Multi-sensory integration in the mouse cortical connectome using a network diffusion model

Abstract

8Having a structural network representation of connectivity in the brain is instrumental in analyzing 9 communication dynamics and information processing in the brain. In this work, we make steps towards 10 understanding multi-sensory information flow and integration using a network diffusion approach. In par-11 ticular, we model the flow of evoked activity, initiated by stimuli at primary sensory regions, using the 12 Asynchronous Linear Threshold (ALT) diffusion model. The ALT model captures how evoked activity that 13 originates at a given region of the cortex "ripples through" other brain regions (referred to as an activation 14 cascade). By comparing the model results to functional datasets based on Voltage Sensitive Dye (VSD) 15 imaging, we find that in most cases the ALT model predicts the temporal ordering of an activation cascade 16 correctly. Our results on the Mouse Connectivity Atlas from the Allen Institute for Brain Science show 17 that a small number of brain regions are involved in many primary sensory streams -the claustrum and the 18 parietal temporal cortex being at the top of the list. This suggests that the cortex relies on an hourglass ar-19 chitecture to first integrate and compress multi-sensory information from multiple sensory regions, before 20 utilizing that lower-dimensionality representation in higher-level association regions and more complex 21 cognitive tasks. 22The final network, denoted as N c , consists of 617 directed edges between 67 nodes -each node corre-126 sponds to one of the ROIs in our model. The density of N c is 13.9%. The distribution of edge weights is 127 skewed, with 80% of the edges having a weight of less than 5 and few edges having a weight of up to 40. The 128 distribution of edge lengths is almost uniform in the 1-7mm range. The diameter of the network (maximum 129 shortest path length between any two nodes) is 7 hops, while a node is on the average about 4 hops away from 130 any other node. The average in-degree of each node is 9.2 connections (σ=3.1), while the out-degree distri-131 bution has the same mean but larger variability (σ=8.3). Additionally, the network N c is strongly clustered, 132 with an average clustering coefficient of 60% [10]. 133The ten primary sensory regions associated with the visual, auditory, gustatory, olfactory systems, as well 134 as six somatosensory regions for different body parts (see Table SI-1), have a special role in our analysis: 135 they are viewed as sources of sensory information in the cortex [4,61]. The location of these ROIs in the 136 Allen Mouse Brain Atlas is shown in Figure 2. 137 6

Unveiling the sensory and interneuronal pathways of the neuroendocrine connectome inDrosophila

“…neurotransmitter) of the neurons involved. Such networks have been recently used successfully in the mouse to model sensory impact on activity in higher brain centers of the thalamus (Shadi et al, 2020)…”