The timecourse of inter-object contextual facilitation.

Area of Science:

• Cognitive Neuroscience
• Visual Perception
• Human Neuroimaging

Background:

• High-level vision research often focuses on individual objects or entire scenes, neglecting intermediate organizational levels like object constellations.
• Object constellations, familiar groupings of associated objects (e.g., plate and spoon), play a role in scene understanding.
• Previous studies indicate that integrating information from multiple objects aids in high-level scene and object recognition.

Purpose of the Study:

• To investigate when the human visual system integrates information across objects to support recognition.
• To use electroencephalography (EEG) to examine the neural basis of object constellation recognition.
• To assess if real-world size information is inferred more effectively from object pairs than isolated objects.

Main Methods:

• EEG was recorded from human participants viewing masked object constellations and isolated object silhouettes.
• Stimuli varied in real-world size (large vs. small) and retinal size.
• Representational Similarity Analysis (RSA) was employed to decode neural patterns related to real-world size information.

Main Results:

• Neural activity patterns encoded real-world size information for object constellations starting around 200 milliseconds post-stimulus onset.
• This real-world size representation was significantly stronger and more specific for object pairs compared to single objects.
• The findings persisted even after controlling for visual similarity effects.

Conclusions:

• The visual system integrates information across objects in familiar constellations, facilitating recognition.
• Inter-object facilitation leads to a distinct high-level neural representation for object pairs compared to single objects.
• Object constellations represent an intermediate, yet crucial, level of visual organization.