Fixation
also: gaze fixation, fixation point
A fixation is a brief pause of the eye on one spot — typically 100–300 milliseconds in reading and scene-perception research — during which visual detail is actually taken in. Between fixations the eye jumps in fast movements called saccades, and almost nothing is perceived mid-jump.
Vision is not a smooth video feed. Sharp detail exists only in the fovea, a region about the size of a thumbnail at arm's length, so the eye samples a scene as a sequence of fixations linked by saccades. What gets fixated gets processed; everything else stays peripheral blur — which is why an element can be 'on screen' and still functionally invisible.
For a web page, the fixation sequence is the real reading order, and it rarely matches the layout order the designer intended. The first handful of fixations decide what a visitor believes the page is about — they land on faces, high-contrast text, and strong focal points long before anyone reads a paragraph.
Fixations are recorded in the lab with infrared eye trackers, then aggregated across viewers into fixation-density maps. Those maps are exactly what saliency models are trained on: datasets like SALICON and MIT300 pair images with real fixation data, and a model like UNISAL — the one behind Heatpoints — learns to predict the density map for an image it has never seen.
In practice you rarely need individual fixations; you need the aggregate: where would a crowd's first fixations concentrate on this design? That is what a predicted attention heatmap shows, and why the hottest zone should coincide with your headline or CTA rather than a decorative photo.
See it on your own page
Free score /100 · no signup · results in seconds
RELATED TERMS