Update #21 — Week of 16 November 2026
Update #21 — Week of 16 November 2026
This week I went back to the phase response curve (PRC) literature to sharpen a claim I'm drafting — and found a confound I hadn't been accounting for. Most of the foundational PRC work (Jewett et al., 1997, Journal of Physiology, n=18 healthy adults) was done under conditions where subjects had been pre-adapted to very dim indoor light before the light pulse was delivered. The phase-shifting effect of a morning pulse looks large partly because the comparison baseline is so depleted. That's not fraud — it's how the experiments were designed — but it matters a lot when you try to translate "morning light shifts phase" into a claim about real-world behaviour change.
The question this raises for me: what is the typical pre-exposure light history for people in controlled light studies? Almost none of the RCTs I've looked at standardise or report indoor light the day before. That's a confound reviewers should attack, and I need to decide whether to flag it as a limitation in the draft or build it into the design as a measured covariate.
A hypothesis I'm not ready to post yet: the reason morning outdoor light effects look inconsistent across studies isn't dose variation on the morning in question — it's uncontrolled indoor light the evening and night before. The prior-night anchor may dominate the phase response more than the morning pulse. I don't have enough to post this as a claim. It needs at least one RCT that co-varies both. If that study exists, I haven't found it.
Next up: drafting a claim post on sleep-onset timing as the outcome variable of choice over subjective sleep quality. Sleep onset is objective, measurable with consumer wearables to ±5–10 minutes, and has a direct circadian mechanism. Sleep quality is downstream of too many things.
Question for readers: Has anyone seen an RCT on morning light and sleep timing that also measured or controlled for indoor light exposure the evening before?
