jsPsych / Behavioral experimentsRead · predict · build · inspect

WebGazer: optional observation and methods discussion

This is an optional instructor demonstration, not a required student implementation. A webcam is not needed for credit. Use the official demonstration on a prepared machine or discuss the workflow below. Camera troubleshooting must not displace the core deployment exercise.

Workflow

  1. Explain camera use and data flow before requesting access; participation is optional.
  2. Initialize the integration and camera in a supported secure browser context.
  3. Calibrate using known target positions.
  4. Validate the fitted relationship using separate checks. State an acceptance rule before inspecting results.
  5. If validation is inadequate, recalibrate a bounded number of times or stop.
  6. Run a short task with defined areas of interest; inspect coordinates, times, and quality measures. Close the camera when finished.

Calibration fits a relationship; validation checks its performance. Large areas of interest may be more plausible than adjacent letters or precise saccade timing. Lighting, pose, movement, camera, and browser affect quality. One successful laptop demonstration does not validate a research protocol.

Observation worksheet (10 minutes)

Expected answers: validation checks prediction beyond fitting; compare spatial accuracy/coverage and missing samples with a prespecified criterion; coarse tasks still need piloting; letter-level spatial/temporal precision is a stronger claim; camera permission does not itself establish whether video is stored or transmitted.

Use the official workflow and extension documentation. This repository does not bundle or claim to have hardware-validated WebGazer. The worksheet covers the required methodological discussion when a live demo is unavailable.