Feedforward and recurrent processing in scene segmentation: electroencephalography and functional magnetic resonance...

Area of Science:

• Neuroscience
• Visual Perception
• Cognitive Science

Background:

• Texture segregation involves distinct processes: boundary detection and surface segregation.
• Neural evidence for texture boundary detection exists in monkey V1, but surface segregation in V1 and human V1 is debated.
• Previous research lacks methods to differentiate boundary detection from surface segregation in humans.

Purpose of the Study:

• To differentiate neural correlates of texture boundary detection and surface segregation in humans.
• To investigate the temporal dynamics and brain regions involved in these processes.
• To determine if these processes occur in early visual areas like V1 in humans.

Main Methods:

• Utilized electroencephalography (EEG) and functional magnetic resonance imaging (fMRI).
• Developed a novel paradigm to distinguish between boundary detection and surface segregation.
• Recorded neural activity and blood-oxygen-level-dependent (BOLD) signals during a texture segregation task.

Main Results:

• EEG data revealed texture boundary detection signals in early visual cortex at 92 msec, spreading to temporal and parietal lobes.
• Surface segregation signals emerged in temporal areas at 112 msec, then spread to parietal and occipital regions.
• fMRI showed correlates of both boundary detection and surface segregation in early visual areas, including V1.
• Both processes were also observed in frontal areas after 208 msec.

Conclusions:

• Texture boundaries are detected via a feedforward mechanism, with signals represented at increasing latencies in higher visual areas.
• Surface segregation appears to follow a reverse hierarchical pattern, driven by feedback signals to early visual areas like V1.