Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy - PubMed
- ️Wed Jan 01 2020
Creative destruction: Sparse activity emerges on the mammal connectome under a simulated communication strategy with collisions and redundancy
Yan Hao et al. Netw Neurosci. 2020.
Abstract
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.
Keywords: Collisions; Communication systems; Connectome; Redundancy; Routing; Sparseness.
© 2020 Massachusetts Institute of Technology.
Conflict of interest statement
Competing Interests: The authors have declared that no competing interests exist.
Figures

Adjacency matrices of the connectomes tested. Mouse includes the ipsilateral whole brain of the mouse (Oh et al., 2014), while monkey1 (Markov et al., 2014) and monkey2 (Bakker et al., 2012) include monkey cortex. All edges (blue dots) are directed and have weight 1. Nonzero (nz) entries are shown in numerical and percentage terms below each matrix.

Illustration of the Treves–Rolls measure of sparseness S for hypothetical histograms of activity data. Shown are histograms that are well approximated by common functions, along with corresponding sparseness values as calculated from Equation 1. Top left: (half) Gaussian, S ≈ 0.64; top right: exponential, S ≈ 0.5; bottom left: generalized Pareto (shape parameter = 0.25), S ≈ 0.33; bottom right: maximal sparseness (closest to zero), S = N−1 , is achieved for a histogram where xn are all zero except for one nonzero entry, indicated here by “High.” A uniform histogram (not shown) has minimal sparseness of S = 1.0 .

(A) Attempted activity (before collisions) and (B) actual activity (net, following collisions) under IS and RW models in the mouse connectome at 10% load. Vertical axis corresponds to nodes and time runs left to right. Colors indicated in the vertical axis in attempted activity plots (A) represent the number of colliding messages at a given node and time step. Net activity per node per time step (B) is binary (black = off, white = active).

Mean of net simulated activity on the mammal connectome as a function of load under IS (solid lines) and RW (dashed lines) models for mouse (circles), monkey1 (squares), and monkey2 (triangles) networks. In IS, activity remains mostly constant across load and is lower compared to the corresponding value for RW at all load levels except 1 message per time step. Activity in RW increases by more than two octaves across loads tested. Shaded areas represent 2 standard deviations of variability from the mean across simulations (note that in some cases this dispersion is smaller than the width of the markers/lines). Data points show significant differences (t test, p < 0.01 ) for corresponding IS and RW measurements at all loads.

Population sparseness of activity as a function of load under IS (solid lines) and RW (dashed lines) models for mouse (circles), monkey1 (squares), and monkey2 (triangles) networks. As with net activity, IS models show relatively constant sparseness of activity across load, with greater sparseness of activity (closer to zero) than RW models at all loads except 1 message per time step. Sparseness in RW decreases by more than a factor of 2. Shaded areas represent 2 standard deviations of variability from the mean of population sparseness across simulations. Data points show significant differences (t test, p < 0.01 ) for corresponding IS and RW measurements at all loads.

