Scientific background

"The eyes are the mirror of the soul" - this yiddish proverb can be taken as one of the fundamental premises of neuroscientific attention research: it is what we look at that we pay most attention to - and it is attention that comes prior to other cognitive processes such as understanding, believing and desiring.

University of Osnabrück, Campus Westerberg

But what is it that attracts our attention? And how do brain mechanisms account for that?

We want to answer these questions by combining eye-tracking studies with computational simulations of the human visual cortex. Eye-tracking studies help us to understand which features of images and scenes work best to attract people's attention, while the predictive performance of our models allows us to draw inferences about how attentional mechanisms in the brain actually work.

Prof. Dr. Peter König

Our research is carried out within the Neurobiopsychology group at the Institute of Cognitive Science, University of Osnabrück. The head of the workgroup is renowned neuroscientist Prof. Dr. Peter König.

Data acquisition with Eye-Tracking

Eye-Tracker as used by the
Neurobiopsychologie workgroup

Eye tracking allows us to measure where and what people look at. By targeted modifications of visual stimuli we investigate underlying mechanisms, trying to find out why they looked where they did. At the Neurobiopsychology lab we use a high precision head mounted system, which measures eye movements with cameras placed below the eyes of an experimental subject, as well as a so-called remote Eye-Tracker, where the camera is placed at the screen. From the recorded eye positions, these devices can compute with high precision which screen position a subject's gaze was directed at while he or she was watching images or other visual stimuli.

Modeling of visual attention

We want to gain a better understanding of what is happening in the brain when we view an image and decide where to look next. To this end, we develop computational models that mimic human gaze behavior. Here, we combine neuroscientific findings about visual processing in the brain with methods of machine learning. It is known that certain neurons in visual brain areas compute image statistics. By calculating these statistics and integrating them,taking into account the data gained from our experimental subjects, we get an idea of how this integration, and the control of attention resulting from it, might be carried out in the brain. Employing machine learning techniques, we can optimize the weighting of individual image features in our models, such that our predictions closely resemble real human gaze behavior.

This is a short selection of some publications by members of the neurobiopsychology group where GoodGaze is developed. For a complete list, see the Neurobiopsychology website .

Salazar RF, Kayser C, König P.
Effects of training on neuronal activity and interactions in primary and higher visual cortices in the alert cat.
Journal of Neuroscience 2004 Feb 18;24(7):1627-36.

Nagel SK, Carl C, Kringe T, Märtin R, König P.
Beyond sensory substitution--learning the sixth sense.
Journal of Neural Engineering 2005 Dec;2(4):R13-26.

Frey HP, König P, Einhäuser W.
The role of first- and second-order stimulus features for human overt attention.
Perception & Psychophysics 2007 Feb;69(2):153-61.

Onat S, Tichacek K and König P (2007).
Integrating audio-visual information for the control of overt attention.
Journal of Vision 7(10):11, 1-16.

Einhäuser W, Rutishauser U, Frady P, Nadler S, König P and Koch Ch
The relation of phase noise and luminance contrast to overt attention in complex visual stimuli.
Journal of Vision 2006 Oct 9;6(11):1148-58.

Einhäuser W, Schumann F, Bardins S, Bart K, Böning G, Schneider E and König P (2007).
Human eye- head co-ordination in natural exploration.
Network: Computation in Neural Systems 18(3):267-297. Electronical publication