The entire protocol runs in the participant's browser. Set up the study and share a link; the rest happens on its own.
Upload your images or videos, set the presentation order (fixed or randomized) and, if you want, enable the implicit response module.
Participants join from their own computer. Gaze is calibrated against their own webcam before starting — no software to install.
Gaze, facial expression and, where applicable, response speed to keywords are recorded — all synced to the same timeline.
Heatmaps, emotion curves, attention by area of interest and statistical significance across stimuli — ready to export as a PDF.
Draw a rectangle over a logo, a price or a face. The system tracks it frame by frame and measures how much attention it received — only during the actual window it was on screen.
It's not a single number: it's the overlay of where they looked, what they felt and how fast they reacted — so you can read the stimulus from several angles at once.
A heatmap per stimulus, weighted by participant — not by raw sample, so no one counts extra just for having a better camera or better lighting.
Draw a rectangle over a logo, a price or a face and track the object frame by frame. Each AOI exists only within its real window — not before, not after.
Seven basic emotions, frame by frame, resampled to a common grid to average across participants without sampling bias.
After the stimulus, timed words answered YES/NO. Speed is the signal: fast reveals automatic association, slow reveals rationalization.
Paired t-test per emotion and per attention, with multiple-comparison correction and effect size — same stimulus, same subject, a clean comparison.
A PDF with tables, statistical significance and a supporting read generated from the results — it does not replace a specialist's analysis.
Metrics are averaged per participant before per sample — a subject with more detected frames doesn't weigh more than one with fewer.
If a participant lost tracking quality, they are excluded and the count is reported — never backfilled with an artificial average.
When several stimuli are compared at once, the false discovery rate is adjusted — not just the raw p-value.
Significance alone isn't enough: every comparison also reports how large the difference is, not just whether it's "significant".
This is not a hardware eye tracker. It's an ordinary webcam, calibrated per subject in the browser. The precision is enough to read directional trend — which general zone the gaze went to — not pixel-precise fixation.
That's why every result is delivered with its context: how many participants, how many were left out for signal quality, and from which device. Honesty about the method is part of the method.
Build the protocol, send the link, and watch results accumulate in real time — with the same statistical rigor as an in-person lab.