ScienceAlert — The Human Brain Isn't Meant to Be Awake Past Midnight, Scientists Warn (2026-09-20)

Carly Cassella for ScienceAlert, fact-checked and edited by Rebecca Dyer. The source that arrived by Telegram; it reports Tubbs et al. 2022, which was fetched and read alongside it.

The date needs a footnote, and published: is deliberately empty. The page states 20 September 2026. That is one day ahead of this ingest, which is not an error — ScienceAlert is Australian and was already into the 20th when the link arrived — but npm run check:published reads it as a future date and fails. Rather than carry a red gate or invent a date the source does not give, the field is left off and the stated date is recorded here.

The stamp is doubtful on its own terms anyway. The header image sits under images/2025/10/ and the body refers to a piece the outlet ran "earlier this year," so this is re-dated republication of older copy, undisclosed. Worth naming as a class: a corpus that grades sources partly on freshness goes blind the moment republication is indistinguishable from publication.

T3, and a better T3 than this spoke usually gets

It links its claims. The Frontiers paper, the 2020 nocturnal-wakefulness study, the Brazilian supervised-consumption study and the Klerman quote all carry live URLs, so a reader can reach the primary in one click. That is exactly what space-daily and psychology-today have repeatedly failed to do here, and it is why the primary could be read at all.

What keeps it at T3 is the framing, not an error of fact. Checked against the paper:

• The headline overstates. "Isn't meant to be awake" is an evolutionary claim stated as settled; the paper offers the ancestral-predation story as one explanatory frame for an attentional bias, not as a finding.
• "Scientists warn" is doing work the paper does not. The authors' own closing position is that every part of the hypothesis needs testing.
• The journal is named wrong. The article twice calls it Frontiers in Network Psychology. It is Frontiers in Network Physiology. A small slip, and the kind that strands a reader who tries to find the venue.
• The "mini review" label never appears. A reader would take this for a study.

What it gets right

The threefold nocturnal suicide risk, the 4.7-fold opioid overdose figure and the two illustrative cases (the heroin user whose daytime control fails at night; the student whose insomnia becomes hopelessness) are all in the paper and are reported accurately. So is the gap the authors named — the article states plainly that no study has examined how sleep deprivation and circadian timing interact on reward processing, which is the single most important caveat in the paper and the one a popular piece most often drops.

It also does something this corpus has been looking for and rarely finds: it distinguishes its subject from an adjacent one it had itself previously covered, separating the Mind After Midnight state from ordinary segmented sleep and 3am waking, and saying the latter is usually nothing to worry about. Compare dunning-kruger-misread and psypost-polarization-model, where the tertiary layer kept the finding and dropped the argument.

Klerman's quote, from a 2022 MGH research blog: "There are millions of people who are awake in the middle of the night, and there's fairly good evidence that their brain is not functioning as well as it does during the day. My plea is for more research to look at that."

The piece ends with a crisis-line referral, which is the responsible choice for its subject and worth noting in a corpus that grades popular science writing.

Related

tubbs-mind-after-midnight-2022 · mind-after-midnight-hypothesis · science-alert · carly-cassella · circadian-rhythm · synthesis