Saliency on a natural scene background: Effects of color and luminance contrast add linearly

Engmann, Sonja; Hart, Bernard Marius ’t; Sieren, Thomas; Onat, Selim; König, Peter; Einhäuser, Wolfgang

doi:10.3758/app.71.6.1337

Cited by 36 publications

(21 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Bottom-up attentional allocation results from the elements of a scene directly attracting the spotlight of attention [53]. This relates to a whole hierarchy of visual features that may be combined linearly or non-linearly [34], [54]–[56]. Additionally, attentional resources can also be directed in a top-down manner by the task and meaning of different image components [2].…”

Section: Discussionmentioning

confidence: 99%

“…Worse, non-linear scaling of a feature has a large effect qualitatively changing its contribution in a linear combination of all features. As a consequence, investigations of properties of integration of different features [34], [55], [56] depend on the chosen scale. For example, is the standard deviation or the variance (square of standard deviation) of luminance in a local region the better measure for luminance contrast?…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

The Contributions of Image Content and Behavioral Relevancy to Overt Attention

et al. 2014

Self Cite

View full text Add to dashboard Cite

During free-viewing of natural scenes, eye movements are guided by bottom-up factors inherent to the stimulus, as well as top-down factors inherent to the observer. The question of how these two different sources of information interact and contribute to fixation behavior has recently received a lot of attention. Here, a battery of 15 visual stimulus features was used to quantify the contribution of stimulus properties during free-viewing of 4 different categories of images (Natural, Urban, Fractal and Pink Noise). Behaviorally relevant information was estimated in the form of topographical interestingness maps by asking an independent set of subjects to click at image regions that they subjectively found most interesting. Using a Bayesian scheme, we computed saliency functions that described the probability of a given feature to be fixated. In the case of stimulus features, the precise shape of the saliency functions was strongly dependent upon image category and overall the saliency associated with these features was generally weak. When testing multiple features jointly, a linear additive integration model of individual saliencies performed satisfactorily. We found that the saliency associated with interesting locations was much higher than any low-level image feature and any pair-wise combination thereof. Furthermore, the low-level image features were found to be maximally salient at those locations that had already high interestingness ratings. Temporal analysis showed that regions with high interestingness ratings were fixated as early as the third fixation following stimulus onset. Paralleling these findings, fixation durations were found to be dependent mainly on interestingness ratings and to a lesser extent on the low-level image features. Our results suggest that both low- and high-level sources of information play a significant role during exploration of complex scenes with behaviorally relevant information being more effective compared to stimulus features.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

The Contributions of Image Content and Behavioral Relevancy to Overt Attention

et al. 2014

Self Cite

View full text Add to dashboard Cite

show abstract

“…In this view, texture contrast, which by definition is a V-shaped function of contrast, rather than first-order contrast drives attention. However, the V-shaped effect of contrast has also been in seeming conflict with the positive correlation of fixations with contrast and with the observation that for contrast modifications that ramp gradually over the whole image (either from α = −1 at one end to α = +1 at the other or between α = 0 and α = +1 or α = −1), fixation probability scales linearly with this contrast modification [50,51]. This linear effect for large-scale ramps was observed notably for the same instruction (‘study the images carefully’) that yielded the V-shaped function of local modifications in Einhäuser & König [35] and is also used here, refuting one concern put forward in Parkhurst & Niebur [38].…”

Section: Discussionmentioning

confidence: 99%

Attention in natural scenes: contrast affects rapid visual processing and fixations alike

Hart

Schmidt

Klein-Harmeyer

et al. 2013

Phil. Trans. R. Soc. B

Self Cite

View full text Add to dashboard Cite

For natural scenes, attention is frequently quantified either by performance during rapid presentation or by gaze allocation during prolonged viewing. Both paradigms operate on different time scales, and tap into covert and overt attention, respectively. To compare these, we ask some observers to detect targets (animals/vehicles) in rapid sequences, and others to freely view the same target images for 3 s, while their gaze is tracked. In some stimuli, the target's contrast is modified (increased/decreased) and its background modified either in the same or in the opposite way. We find that increasing target contrast relative to the background increases fixations and detection alike, whereas decreasing target contrast and simultaneously increasing background contrast has little effect. Contrast increase for the whole image (target + background) improves detection, decrease worsens detection, whereas fixation probability remains unaffected by whole-image modifications. Object-unrelated local increase or decrease of contrast attracts gaze, but less than actual objects, supporting a precedence of objects over low-level features. Detection and fixation probability are correlated: the more likely a target is detected in one paradigm, the more likely it is fixated in the other. Hence, the link between overt and covert attention, which has been established in simple stimuli, transfers to more naturalistic scenarios.

show abstract

“…In line with object-based accounts (Einhäuser et al, 2008b; Nuthmann and Henderson, 2010) of fixations in natural scenes, such extreme modifications may be interpreted as being such an oddity relative to the background that they gain object or proto-object quality and thus attract attention. Such an interpretation is further supported by the fact that for large scale contrast modifications, the effect of contrast on salience (operationalized as the probability to fixate a certain contrast level) is linear (Einhäuser et al, 2006; Engmann et al, 2009). In a recent paper ('t Hart et al, 2013), we extended the contrast-modification paradigm to images that contain a single nameable object (rather than only foliage), and found the V-shape of contrast on fixation probability to persist when a random location is modified.…”

Section: Introductionmentioning

confidence: 98%

Fixations on objects in natural scenes: dissociating importance from salience

Hart¹,

Schmidt²,

Roth³

et al. 2013

Front. Psychol.

Self Cite

View full text Add to dashboard Cite

The relation of selective attention to understanding of natural scenes has been subject to intense behavioral research and computational modeling, and gaze is often used as a proxy for such attention. The probability of an image region to be fixated typically correlates with its contrast. However, this relation does not imply a causal role of contrast. Rather, contrast may relate to an object's “importance” for a scene, which in turn drives attention. Here we operationalize importance by the probability that an observer names the object as characteristic for a scene. We modify luminance contrast of either a frequently named (“common”/“important”) or a rarely named (“rare”/“unimportant”) object, track the observers' eye movements during scene viewing and ask them to provide keywords describing the scene immediately after. When no object is modified relative to the background, important objects draw more fixations than unimportant ones. Increases of contrast make an object more likely to be fixated, irrespective of whether it was important for the original scene, while decreases in contrast have little effect on fixations. Any contrast modification makes originally unimportant objects more important for the scene. Finally, important objects are fixated more centrally than unimportant objects, irrespective of contrast. Our data suggest a dissociation between object importance (relevance for the scene) and salience (relevance for attention). If an object obeys natural scene statistics, important objects are also salient. However, when natural scene statistics are violated, importance and salience are differentially affected. Object salience is modulated by the expectation about object properties (e.g., formed by context or gist), and importance by the violation of such expectations. In addition, the dependence of fixated locations within an object on the object's importance suggests an analogy to the effects of word frequency on landing positions in reading.

show abstract

Saliency on a natural scene background: Effects of color and luminance contrast add linearly

Cited by 36 publications

References 64 publications

The Contributions of Image Content and Behavioral Relevancy to Overt Attention

The Contributions of Image Content and Behavioral Relevancy to Overt Attention

Attention in natural scenes: contrast affects rapid visual processing and fixations alike

Fixations on objects in natural scenes: dissociating importance from salience

Contact Info

Product

Resources

About