Knowledge-driven perceptual organization reshapes information sampling via eye movements.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computer Vision

Background:

• Human eye movements are crucial for environmental exploration.
• The role of image-computable features versus object representations in guiding gaze is debated.
• Emerging theories suggest object knowledge reconfigures feature space for object perception.

Purpose of the Study:

• To investigate whether object-oriented eye movements are driven by image-computable features or their integration into object representations.
• To explore how acquired object knowledge influences visual exploration patterns.

Main Methods:

• Recorded human eye movements while participants viewed stimuli.
• Stimuli were initially perceived as meaningless patches, later organized into coherent objects through acquired knowledge.
• Compared eye movement patterns before and after object knowledge acquisition.

Main Results:

• Fixations became more object-centered, less dispersed, and more consistent after images were organized into objects.
• Gaze guidance shifted from exploratory sampling to exploiting object-related image areas.
• These effects were observed from the initial fixations onward.

Conclusions:

• Eye movements are not solely determined by knowledge-dependent object representations.
• Gaze guidance results from an integration of object representations and image-computable features.
• Visual information sampling is dynamically guided by the interplay between image features and knowledge-driven perceptual organization.