The study

What is being tested

What came before

The confirmatory session is 200 trials long, at 0.005 one-sided on the V-test (modified Rayleigh), wrapped error, hypothesised direction 0°, one-sided. Those three numbers were fixed before any human was run, and the reason is the history below.

1. Early reports

   A long history of informal accounts of noticing an unseen gaze — almost none of it instrumented, and almost none of it blinded.

2. The first controlled tests

   The earliest laboratory tests, around the turn of the twentieth century, treated the belief as a superstition to be dismantled — and found nothing beyond chance.

3. The remote-staring literature

   Later work reported positive results, but the strongest of them came from designs with trial-by-trial feedback and imperfectly random sequences. When careful replications removed both, the effect disappeared, and the field’s own meta-analysis shrinks toward zero in its best-evidence subset.

4. Why it stayed inconclusive

   Feedback that teaches, sequences that are not truly random, yes/no response bias, experimenter effects, ordinary gaze cues leaking through imperfect shielding, and analysis decisions made after seeing the data. The full account, with references, is a Research Room document.

5. The methodological gap

   Every controlled study located by the review asks a binary question: “Am I being stared at — yes or no?” None of them measured the direction of an unseen observer.

6. What this study asks instead

   A directional question: from which direction is the unseen observer located? The receiver commits a continuous bearing around the ring, so a yes/no bias has no preferred direction to exploit, and chance is one sector of several — not one flip of two.

What either outcome would mean

A positive GEO result would mean that a condition-blinded receiver's reported bearings contain statistically detectable directional information about the true location of the unseen observer, beyond what chance predicts, under the pre-specified analysis. The confirmatory endpoint is the receiver's committed directional bearing; eye tracking is not the primary confirmatory endpoint. By itself, the result identifies no physical carrier and no mechanism.

A null result is meaningful in the same plan: under the tested conditions, at the locked power and analysis, no detectable directional effect was observed — which constrains how large such an effect could be. Both outcomes are reportable. That posture is built into the frozen plan rather than decided after the numbers arrive.

What a GEO result would establish — and what comes next

GEO — the confirmatory test

Whether a condition-blinded receiver can localize the direction of an unseen observer above chance. That is the whole confirmatory claim.

The interpretation boundary

A positive GEO result alone does not identify a physical carrier or mechanism, and this protocol asserts none.

Cue discrimination — KO-OPT

Materials and optical controls test whether any observed directional effect follows the predicted condition/material signature rather than ordinary sensory cue leakage.

Future instrumentation

If a directional effect is reproducibly observed, follow-on work may instrument the receiver more deeply — eye tracking or other physiological measurements — to investigate how the directional response manifests. That instrumentation is exploratory, not part of the current confirmatory endpoint.