Impoverished second-order input to global linking in human vision

Area of Science:

• Neuroscience
• Visual Perception
• Computational Neuroscience

Background:

• Global visual processing involves integrating information across larger spatial regions than individual receptive fields.
• Network operations between local detectors are crucial for tasks like contour integration and motion trajectory detection.
• The role of second-order visual detectors in these global operations remains unclear.

Purpose of the Study:

• To investigate whether network operations occur between second-order visual detectors.
• To compare the contribution of first-order and second-order detectors to global visual tasks.

Main Methods:

• Compared human observer performance on contour integration and motion trajectory detection tasks.
• Used stimuli composed of either first-order or second-order elements, equated for visibility.
• Assessed performance when identifying extended contours or motion trajectories embedded in noise.

Main Results:

• First-order elements showed significant linking interactions, consistent with previous findings.
• Second-order elements exhibited little to no linking interaction between local detectors.
• Performance was near chance for tasks involving second-order elements, indicating poor integration.

Conclusions:

• Network operations crucial for global visual tasks receive limited input from second-order visual detectors.
• Unlike first-order detectors, second-order detectors do not effectively support global contour and motion integration.
• This suggests a fundamental difference in how first-order and second-order attributes contribute to visual scene understanding.