.

There were narrow exceptions. Cannabinoids reduced withdrawal symptoms in people with cannabis use disorder, showed some signal for reducing autistic traits (very low certainty), and offered modest help with insomnia. But the most popular reason people take medical cannabis — anxiety and depression — is precisely the one with the weakest evidence base.

Important caveats: for depression specifically, the negative finding reflects an absence of randomized trial evidence, not a clean disproof. Large reviews of heterogeneous literature can mask subgroup effects — different THC:CBD ratios, delivery routes, and patient populations all get averaged together. Some researchers have pushed back, noting that specific compounds like cannabidiol (CBD) may yet show promise when studied in isolation. The next useful analyses will be the ones that stratify by dose, formulation, and symptom cluster — a null result for "cannabinoids as a broad class" can hide narrow, mechanism-based benefits. According to NPR's reporting, prescription rates in countries like Australia have outpaced clinical evidence, with sales increases that worry public-health researchers.

The experience of ADHD — physically awake but mentally somewhere else — may have a literal neural basis that goes deeper than dopamine.

A study from Monash University and the Paris Brain Institute, published in the Journal of Neuroscience, found that adults with ADHD show more "local sleep" activity during tasks than neurotypical adults. Local sleep is exactly what it sounds like: slow brain waves — normally the signature of deep sleep — briefly appearing in localized cortical regions while the person remains globally awake. In the ADHD group, these waves clustered in the fronto-central area (the brain's executive control hub) and appeared right before a person missed a button press or showed inconsistent reaction times.

The concept of local sleep was previously associated with sleep deprivation in non-ADHD brains. The possibility that ADHD involves chronic microbursts of cortical sleep during waking hours reframes the condition as partly a disorder of arousal regulation, not just attention or impulse control. Methodological advances are helping make the claim testable: teams have developed rolling-window detection algorithms (RoWDI) that scan EEG data for sleep-like signatures and flag brief intrusions correlated with performance drops.

The study was cross-sectional (63 adults, medication-free), so it shows a link but doesn't prove causality. But it raises a practical question with treatment implications: if ADHD involves sleep dysregulation, are stimulants partly working through their arousal-promoting effects rather than purely through dopamine? And could sleep-targeted interventions — like auditory stimulation during overnight sleep, which boosts slow waves in neurotypical populations — reduce daytime local sleep intrusions in ADHD?

Consciousness theories usually live in philosophy seminars, but a new computational proposal is explicitly aimed at engineering.

The "Dual-Laws Model" — a preprint, not yet peer-reviewed — argues that instead of chasing the mysterious "what it feels like" side of consciousness, we should define it through two functional capacities: autonomy in constructing one's own goals, and the ability to mentally decouple from current sensory input. A conscious system, in this framing, isn't just reacting — it can pause, simulate, and re-decide based on internally generated scenarios. The authors lay out seven questions any theory of consciousness should answer, then show how their model could, in principle, be implemented in artificial agents.

Neuroscience enters because similar capacities in humans map onto prefrontal-hippocampal circuits for prospective simulation and large-scale "global workspace" broadcasting. Even if you're skeptical, this is valuable as a design-oriented theory that forces engineers to say which ingredients of brain-like cognition they are — and are not — trying to recreate. Watch for critiques comparing this to Integrated Information Theory and Global Workspace Theory, and for small toy implementations that test whether the two functional capacities produce behavior we intuitively label "conscious."

Three drugs walk into a brain and break the same thing; a 25-year-old plumbing failure finally helps explain why someone hears voices; and the world's largest cannabis review tells millions of patients their medicine lacks evidence of benefit for mental health and carries side effects. Meanwhile, somewhere in a preprint, the hippocampus is quietly applying for a second job in chronic pain — because apparently storing every embarrassing memory you have wasn't enough work.

Stay curious. Stay skeptical. Especially of your own brain's cleaning schedule.

If someone you know would enjoy this, send it their way — brains are better when they're not thinking alone.

The FTC said it would treat certain AI practices as "unfair" under Section 5 of the Federal Trade Commission Act — with direct implications for diagnostic tools and clinical decision-making in psychiatry and neurology. Read → FTC Draws a Line: "Unfair" AI Is Now an Enforcement Target Under Section 5