A Neural Network Model Of Multisensory Interaction

Multisensory integration is the study of how information from different sensory modalities such as sight, sound, tactile be integrated by the nervous system and finally form the perception of events. Multisensory integration can facilitate human and animals to extract the weak signals from complex, noisy environments.Stochastic resonance is common phenomenon in nonlinear systems. The noise is considered harmful to signals in traditional view while the noise can improve the detection of weak stimulti and enhance system performance when stochastic resonance happens. So the noise is not always harmful, to master their rules, we can benefit from it.Both multisensory integration and stochastic resonance are characteristics of nervous system. In fact, most of time, when multisensory integration happens, there may be stochastic resonance in the process. The purpose of this paper is to study the ordinary multisensory integration as well as multisensory integration in noisy environment.We use modeling and experimental methods. A neural network model is presented. The model includes three neural areas:two unimodal areas and one multisensory area. The two unimodal areas communicate via synaptic with multisensory area. This model can mimic the main characteristics of multisensory integrations:multisensory enhancement, inverse rule and spatial rule. Then the noise is added to this model. We find that in certain range of noise intensity, the activities of multisensory neurons first enhance and then weaken with the increase of noise intensity. This can be considered as stochastic resonance. We have also designed an event-related desynchronization experiment to test whether the noise can enhance the EEG (βandμ) suppression. Different intensities of white noises are applied to subjects, and subjects are asked to move their fingers. We find that the suppression value reach maximum under certain intensity of noise.This paper has studied the ordinary multisensory integration and multisensory integration in noisy environment. We confirm that there can be stochastic resonance in the process multisensory integration. The research of combination of multisensory integration and stochastic resonance is a useful exploration.