Abstract:We investigated the dynamical behaviors of bimodular continuous attractor neural networks, each processing a modality of sensory input and interacting with each other. We found that when bumps coexist in both modules, the position of each bump is shifted towards the other input when the intermodular couplings are excitatory and is shifted away when inhibitory. When one intermodular coupling is excitatory while another is moderately inhibitory, temporally modulated population spikes can be generated. On further increase of the inhibitory coupling, momentary spikes will emerge. In the regime of bump coexistence, bump heights are primarily strengthened by excitatory intermodular couplings, but there is a lesser weakening effect due to a bump being displaced from the direct input. When bimodular networks serve as decoders of multisensory integration, we extend the Bayesian framework to show that excitatory and inhibitory couplings encode attractive and repulsive priors, respectively. At low disparity, the bump positions decode the posterior means in the Bayesian framework, whereas at high disparity, multiple steady states exist. In the regime of multiple steady states, the less stable state can be accessed if the input causing the more stable state arrives after a sufficiently long delay. When one input is moving, the bump in the corresponding module is pinned when the moving stimulus is weak, unpinned at intermediate stimulus strength, and tracks the input at strong stimulus strength, and the stimulus strengths for these transitions increase with the velocity of the moving stimulus. These results are important to understanding multisensory integration of static and dynamic stimuli.

What problem does this paper attempt to address?

The paper primarily explores the behavioral characteristics of bimodular Continuous Attractor Neural Networks (CANNs) in processing static and dynamic stimuli. The core objective of the study is to understand the dynamic behavior of bimodular CANNs under different parameter configurations, particularly focusing on the variations in connection weights between interacting modules, input strength, and input differences. Specifically, the paper addresses the following key issues: 1. **Impact of connections between different modules**: The study finds that when bumps coexist in the two modules, their positions change based on the interconnections between the modules. If the interconnections are excitatory, the bumps move towards the input of the other module; if inhibitory, they move away from the input. When one connection is excitatory and the other is moderately inhibitory, periodic population spike activity can be generated. 2. **Population spikes and transient spikes**: When the connection from one module to another is excitatory and the reverse connection is moderately inhibitory, periodic population neuron activity (i.e., population spikes) can be observed. When the inhibitory connection becomes stronger, transient spike activity appears, which is useful for encoding transient information. 3. **Mechanism of bump height variation**: The height of the bumps is influenced by two effects: the excitation effect and the displacement effect. The excitation effect refers to the bump height in the module receiving excitatory input being generally higher than that in the module receiving inhibitory input. The displacement effect indicates that the interaction between modules causes a shift in bump position, which may weaken the effectiveness of the input, thereby affecting the bump height. 4. **Role as a multisensory integration decoder**: By extending the Bayesian framework, the paper demonstrates how excitatory and inhibitory connections can be encoded as attractive and repulsive prior knowledge. When the input differences are small, the bump positions in the modules can decode the posterior mean; when the input differences are large, multiple stable states may exist, and the less stable states can be accessed by introducing more stable inputs with a delay. In summary, this study provides a theoretical foundation for understanding how the brain processes information from different sensory channels and reveals the complex behavior of bimodular CANNs in simulating the process of multisensory information integration.

Bimodular continuous attractor neural networks with static and moving stimuli

Multiperiodicity of Periodically Oscillated Discrete-Time Neural Networks with Transient Excitatory Self-Connections and Sigmoidal Nonlinearities

Distinct excitatory and inhibitory bump wandering in a stochastic neural field

Multiscale Motion and Deformation of Bumps in Stochastic Neural Fields with Dynamic Connectivity

Bump attractor dynamics underlying stimulus integration in perceptual estimation tasks

Bump attractors and waves in networks of leaky integrate-and-fire neurons

Continuous Attractors and Population Decoding Multiple-Peaked Activity

Dynamical Properties of Continuous Attractor Neural Network with Background Tuning

Multiple scroll attractors and multistability in the collective dynamics of a four chain coupled Hopfield inertial neuron network: Analysis and circuit design investigations.

A Moving Bump in a Continuous Manifold: a Comprehensive Study of the Tracking Dynamics of Continuous Attractor Neural Networks.

Non-equilibrium statistical mechanics of continuous attractors

Computing with Continuous Attractors: Stability and Online Aspects

CMCI: A Robust Multimodal Fusion Method for Spiking Neural Networks

Spontaneous Motion on Two-dimensional Continuous Attractors

Neural information processing with feedback modulations.

Dynamics and computation of continuous attractors.

Robustly encoding certainty in a metastable neural circuit model

Dynamics Of Neural Networks With Continuous Attractors

Antagonistic Cross-Links and Reciprocal Couplings Emerge in Optimal Multisensory Integration