(A) Net activity and (B) population sparseness for empirical and randomized networks of the monkey2 and the (thresholded) mouse under IS and RW models. In IS models, both the empirical monkey2 network (triangles) and the (thresholded) mouse network (circles) show lower activity and greater sparseness (showing a difference in percentage terms of between 2–7%) compared to corresponding randomized networks. Randomized RW models show largely the same behavior as the empirical networks (differing by less than 1%) in terms of activity and sparseness. Filled symbols indicate significant differences (t test, p < 0.01) between empirical and corresponding randomized networks, whereas open symbols indicate no significant difference.
Similar articles
-
Effects of packetization on communication dynamics in brain networks.
Fukushima M, Leibnitz K. Fukushima M, et al. Netw Neurosci. 2024 Jul 1;8(2):418-436. doi: 10.1162/netn_a_00360. eCollection 2024. Netw Neurosci. 2024. PMID: 38952819 Free PMC article.
-
Editorial: Network Communication in the Brain.
Graham D, Avena-Koenigsberger A, Mišić B. Graham D, et al. Netw Neurosci. 2020 Nov 1;4(4):976-979. doi: 10.1162/netn_e_00167. eCollection 2020. Netw Neurosci. 2020. PMID: 33195944 Free PMC article.
-
A spectrum of routing strategies for brain networks.
Avena-Koenigsberger A, Yan X, Kolchinsky A, van den Heuvel MP, Hagmann P, Sporns O. Avena-Koenigsberger A, et al. PLoS Comput Biol. 2019 Mar 8;15(3):e1006833. doi: 10.1371/journal.pcbi.1006833. eCollection 2019 Mar. PLoS Comput Biol. 2019. PMID: 30849087 Free PMC article.
-
Seguin C, van den Heuvel MP, Zalesky A. Seguin C, et al. Proc Natl Acad Sci U S A. 2018 Jun 12;115(24):6297-6302. doi: 10.1073/pnas.1801351115. Epub 2018 May 30. Proc Natl Acad Sci U S A. 2018. PMID: 29848631 Free PMC article.
-
Park BY, Vos de Wael R, Paquola C, Larivière S, Benkarim O, Royer J, Tavakol S, Cruces RR, Li Q, Valk SL, Margulies DS, Mišić B, Bzdok D, Smallwood J, Bernhardt BC. Park BY, et al. Neuroimage. 2021 Jan 1;224:117429. doi: 10.1016/j.neuroimage.2020.117429. Epub 2020 Oct 7. Neuroimage. 2021. PMID: 33038538
Cited by
-
Effects of packetization on communication dynamics in brain networks.
Fukushima M, Leibnitz K. Fukushima M, et al. Netw Neurosci. 2024 Jul 1;8(2):418-436. doi: 10.1162/netn_a_00360. eCollection 2024. Netw Neurosci. 2024. PMID: 38952819 Free PMC article.
-
Brain network communication: concepts, models and applications.
Seguin C, Sporns O, Zalesky A. Seguin C, et al. Nat Rev Neurosci. 2023 Sep;24(9):557-574. doi: 10.1038/s41583-023-00718-5. Epub 2023 Jul 12. Nat Rev Neurosci. 2023. PMID: 37438433 Review.
-
Editorial: Network Communication in the Brain.
Graham D, Avena-Koenigsberger A, Mišić B. Graham D, et al. Netw Neurosci. 2020 Nov 1;4(4):976-979. doi: 10.1162/netn_e_00167. eCollection 2020. Netw Neurosci. 2020. PMID: 33195944 Free PMC article.
References
-
- Abdelnour F., Voss H. U., & Raj A. (2014). Network diffusion accurately models the relationship between structural and functional brain connectivity networks. NeuroImage, 90, 335–347. DOI: 10.1016/j.neuroimage.2013.12.039, PMID: 24384152, PMCID: PMC3951650 - DOI - PMC - PubMed
-
- Achard S., Salvador R., Whitcher B., Suckling J., & Bullmore E. D. (2006). A resilient, low-frequency, small-world human brain functional network with highly connected association cortical hubs. Journal of Neuroscience, 26(1), 63–72. DOI: 10.1523/JNEUROSCI.3874-05.2006, PMID: 16399673, PMCID: PMC6674299 - DOI - PMC - PubMed
-
- Aeschbach D., Matthews J. R., Postolache T. T., Jackson M. A., Giesen H. A., & Wehr T. A. (1997). Dynamics of the human EEG during prolonged wakefulness: evidence for frequency- specific circadian and homeostatic influences. Neuroscience Letters, 239(2–3), 121–124. DOI: 10.1016/S0304-3940(97)00904-X - DOI - PubMed
-
- Avena-Koenigsberger A., Mišić B., Hawkins R. X., Griffa A., Hagmann P., Goñi J., & Sporns O. (2017). Path ensembles and a tradeoff between communication efficiency and resilience in the human connectome. Brain Structure and Function, 222(1), 603–618. DOI: 10.1007/s00429-016-1238-5, PMID: 27334341 - DOI - PubMed
LinkOut - more resources
Full Text Sources