Modelling Audio-Visual Multisensory Integration in Alpha-Gamma Bands

We have utilized a neural mass model to simulate the audio-visual multisensory integration. The model can perform some behavioral functions and works in oscillatory conditions similar to EEG (Electroencephalography). It consists of four interconnected neural populations (pyramidal, excitatory interneurons, and fast and slow inhibitory interneurons) which make up one cortical column, equivalent to an ROI (Region of Interest), that can produce oscillations at a certain frequency, similar to an EEG frequency band. We are concerned with conscious neural processing and attention which is why our focus is only on two frequency bands: Alpha and Gamma. The gamma rhythm is typically attributed to conscious neural processing while alpha rhythm is associated with attention mechanism i.e., to inhibit the functioning of a certain neural population. The interconnection of more than one ROI makes up a certain brain area dedicated to specific function. In our model, we have used four brain areas: Two for unisensory processing (one for auditory and one for visual modality), one for multisensory processing, and one for generation of an alpha rhythm. The unisensory and multisensory areas comprise of 180 ROIs to account for 180 degrees in the azimuthal space, while for alpha rhythm generation we used just one ROI which sends its output to other areas. The model can perform different behavioral functions: It can solve causal inference problem in the multisensory area (in case of dual modality inputs) or in unisensory area (in case of multiple unimodal inputs). It can also simulate ventriloquism effect which elaborates the bias or shift in perception of auditory and visual position when the two stimuli are closer to each other in azimuthal space. Finally, the model simulates attention modulation i.e., the human ability to focus on certain stimuli whilst inhibiting other ambient information.