This title appears in the Scientific Report :
2016
From single-cell spiking to large-scale interactions: A multi-scale spiking network model of macaque cortex
From single-cell spiking to large-scale interactions: A multi-scale spiking network model of macaque cortex
Even in the absence of a particular sensory stimulus or behavioral task, mammalian cortex features complex dynamical behavior on multiple scales. On the single-cell level, neurons spike irregularly with increasing intrinsic time scales along the visual hierarchy [3]. Cortical layers and cell types s...
Saved in:
Personal Name(s): | Schmidt, Maximilian (Corresponding author) |
---|---|
Bakker, Rembrandt / Shen, Kelly / Bezgin, Gleb / Hilgetag, Claus-Christian / Diesmann, Markus / van Albada, Sacha | |
Contributing Institute: |
Computational and Systems Neuroscience; INM-6 Computational and Systems Neuroscience; IAS-6 |
Imprint: |
2016
|
Conference: | Bernstein Conference 2016, Berlin (Germany), 2016-09-21 - 2016-09-23 |
Document Type: |
Poster |
Research Program: |
Brain-inspired multiscale computation in neuromorphic hybrid systems Supercomputing and Modelling for the Human Brain Brain-Scale Simulations Connectivity and Activity Theory, modelling and simulation The Human Brain Project |
Publikationsportal JuSER |
Even in the absence of a particular sensory stimulus or behavioral task, mammalian cortex features complex dynamical behavior on multiple scales. On the single-cell level, neurons spike irregularly with increasing intrinsic time scales along the visual hierarchy [3]. Cortical layers and cell types show heterogeneous average firing rates. On a global scale, cortical areas form clusters with covarying activity, so-called resting-state networks [5]. It is an open question how these multi-scale activity patterns emerge from the complex structure of cortex, with connectivity that is specific to cell types, layers, and areas.To address this question, we present a spiking multi-scale model of macaque visual cortex connected according to a new connectivity map based on multiple sources. It thus makes the nontrivial step from a population description [Chaudhuri et al.] to a spiking description of this system. Visual cortex serves as an example system on which we develop the necessary tools to simulate spiking networks of this scale with realistic population-level connectivity, enabling analogous models of further cortical regions in future.The network connectivity combines data from numerous tracing studies, collected in a new release of the CoCoMac database, with a recent quantitative dataset [6] and is refined using dynamical constraints [2]. The model for each area is an adaptation of a microcircuit model of early sensory cortex [1] to the specific laminar structure of the given area.Simulations on the JUQUEEN supercomputer show that plausible dynamics emerge from the network, mimicking experimental findings on cortical dynamics on multiple scales, from single-cell spiking properties [3] to the large-scale interaction patterns of cortical areas [5]. Furthermore, cortico-cortical interactions propagate predominantly in the feedback direction, akin to experimental observations during sleep or visual imagery [4]. Our study demonstrates that such multi-scale dynamics can emerge from the connectivity on the level of areas and populations, thereby separating network effects from the influence of diverse intrinsic dynamics of single cells. |