% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@PHDTHESIS{Albers:1011057,
author = {Albers, Jasper},
othercontributors = {Diesmann, Markus and Offenhäusser, Andreas},
title = {{S}patial connectivity patterns shape dynamics of
large-scale spiking neural networks of visual cortex},
school = {RWTH Aachen University},
type = {Dissertation},
address = {Aachen},
publisher = {RWTH Aachen University},
reportid = {RWTH-2025-04522},
pages = {1 Online-Ressource : Illustrationen},
year = {2025},
note = {Veröffentlicht auf dem Publikationsserver der RWTH Aachen
University; Dissertation, RWTH Aachen University, 2025},
abstract = {Unraveling the inner workings of the brain has been one of
the most fascinating research topics in human society for
centuries. Evolving from the early discoveries of neurons by
Golgi and Cajal, the field of neuroscience has made
significant progress in examining the anatomy, physiology
and functionality of the human brain. A driving factor for
this research is the explosion in the availability of
detailed anatomical and physiological data from experiments
over the past decades, as well as advances in the
availability of computational resources. Combined, these
changes enable ever more precise studies of both brain
connectivity and function conducted in the field of
Computational Neuroscience. This thesis aims to further our
understanding of the link between structure and dynamics in
the brain, which fundamentally underlies its
structure-function relationship. Specifically, a focus is
set on studying the impact of spatial connectivity, i.e.,
the distance-dependent wiring principles of neurons, on
cortical activity of the visual regions. Choosing a
computational approach, detailed anatomical knowledge is
integrated into biologically constrained models of cortical
circuits, allowing for an investigation of the relationship
of structure and dynamics in silico. First, the
technological foundations are established that enable
efficient simulations of large-scale neural networks at
realistic neural and synaptic density. An open-source
software platform is conceptualized and implemented that
facilitates the automated execution and analysis of
performance benchmarks of neural network simulators, which
are the backbone of simulations in modern Computational
Neuroscience. Thereby, development is aided that optimizes
the time-to-solution in such simulations, ultimately
allowing to integrate ever more anatomical detail into
cortical network models. Second, this thesis consults a
broad range of experimental data to construct a
comprehensive spatial neural network model of the primary
visual cortex (V1) of macaque monkey. Simulations reveal
that the model of V1 exhibits strongly pathological
activity: it is unable to exist in a balanced dynamical
state in which the statistics of the activity resemble
experimentally observed states. Revisiting the underlying
connectivity and linking it to the dynamics of a model of
local cortex exposes that the inherent recurrent targeting
patterns stemming from low-resolution data based on light
microscopy (LM) necessarily lead to diverging activity. In
contrast, the implausible dynamics are remedied by replacing
the LM data set by a recent high-resolution data set based
on electron microscopy (EM). Substituting the LM
connectivity by EM connectivity in the model of V1 resolves
the unbalanced activity, yielding a model that exhibits
biologically plausible activity without further fine-tuning.
Thus, the derived model can act as a platform for future
research to study the dynamical and functional implications
of spatial connectivity, shedding light on optimal wiring
strategies employed by the visual cortex in the brain.
Third, a novel predictive connectivity rule is developed
that explicitly takes the spatial distributions of synapses
relative to the cell bodies into account. To compare the
predicted connectivity to an experimental ground truth, a
general method for comparing the recurrent connection
structure of networks is devised based on matrix
decomposition. A subsequent analysis demonstrates that the
precise spatial connectivity rule outperforms a classical
approach based on the distance between neurons. Thereby, it
paves the way for incorporating the intricacies of local
spatial connectivity into the next generation of modeling.
Overall, this work provides a thorough account on spatial
connectivity in cortical neural networks. It thus
contributes to uncovering the structure-dynamics-function
triad of the brain, ultimately leading to a deeper
understanding of its working mechanisms.},
cin = {535500-2 ; 934910 / 130000},
ddc = {530},
cid = {$I:(DE-82)535500-2_20140620$ / $I:(DE-82)130000_20140620$},
pnm = {ACA - Advanced Computing Architectures (SO-092) / HBP SGA3
- Human Brain Project Specific Grant Agreement 3 (945539) /
BMBF 03ZU1106CB - NeuroSys: Algorithm-Hardware Co-Design
(Projekt C) - B (BMBF-03ZU1106CB) / Impuls- und
Vernetzungsfonds (IVF-20140101) / NeuroSys:
Algorithm-Hardware Co-Design (Projekt C) - B
(BMBF-03ZU2106CB)},
pid = {G:(DE-HGF)SO-092 / G:(EU-Grant)945539 /
G:(DE-Juel1)BMBF-03ZU1106CB / G:(DE-HGF)IVF-20140101 /
G:(DE-Juel1)BMBF-03ZU2106CB},
typ = {PUB:(DE-HGF)11},
doi = {10.18154/RWTH-2025-04522},
url = {https://publications.rwth-aachen.de/record/1011057},
}
guest ::