Contributed talks II: Colour-selective regions of visual cortex are responsive to the colour statistics of objects

Area of Science:

• Neuroscience
• Computational Vision
• Psychophysics

Background:

• Previous research suggests objects possess distinct color properties compared to their backgrounds.
• Understanding how the brain processes these color statistics is crucial for object recognition.

Purpose of the Study:

• To investigate the color statistics of objects versus backgrounds in natural scenes.
• To identify brain regions responsive to these object color statistics using functional magnetic resonance imaging (fMRI).
• To determine if the ventral visual pathway responds to object color probability (OCP) independently of other scene information.

Main Methods:

• Analysis of the Natural Scenes Dataset (NSD) to quantify color differences between object classes and backgrounds.
• fMRI study correlating brain activity (BOLD responses) with calculated Object Colour Probability (OCP).
• Utilized independent functional localizers to identify color- and object-selective visual areas.

Main Results:

• Object pixels were confirmed to be warmer, redder, more saturated, and darker than background pixels.
• Object Colour Probability (OCP) correlated with BOLD responses primarily in the ventral visual pathway.
• OCP of color patches correlated with BOLD signals in color-selective, but not object-selective, visual regions in a simplified visual task.

Conclusions:

• The ventral visual pathway is sensitive to the statistical regularities of object colors in natural scenes.
• Color-selective brain regions show a specific response to object color probability, suggesting a role for color statistics in visual processing.