Sensory detection of signals from the physical surroundings, along with abilities in various cognitive domains such as inference, reasoning, decision making, and problem solving, enable us to perceive and interact with our environment. This ability to think is founded in the a priori knowledge of concepts. Different domains of cognition are realized by different forms of such abstract knowledge.

Every scene is a perceptual construction that constitutes a part of our perceptual world. At the same time, each scene is the instantiation of an abstract a priori concept in our memory. The components of a scene are arranged with a particular logic to create a specific function. The layout of a scene, i.e. the symbol of its concept, refers to an abstract form that summarizes the minimum essential information within its concept, which recognizes the scene pre-attentively. Any abstract concept, e.g. the perceptual memory of a scene, is represented in a network of neural populations. When faced with a scene, its conceptual network first detects the overall scheme of the scene and then directs attention, by planning a saccade sequence, to different points in the scene, so that it forms the perception of the scene as an instance of its abstract concept.

The question is how the prior knowledge of the concept of a scene is controlling the pattern of eye movements in a scene. To answer this, we have designed a 3D complex scene (different concepts are combined to make a functional scene) in virtual reality, in order to study the relation between the highlighted concept in a scene and the pattern of saccades planned. We have two groups of subjects. Each group is required to perform a defined task, including recalling the whole or part of the scene, or recalling a part of the scene and finding a specific object in the scene. We have analyzed the correlations between the nature of the task, and the nature of the concept of the scene, with the sequence of fixation points, in order to answer the proposed question.

-