Description
Eye-track many viewers watching the same video and measure frame by frame how spread out their gaze is, showing when attention converges and how cuts disrupt it.
For each video frame, the gaze points of all viewers are clustered or enclosed in an ellipse holding most of them. Low dispersion means viewers look at the same place at the same time, called attentional synchrony, which indicates attention driven by the narrative or by strong salience. Adults converge strongly on edited video; infants are more dispersed, especially just after a camera cut, and converge over the course of a shot, suggesting they need more time to process each scene. Plotting dispersion against time since each cut gives recovery curves. Because strong bottom-up features such as motion also produce convergence, results should be compared with a saliency model. For UX, the metric turns eye-tracking of videos, animations, tutorials or dynamic dashboards into a time series showing which moments hold shared attention and which scatter it.
Attributes
- Discipline
- Neuro, Biometrics & Affect
- Loss type
- Unmigrated
- Stage
- Evaluation, Design
- Data
- Quantitative
- Effort
- Weeks
Steps
- Eye-track a group of viewers (for example 20 or more) watching the same video or animated sequence.
- Align gaze samples to video frames.
- For each frame compute dispersion, such as the area of the ellipse covering most gaze points or the number of gaze clusters.
- Mark events such as cuts, scene changes, captions or calls to action on the timeline.
- Plot dispersion against time since each event and compare viewer groups or video versions.
- Compare with a computational saliency model to separate story-driven from motion-driven convergence.
Sources
- Handbook of Child Psychology Vol 2 (Lerner et al., 2015)
Also known as
- Between-viewer gaze dispersion on dynamic video
- Attentional synchrony
- Gaze coherence
Categories
- Analytics & Behaviour
- Neuro, Biometrics & Affect