Brain Functioning – PGO en Colleges
College 1- Brain networks
Graph analysis
Node = a neuron. Line = the
connection between the neurons. The
degree is the number of edges
connected to each node. Only start
and end can have an odd degree, the
rest should be even.
Graph: squares = edges and axis =
nodes.
Many studies did not replicate
linking brain and behaviours. But now there are more datasets available, more and more studies are
beginning to find some correlations.
1
, PGO 1 – A matter of approach
Leerdoel 1: How has the neuropsychological research changed over time?
Leerdoel 2: What are the key objectives and methodologies of neuropsychology?
Leerdoel 3: What is network science and how can it be used to study brain functioning?
Leerdoel 4: How are brain networks relevant for understanding brain disorders?
https://maastrichtuniversity.zoom.us/j/93496264374?pwd=cVF6VGFaYUwvWlF2cmMrcHhPbDA5Zz0
9 → Zoom meeting.
https://drive.google.com/drive/folders/1jQheXMj6SVWJ3DweINooI5LAf2u2s5cQ → Dropbox
https://drive.google.com/drive/folders/1Pe-F0bkK1Y_8qSONdcQhadybGMperIZo → Literatuur
beide vakken.
Artikel: Luria – Neuropsychology in the local diagnosis of brain damage.
Part 1
The study of the function of separate parts of the brain began with observing speech pathology (o.a.
Broca – producing- and Wernicke – understanding -). After decades of research, local significance
was attributed to very general changes such as language, mental activity and character changes.
This confidence in the clearly local significance of the disturbance of higher mental functions met,
however, with a series of substantial objections: disturbances of higher orders of the brain are very
unlikely to be given a specific area in the brain. It is more a regression of functional organization.
Local diagnosis began to appear unfounded.
Two kinds of investigators:
1. Narrow localisation: believed that elementary and higher functions are an immediate
function of the narrowly limited parts and spoke of ‘zones’ several mental processes are
organised. Loss of these functions would be a symptom of damage in a corresponding zone.
2. Anti-localisation: believed that elementary functions are related to narrowly limited parts of
the brain, they related higher (mental) functions to the brain as a whole, thereby tying them
in a direct fashion with the "brain mass" and considered that the disturbance of these
functions is an unequivocal symptom of the massiveness or volume of brain damage.
Elementary and higher mental processes are both part of hierarchically structured centres.
Therefore, disturbances of these functions can be accounted for by the destruction of different links
in this hierarchical system, and are symptoms of multiple significance (= the local significance of
which may be made more exact only as a result of special neurological analysis.).
Only special analyses of the structure of these functions and those physiological mechanisms behind
them, helps us to understand the complex structure and specify the factors.
2
,The hierarchical order: when one link is damaged, the higher mental function may suffer, even when
the centres differ in localisation. However, when one links is lost, the whole system is different and
the symptoms of disturbance of another mental function will have a complete other structure,
depending on the location of the damage.
In this way we come again to the position that there is a multiple significance of symptoms, but this
position does not in any way deprive these symptoms of their local value. In order to provide a
correct evaluation of a symptom and its local significance it is only necessary to carry out a
qualitative analysis of the structure of the symptom.
Part 2
Example of the fact that some psychological function does not belong to one part of the brain:
writing. The naïve idea existed of a zone in the brain for writing movements. But lesions among all
parts of the brain can result in writing disturbances. Multiple significance however does not mean we
cannot use it for local diagnostics → The essence is that disturbances of writing have very different structure
in different localisation of cortical defect, and a careful neuropsychological analysis can single out different
factors underlying different forms of writing defects; such qualification of writing defects makes it possible to
use disturbances of writing as a mean for topical diagnostics of brain injury.
Writing example: Showed that it is impossible to relate a function to some limited part of the brain
or to conclude that a disturbance directly indicates a specific lesion. Mental functions are complex
dynamic structures of cortical zones working together. Each of these zones contributes its own
factor to the making of a functional system. From this it should be clear that local damage of the
brain cortex should not be related to a symptom (which might be of multiple significance and might
have been evoked by damage in various locations) but to a factor which leads to the origination of a
symptoms.
Symptoms of disturbance of any higher cortical function may be used for local diagnosis of brain
damage, but such a diagnosis can be carried out only when there is qualitative analysis or evaluation
of the symptom. Such evaluation of the symptom is the first task of neuropsychology.
