fMRI orientation decoding in V1 does not require global maps or globally coherent orientation stimuli.

Area of Science:

• Neuroscience
• Visual Perception
• Neuroimaging

Background:

• Functional magnetic resonance imaging (fMRI) at 3 Tesla can decode visual orientation from V1 data.
• This capability may stem from columnar-level information or global orientation-preference maps.
• Global maps could arise from bottom-up processing or top-down/recurrent mechanisms.

Purpose of the Study:

• To investigate if fMRI orientation decoding requires globally coherent stimuli or global V1 activity patterns.
• To determine if fine-grained fMRI signals contribute to orientation decoding.

Main Methods:

• Used fMRI with 2-mm voxels to decode orientation from standard and patch-swapped gratings and spirals.
• Patch-swapped stimuli presented opposite orientations globally while appearing globally similar.
• Applied spatial high-pass filtering to analyze fine-grained fMRI pattern components.

Main Results:

• Orientation was robustly decodable from all stimulus types, including patch-swapped variants.
• Decoding remained strong after spatial high-pass filtering, indicating reliance on fine-grained fMRI patterns.
• Global orientation preference maps were present for standard stimuli but weak/absent for patch-swapped stimuli.

Conclusions:

• fMRI orientation decoding does not necessitate globally coherent visual stimuli.
• Fine-grained fMRI signal components are crucial for decoding visual orientation.
• Global orientation preference maps in V1 likely depend on globally coherent orientations and potentially top-down processing.