Celastrus paniculatus is a climbing shrub native to India, used in Ayurveda for centuries as a cognitive tonic. The seed oil carries an old name, Jyotishmati, roughly “luminous,” and the seeds themselves hold sesquiterpenes and alkaloids. Animal work suggests the oil may raise acetylcholine availability, and that is the thread that makes it interesting for dreaming: cholinergic tone is one of the few levers with a plausible line to REM. That link has never been tested in humans for dream effects, so the rationale going in was extrapolation, not evidence.
The protocol ran seven consecutive nights, taken before bed with no journaling and no technique. Two things changed across the week rather than staying fixed, which makes this less clean than a standard run: the dose climbed from one gram to two, and the preparation moved from chewed seeds with water to seed material blended and soaked in olive oil. That is a deviation from the single-variable standard, but it turned out to be the most informative part.
The early single-gram nights were unremarkable — recall and coherence sat at baseline, sleep was ordinary. The signal, such as it is, appeared later. Recall rose above baseline on three nights, and the strongest of those were the two-gram nights where the seed was soaked in olive oil. On those nights recall, narrative coherence, and subjective sleep quality all ticked up together, and dreams were easier to reach into on waking — more retrievable detail, held longer. The two flat nights in the run both had alcohol beforehand, which is the more likely explanation for them than the compound.
No lucid dream occurred on any night, and hypnagogic imagery and emotional intensity held at baseline throughout. So this is not a lucidity compound on this protocol, and the effect that did show up is modest and confounded by the shifting dose and preparation.
The most useful read is that preparation may matter more than the plant. Celastrus is described as fat-soluble in the traditional accounts, and the recall bump clustered on the oil nights rather than the chewed ones. That is a hypothesis, not a finding — the run varied too many things at once to call it. The clean next test is obvious: hold dose and timing fixed at two grams, vary only fat-based versus water-based preparation, and see whether the oil is doing the work. Filed as a weak positive, worth isolating.