Part 3
A second task of neuropsychology is making a comparison of different symptoms and discovering a
general factor underlying them. The hypothesis is that when there is a local lesion which causes the
loss of some factor, all the functional systems which have this factor suffer and at the same time all
the systems without this factor are preserved → this is the principcal of double dissociation. It helps
us to raise probabilities of a correct local diagnosis of the damage.
Also, syndrome analysis allows us to establish physiological differences between seemingly similar
functions (I) and the physiological similarities between seemingly different functions (II).
Example I: phonemic and musical hearing. Likely to be similar, but are in fact dissociated: you can
have one without the other and vice versa.
Example II: space orientation, numerical schemes and logical-grammatical operations. Unlikely to be
similar, but are in fact associated. There is a general underlying factor for all three: organisation of
elements into simultaneous spatially oriented schemes.
The discovery of different factors which underlie quasi similar functions and of general factors which
unite seemingly different psychological processes is one of the important tasks of neuropsychology.
It is not easy to dynamically localize brain structures. Lesion are often broad and not narrow and
precise. Nevertheless, analysis of the change in higher mental functionals due to local lesions can
enrich the possibility for local diagnosis.
3
,Summary
A precise and early local diagnosis of a circumscribed brain injury is one of the most important
problems of the neurological and neurosurgical clinic. That problem can be solved only on the basis
of an adequate neuropsychological theory and with the application of a series of special
psychological methods. Every psychological function can be regarded as a complicated functional
system which is a result of a constellation of simultaneous and successive partecipations of several
cortical zones. Each of these zones takes part in the realization of this functional system providing a
highly specialised factor which is included in the course of psychological function. That is why lesions
of every cortical zone can result in a disorganization of the functional system, but with different
localisation of the primary lesion the type of the functional disturbances becomes different,
according to the special factor, disturbed by the local lesion. The paper gives several examples of the
multiple significance of symptoms observed by neuropsychological analysis of local brain lesions and
describes methods which can be used for their application for local diagnostics of the injury.
Boek: Fornito – Fundamentals of brain network analysis: Chapter 1.
Current thinking of the brain networks is the concept if connectome. It defines a
matrix representing all possible pairwise anatomical connections between neural
elements of the brain. It is a cellular wiring diagram of the brain. Colored elements
represent a projection from the region listed in the column to the region listed in
the row. The size of the dots in each matrix element is proportional to the
projection distance and darker colors indicate stronger average reported
connectivity strength.
It is not that in the days of Brodmann, Wernicke, etc. this type of evidence did not exist. There has
simply been an eascendancy lately due to two factors:
• Recent years have seen rapid growth in the science of networks in general; not limited to
psychology but in all kinds of areas.
• There has been a technological evolution of methods to visualize the brain organization; e.g.
optogenetics, tract tracing, etc.
Focus is on graph theory → a way of modeling. It estimates and simulates the topology and
dynamics of brain networks. It is a branch of math concerned with understanding systems of
interacting elements. A graph consist of nodes linked by edges.
1.1 GRAPHS AS MODELS FOR COMPLEX SYSTEMS
1.1.1 A brief history of graph theory
Complex systems have properties that are neither completely random nor completely regular,
instead showing nontrivial characteristics that are indicative of a more elaborate, or complex,
organization; these are all around is (molecules, infrastructure, brains, etc.).
The first use of a graph system was a real-world system: Euler and the river banks. The importance
here is that Euler was not into the details of the geography, but rather focused on the links between
system elements.
Another example was by Erdos-Reny who designed a graph in which there are N nodes and a
uniform probability p. If p is close to 1 the graph is densely connected and vice versa for p close to 0.
4
,Both of the above are binary undirected graphs. Binary = edges are either absent or present or,
equivalently, the edge weight is either zero or one. Unidirected = edges connect nodes
symmetrically; no distinction is made between the source and target of a connection
Watts and Strogatz began their analysis with a regular lattice of N nodes and selected randomly
edges connecting nodes i and j. They focused on two things: clustering
coefficients and characteristic path length. The seminal work of Watts
and Strogatz (1998) identified a continuum of network topologies
ranging from completely regular and lattice-like (left) to completely
random (right). Interposed between these extremes is a class of
networks with a so-called small-world topology, which can be generated by randomly rewiring (with
probability, pWS) an arbitrary proportion of edges in a regular network.
