A continuous attractor network model without recurrent excitation: maintenance and integration in the head direction...

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Neural Network Modeling

Background:

• The head direction system provides an internal sense of direction, crucial for navigation.
• Continuous attractor networks offer a framework for stable representation of continuous variables like head direction.
• Experimental data from head direction cells motivates the development of detailed network models.

Purpose of the Study:

• To investigate a three-population network model as a continuous attractor network for head direction representation.
• To determine the conditions for stable spatial activity profiles and accurate integration of angular velocity inputs.
• To compare the model's dynamics with experimental recordings of head direction cells.

Main Methods:

• Analytical methods applied to a simplified threshold-linear neuron model.
• Numerical simulations of a large spiking neuron network.
• Comparison of model dynamics with experimental data from rat head direction cells.

Main Results:

• The model successfully stores head direction as a spatial activity profile without continuous external input.
• Conditions for emergence of spatial selectivity and reliable integration of angular velocity were identified.
• The network demonstrates rapid updating of directional representation, consistent with experimental observations.

Conclusions:

• The proposed three-population network model effectively captures key dynamics of the biological head direction system.
• This model provides a mechanistic explanation for how head direction cells maintain and update directional information.
• The findings support the role of continuous attractor networks in neural computation for spatial orientation.