A recurrent circuit implements normalization, simulating the dynamics of V1 activity.

Area of Science:

• Neuroscience
• Computational Neuroscience

Background:

• The normalization model explains neural activity in systems like the primary visual cortex (V1).
• It involves dividing neuronal responses by a weighted sum of activity from a neuron pool.
• However, its link to recurrent amplification, weighted connections, and complex dynamics remains unclear.

Purpose of the Study:

• To investigate how normalization arises from recurrent amplification in neural circuits.
• To understand the emergence of weighted normalization within recurrent networks.
• To explore how normalization contributes to complex neural dynamics, such as gamma oscillations in V1.

Main Methods:

• Developed a family of recurrent circuit models composed of coupled neural integrators.
• Implemented normalization through recurrent amplification with adjustable weights.
• Simulated models to analyze the emergence of normalization properties and neural dynamics.

Main Results:

• The proposed models successfully implement normalization via recurrent amplification.
• These models demonstrate how arbitrary normalization weights can arise from recurrent circuit interactions.
• The models can recapitulate key experimental observations of neural activity dynamics in V1, including gamma oscillations.

Conclusions:

• Recurrent circuit models provide a mechanistic explanation for normalization in V1.
• These models bridge the gap between normalization, recurrent amplification, weighted connections, and neural dynamics.
• The findings offer insights into the computational principles underlying neural processing in the visual cortex.