Do we look at what matters, or at what stands out?
Objects that people consider important in a scene are looked at more and more centrally, and boosting an object's contrast draws the eyes to it without changing that underlying importance effect.
Source
Fixations on objects in natural scenes: dissociating importance from salience
Study at a glance
- Design
- Human experiment — Eye-tracking during 3-s free viewing of photographs in which contrast of one object ('common' = often named as characteristic, or 'rare') was decreased (Exp 1) or increased (Exp 2), followed by typing up to 5 keywords describing the scene
- N
- N=36 · 36 volunteers in total: 16 in Experiment 1 (contrast decreases) and 20 in Experiment 2 (contrast increases plus neutral images); each scene was seen in each condition by four observers
- Population
- Healthy German-speaking adult volunteers naive to the study's purpose
- Outcome
- Proportion of fixations landing on the object, whether the object was mentioned in keywords, and fixation distance from the object's centre
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Participants viewed art photographs of everyday scenes for 3 seconds while their eye movements were recorded, then typed up to five keywords describing the scene. In each scene the researchers picked a 'common' object that earlier viewers often named as characteristic and a size-matched 'rare' object seldom named, and altered the luminance contrast of one of them. Experiment 1 reduced contrast by 20% or 80%; Experiment 2 increased it by 75% or 150% relative to a lower-contrast neutral version.
What they found
Common objects drew a larger share of fixations than rare objects in both experiments. Raising an object's contrast increased fixations on it in a roughly linear way whether it was common or rare, but lowering contrast had almost no effect. Any contrast increase made rare objects as likely to be mentioned as common ones, regardless of its size, yet common objects still received more fixations. Fixations on common objects also landed closer to the object's centre than fixations on rare objects.
The limits
What it doesn't show
The samples were small (16 and 20 observers, with four observers per scene and condition), and the authors acknowledge a contrast effect on where fixations land within objects might appear with more power. The scenes were artistically composed photographs chosen to contain many objects, so the findings may not apply to arbitrary real-world scenes, and hidden object properties such as foreground versus background position were not fully controlled; common objects tended to sit lower in the image. The two experiments used different neutral baselines, so they cannot be directly compared with each other. Importance was measured only by naming in keywords, which may be influenced by language.
Key terms
- Salience
- Here, how likely an object is to be fixated; more broadly, how much something stands out and attracts attention.
- Object importance
- How characteristic an object is for a scene, measured here by how many observers named it as a keyword.
- Luminance contrast
- The difference in brightness between parts of an image; a low-level visual feature used in saliency models.
- Fixation
- A pause of the eyes on one location between rapid eye movements (saccades), used as a proxy for where attention is directed.
- Saliency map model
- A computational model predicting where people will look from low-level features like contrast, colour and orientation.
Flashcards
0 of 11 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What happened to fixations on an object when its contrast was reduced?
Common questions
If making an object brighter draws the eyes, why don't saliency models fully explain gaze?
Because common, meaningful objects kept attracting more fixations than rare ones even when contrast was changed, showing that higher-level knowledge about the scene also guides where we look.
Why didn't lowering contrast reduce fixations?
The authors suggest two effects cancel out: a dimmer object is less visible, but it also looks odd compared with the rest of the scene, which can attract attention.
What does 'dissociating importance from salience' mean here?
Contrast changes altered how important a rare object seemed (it got mentioned as often as the common one) without erasing the fixation advantage of the originally important object, so the two measures came apart.
More on Attention