Later, Barabasi and Albert introduced anther model that built a complex graph by adding nodes
incrementally. New nodes connect preferentially to existing nodes that already have a large number
of connections and thus represent putative network hubs. By this generative process of preferential
attachment, the “rich get richer’’. A scale-free degree distribution means that the probability of
finding a very high degree hub node in the graph is greater than would be expected if the degree
distribution was unimodal
A third major development in the application of graph theory to real-world systems has been the
discovery that such networks are modular—they can be nearly decomposed into subsets of nodes
that are more densely connected to each other than to nodes in any other modules.
Three concepts are very important for complex network science: small-worldedness, degree
distribution and modularity.
1.1.2. Space, time and topology
Topology is an important aspect of how networks are organised, like rotating, skewed, etc. But other
aspects such as space and time matter too. For brain networks, the three-plus-one dimensions of
space and time must be incorporated with the fifth dimension of topology. Another important aspect
to take into accounts are the cost → drives the physical location and connectivity of nodes, such that
spatially proximal nodes have a higher probability of connectivity than spatially distant nodes.
Complex networks are constant trade-offs of the costs minimalization and topological value.
1.2 GRAPH THEORY AND THE BRAIN
Each row or column representing a different brain region in the matrix is drawn as a node in the
graph, and the values in each matrix element are drawn as edges
1.2.1. The neuron theory and connectivity at the microscale
Two neuroscientisst: first Golgi with his neural staining method that showed that there was a
continuous syncytial connection between the cell bodies. Second there was Cajal, who showed with
silver impregnation that neurons were discrete cells that contacted each other via synaptic junctions.
Electron microscopy showed later that Cajal’s idea was in favour. It is now accepted that synaptic
junctions are generally points of close contiguity, but not continuity, between connected neurons.
Cajal argued that the brain network organization is driven by the minimization of axonal wiring cost
(conserving material and space) and by minimization of the conduction delay in the information
transmission (conserving time)/also called maximization of integrative topology.
5
, 1.2.2. Clinicopathological correlation and connectivity at the macroscale
Network diagrams were drawn, e.g. by Wernicke, to summarize white matter connections between
cortical area and to explain how symptoms of brain disorders could be related to pathological
lesions. Language is a good example of such a model: language production = frontal and language
comprehensions is temporal. A lesion in the frontal region indeed predicted difficulties with
repeating words. The nineteenth century diagrams of large-scale brain network organization drawn
by these pioneers, comprising a few spatially circumscribed areas (nodes) interconnected by white
matter tracts (edges), set the scene for graph theoretical analysis of nervous systems at the
macroscopic scale.
An important difference between the macroscopic and microscopic anatomists was the quality of
data. The new methods used by Cajal etc made it possible to come up with very high-detailed data,
opposed to macroscopic anatomists. Wernicke’s work was based on several patients and their
symptoms. That is why it took a long time for this type of connectionism to be accepted. As the years
went by, more and more researchers found evidence in light of macroscale connectivity.
1.2.3. The dawn of connectomics
The first brain graphs were a number of cortical and subcortical nodes interconnected by axonal
edges in the macaque monkey and cats using injections. The next step was not to study mammals,
but C. elegans. The nervous system of this animal is quite limited so easy to map. They showed that
this network had a short characteristic path length and high clustering; in other words, its global
organization conformed to a small-world topology.
These and other early graph theoretical studies of brain networks demonstrated proof of concept—
they showed that the mathematical tools of graph theory were applicable to suitably simplified
nervous systems. Only around 2005 they started to graph theoretic techniques to human
neuroimaging data.
1.2.4. Neuroimaging and human connectomics
Functional connectivity = a statistical dependence between the time series of measures
neurophysiological signals. The basic concept is that two locations can be functionally connected if
they have coherent or synchronized dynamics. Resting state networks: around 10 spatially
distributed large-scale neural systems that show coherent low-frequency oscillations.
Networks around the brain are small-worlded, contain hubs, have a hierarchical structure and
minimized wiring costs. The first graph theoretic analyses of human brains have been based on MRI
and structural covariance. There are two types of connectivity:
• Structural/anatomical → axonal projection from one cell to another. Such axonal projections
are expected to change only slowly over time.
• Functional → a statistical measure of synchronized activity that does not necessarily imply
an underlying anatomical connection, and which can change rapidly over time.
• Link → they are correlated but do not have to be identical. As functional connectivity is
averaged over longer time periods, it may converge onto structural connectivity, although
they are still different measures and may thus yield connectomes with different values of
some topological parameters.
6
