There's a new tool in Training. It flashes a target for a fixed, very short exposure — as little as 50 milliseconds — then covers it, and asks what you perceived.
Credit where it's due. Lyn Buchanan showed a tachistoscope program he'd built on a recent webinar and openly invited anyone to make an updated version — so we did. The tachistoscope itself is much older, a standard piece of perceptual-psychology equipment. What we took from Lyn was the idea of pointing one at remote viewing practice, and the nudge to go and build it.
What it's for
Remote viewing asks you to work with impressions that arrive before analysis catches up. The hard part usually isn't receiving something — it's catching it before the conscious mind tidies it into something more sensible.
A tachistoscope drills exactly that. The exposure is deliberately too short to study. You can't reason your way through it. Whatever you report is what actually landed.
How a trial runs
- A fixation cross settles your eye.
- The target flashes for the exposure you chose.
- A high-contrast mask covers it.
- You type whatever impression you're left holding.
The mask isn't decoration. Without it, the afterimage on your retina lets you keep reading the picture after it's physically gone, and the exercise stops measuring anything. It's there to end the exposure honestly.
Exposures run from 1000ms down to 50ms. Your browser also times itself, so each trial records the exposure it actually achieved rather than the one it was asked for.
About the scoring — please read this part
The scoring is rough, and you should not take it too seriously. It's genuinely useful as a nudge, and genuinely not a verdict. Here's exactly how it works, so you can judge it for yourself.
Every image carries a hand-written list of descriptor words — things like tall vertical structure, jagged peak, blue white. When you submit, your words are matched against that list by a set of plain, deterministic rules, in order: exact match, then word stems (towers → tower), then a synonym lexicon, then how the word sounds (watter → water), then a tight typo allowance.
There is no AI reading your response. Nothing is interpreting what you meant. It is literal word matching, made a bit more forgiving. That has one real advantage: it's completely auditable — every line of the breakdown tells you which rule fired, so you can always see why something did or didn't count, and tell us when it's wrong.
Two things this means in practice:
- The word list isn't exhaustive. Roughly six descriptors per image, written by hand. You can perceive something entirely real about a picture — a mood, a detail, a relationship between parts — and get no credit simply because nobody thought to write that word down. That's a gap in our list, not a miss on your part.
- The synonym lexicon is partial. It currently holds about 360 words grouped into 66 clusters — colours, shapes, scale, texture, patterns, materials, light and motion — so huge credits massive, coiled credits spiral, gray credits grey. That covers a bit under half the vocabulary in the image descriptions. The rest still needs a fairly close word match. We deliberately keep true synonyms separate where they matter: white will never credit grey, because crediting a wrong colour would make the number meaningless.
So: treat recall and precision as a rough signal over many trials, not a judgment of any single one. The number going up over weeks means something. One low score means very little — it might just mean you used a word we haven't taught it yet.
The part actually worth your attention is the breakdown and the image, not the percentages.
You see the target afterwards
Once a trial is scored, the image appears next to your breakdown. This is where most of the learning actually happens. You'll notice you wrote "tower" where the descriptor said "tall vertical structure", or that the thing you dismissed as noise was the dominant element in the frame.
How to use it
- Start slow. Begin at 1000ms and work down as you get comfortable. 50ms is not a beginner setting.
- Report more than images. If what landed was a texture, a temperature, a sense of scale or motion, write it down.
- Don't tidy it up. "Something red, low, wide" is a better report than a guess at which object it was. Describing beats naming.
- Read the breakdown, not the number. It tells you what you missed and what you invented — the percentage tells you neither.
- Keep sessions short. Ten trials is the default for a reason. This is tiring in a way that sneaks up on you.
What it isn't
This isn't a psi test, and it isn't scored as one. It also won't improve the underlying ability — nothing here claims to make anyone more psychic.
What it trains is the step immediately downstream of that: catching an impression, holding onto it, and getting words onto it before analysis reshapes it into something more reasonable. If the signal is there, this is practice at digesting it.
Tachistoscope results are solo training. They never feed your blind accuracy, your calibration, or your rank — those come only from real blind work, and they stay that way.
Where to find it, and what's next
Training → Tachistoscope. It runs against a pool of 436 freely-licensed images put together for this purpose, chosen so a single flash carries a strong, readable shape rather than fine detail you'd need time to study.
If enough people use it, we'll look at adding AI scoring as a feature — something that can actually read your description against the image and judge meaning rather than matching words. That's the real fix for everything described above. For now, it's honest word matching, and we'd rather tell you that than dress it up.
In the meantime, the single most useful thing you can do is tell us when the scoring gets it wrong. If it missed something you clearly perceived, post the word here — the lexicon grows directly from that, and every word added helps everyone.