Optimal decoding of correlated neural population responses in the primate visual cortex

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Processing

Background:

• Environmental stimuli activate large neuronal populations in sensory cortex.
• The mechanism for reading out these distributed neural responses to guide behavior is not fully understood.

Purpose of the Study:

• To investigate how population neural responses in the primary visual cortex (V1) are decoded for behavior.
• To identify efficient decoding rules for visual information processing.

Main Methods:

• Voltage-sensitive dye imaging was used to measure population responses in the V1 of monkeys.
• Various decoding rules were evaluated for their ability to detect visual targets from neural activity.

Main Results:

• Small visual targets evoked widespread V1 responses with highly correlated variability across distant sites.
• A decoding rule employing spatially antagonistic center-surround summation proved most efficient.
• This optimal decoder surpassed the monkey's performance in the visual detection task.

Conclusions:

• Correlated neural variability in V1 poses challenges for traditional decoding methods.
• Spatially antagonistic center-surround summation offers an efficient strategy for decoding neural population responses.
• Inhibitory mechanisms may play a crucial role in efficient neural decoding.