Faces and text attract gaze independent of the task: Experimental data and computer model.

By Moran Cerf, E. P. Frady, Christof Koch. Journal of Vision, 2009

419 citations recorded by the source corpus.

Previous studies of eye gaze have shown that when looking at images containing human faces, observers tend to rapidly focus on the facial regions. But is this true of other high-level image features as well? We here investigate the extent to which natural scenes containing faces, text elements, and cell phones-as a suitable control-attract attention by tracking the eye movements of subjects in two types of tasks-free viewing and search. We observed that subjects in free-viewing conditions look at faces and text 16.6 and 11.1 times more than similar regions normalized for size and position of the face and text. In terms of attracting gaze, text is almost as effective as faces. Furthermore, it is difficult to avoid looking at faces and text even when doing so imposes a cost. We also found that subjects took longer in making their initial saccade when they were told to avoid faces/text and their saccades landed on a non-face/non-text object. We refine a well-known bottom-up computer model of saliency-driven attention that includes conspicuity maps for color, orientation, and intensity by adding high-level semantic information (i.e., the location of faces or text) and demonstrate that this significantly improves the ability to predict eye fixations in natural images. Our enhanced model's predictions yield an area under the ROC curve over 84% for images that contain faces or text when compared against the actual fixation pattern of subjects. This suggests that the primate visual system allocates attention using such an enhanced saliency map.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/14d146b099b12af9b6d7c13342d9ca9db9984275

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

Citations in the corpus

SALICON: Saliency in Context (2015, background): “While instructions are known to affect eye movement patterns, most common in eye-tracking dataset is the use of a so-called “free-viewing” task [5, 18, 4, 27] due to its taskfree nature.”
Quantitative Analysis of Human-Model Agreement in Visual Saliency Modeling: A Comparative Study (2013, methodology): “Context [3], [35], [45], gain modulation of features for target detection [4], [5], [80], and use of target detectors tuned to specific objects [50], [69] has been used for modeling topdown attentional effects.”
A Benchmark of Computational Models of Saliency to Predict Human Fixations (2012, background): “To facilitate comparisons, several authors have produced openly available databases of images with fixations from eye tracking experiments [41] [42] [10] [6] [37] [61] and executable code of models that others can use.”
Depth Matters: Influence of Depth Cues on Visual Saliency (2012, background): “Most eye tracking datasets [7,12,13,5] are constructed for 2D scenes and most saliency models only investigate the cues from 2D images or videos.”
Eye tracking assisted extraction of attentionally important objects from videos (2015, background): “It is well known that faces attract visual attention [9].”
A deep multi-level network for saliency prediction (2016, background): “Similarly to Cerf et al. [7], Judd et al. [19] propose an approach that combines low-level features (color, orientation and intensity) with high-level semantic information (i.e. the location of faces, cars and text) and show that this solution signiﬁcantly improves the ability to predict eye…”
Action from Still Image Dataset and Inverse Optimal Control to Learn Task Specific Visual Scanpaths (2013, background): “Most of these AOIs fall on faces, an effect that has also been noted in [6].”
Predicting human gaze beyond pixels. (2014, methodology): “) and text have been shown to be salient (Cerf et al., 2009), and are explicitly defined as objects.”
Processing of native and foreign language subtitles in films: An eye tracking study (2012, background): “…it has been shown that text in a visual scene is also salient and therefore likely to attract participant’s gaze, even when it is not relevant to the task at hand (Cerf, Frady, & Koch, 2009), which could account for the reading of the subtitles even in the conditions where the subtitles are in FL.”