 |
Center for Nonlinear Studies |
When viewing a visual scene, human observers typically allocate attention to different regions within the stimulus. The selection of image locations is guided by bottom-up (stimulus-driven) and top-down (goal-directed) mechanisms. Solving a task is a typical top-down influence and strongly affects selection behavior. From studies investigating task performances we know that the informativeness of image parts strongly depends on the task.
I will introduce an experimental paradigm to investigate the relationship between task dependent information of image parts and the observers attention directed toward them. The paradigm is based on the Bubbles paradigm introduced by Gosselin and Schyns (2001). Image parts from forest and face images are shown to subjects in different combinations and at different locations in combination with classification tasks.
Measuring subjects' classification responses, allows us to, first, estimate task dependent information of different image parts and to evaluate different models of information integration. Measuring eye movements, furthermore, allows us to investigate the relationship between estimated information of image parts and allocation of attention.
Host: Garrett Kenyon