Predictive feedback can account for biphasic responses in the lateral geniculate nucleus

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Processing

Background:

• Biphasic neural response properties, characterized by shifting optimal stimulus patterns, are observed in visual areas like LGN, V1, and MT.
• These properties suggest complex temporal dynamics in neural information processing.

Purpose of the Study:

• To develop and simulate a hierarchical predictive coding model.
• To capture and explain the temporal variations in neuronal response properties within the LGN-V1 circuit.

Main Methods:

• A hierarchical predictive coding model was developed and simulated.
• The model was trained on natural images to replicate LGN-V1 connectivity.
• Neuronal response profiles and feedback mechanisms were analyzed.

Main Results:

• The model successfully replicated the LGN-V1 connectivity, linking V1 receptive fields to LGN center-surround cell properties.
• Simulated LGN neurons exhibited biphasic spatio-temporal response profiles, consistent with experimental findings in cats.
• A phase-reversed feedback pattern was observed, where aligned LGN receptive fields of the same polarity decreased responses, while opposite polarities increased responses.

Conclusions:

• The model's results support the hypothesis that predictive feedback is a fundamental coding strategy in the brain.
• The findings elucidate the mechanisms underlying biphasic neural responses in the visual system.