A selective diffusion model of brain network activity

Graham, Daniel J.; Hu, Yan

doi:10.32470/ccn.2018.1195-0

Cited by 2 publications

(4 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Collision rules are a dimension worth exploring but as an initial study, we adopted a simplified model that promotes interpretability. Graham and Hao, 2018). With further study of explicit models we may be able to understand fine-grained activity patterns such as regional differences and correlations in activity.…”

Section: Discussionmentioning

confidence: 99%

“…Though it has not been well recognized, successful shortest path routing requires knowledge of current network traffic: a short path is not necessarily short if there is congestion. Even when short path models include small buffers, they show substantial message loss due to collisions (Graham and Hao, 2018). In any case, shortest path routing is considered implausible because it requires global knowledge of network architecture to select shortest paths for all signals (Seguin et al, 2018;but see Mišić et al, 2015).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy

Graham

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Collision dynamics in brain network communication have been little studied. We describe a novel interaction that shows how nonlinear collision rules can result in efficient activity dynamics on simulated mammal brain networks. We tested the effects of collisions in "information spreading" (IS) models in comparison to standard random walk (RW) models. Simulations employ synchronous agents on tracerbased mesoscale mammal connectomes at a range of signal loads. We find that RW routing models have high average activity, which increases substantially with load. Activity in RW models is densely distributed over nodes: a substantial fraction are highly active in a given time window, and this fraction increases with load. Surprisingly, while IS routing models make many more attempts to pass signals, they show lower net activity due to collisions compared to RW. Activity in IS increases relatively little over a wide range of loads. In addition, IS models have greater sparseness (which decreases slowly with load) compared to RW models. Results hold on two networks of the monkey cortex and one of the mouse wholebrain. We also find evidence that activity is lower and more sparse for empirical networks compared to degree-matched randomized networks in IS, suggesting network topology supports IS routing.A given topology can support a range of routing strategies, giving rise to different observable dynamics (see, e.g., Friston, 2011). To begin to understand what routing strategy is in use in the brain, one can look to design considerations in engineered systems (Graham and Rockmore, 2010;Graham 2014;Navlakha et al., 2015;Fornito et al., 2016). A fundamental challenge for any large-scale communication system is the management of collisions. Collisions are the price paid for the ability to route signals selectively and dynamically to many possible destinations. Collision dynamics are emergent: they are a nonlinear effect of node dynamics, topology, and current traffic. All large-scale engineered communication systems have a means of managing collisions, through redress, arbitration, and other strategies (see e.g., Kleinrock, 1976;Mišić and Mišić, 2014).

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy

Graham

2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…But this is generally possible only when collisions and other nonlinear interactions are ignored. As a result, explicit models capable of accounting for collisions have been relatively little studied to date (see Graham & Hao, 2018). With further study of explicit models we may be able to understand fine-grained activity patterns such as regional differences and correlations in activity.…”

Section: Emergent Sparseness On the Mammal Connectomementioning

confidence: 99%

“…Though it has not been well recognized, successful shortest path routing requires knowledge of current network traffic: a short path is not necessarily short if there is congestion. Even when short path models include small buffers, they show substantial message loss due to collisions (Graham & Hao, 2018). In any case, shortest path routing is considered implausible because it requires global knowledge of network architecture to select shortest paths for all signals (Seguin, Razi, & Zalesky, 2018, but see Mišić et al, 2015).…”

Section: Introductionmentioning

confidence: 99%

Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy

Graham

2020

Network Neuroscience

Self Cite

View full text Add to dashboard Cite

Signal interactions in brain network communication have been little studied. We describe how nonlinear collision rules on simulated mammal brain networks can result in sparse activity dynamics characteristic of mammalian neural systems. We tested the effects of collisions in “information spreading” (IS) routing models and in standard random walk (RW) routing models. Simulations employed synchronous agents on tracer-based mesoscale mammal connectomes at a range of signal loads. We find that RW models have high average activity that increases with load. Activity in RW models is also densely distributed over nodes: a substantial fraction is highly active in a given time window, and this fraction increases with load. Surprisingly, while IS models make many more attempts to pass signals, they show lower net activity due to collisions compared to RW, and activity in IS increases little as function of load. Activity in IS also shows greater sparseness than RW, and sparseness decreases slowly with load. Results hold on two networks of the monkey cortex and one of the mouse whole-brain. We also find evidence that activity is lower and more sparse for empirical networks compared to degree-matched randomized networks under IS, suggesting that brain network topology supports IS-like routing strategies.

show abstract

A selective diffusion model of brain network activity

Cited by 2 publications

References 24 publications

Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy

Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy

Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy

Contact Info

Product

Resources

About