High frequency neurons determine effective connectivity in neural networks
Jan. 15, 2018
Understanding how information is routed and processed
in the brain is essential to understand its operation.
However, this is not an easy task: the brain is a extremely
complex network containing billions of neurons, each with its
own connections. In addition, the information channels are not
static; the connections between neurons are dynamic and while
some can be reinforced others can be weakened resulting in a
changing functional connectivity.
An international study, with the participation of a researcher
from the Institute of Interdisciplinary Physics and Complex
Systems (IFISC-UIB / CSIC) has proposed a computational model
of interacting neurons to analyze how the functional
connectivity can be manipulated.
The authors simulated an interconnected network of 11 neuronal
populations (11 nodes), each of which was composed by 80
excitatory and 20 inhibitory neurons. The system was initially
homogeneous, i.e., all populations oscillated with the same
average frequency. The connections between nodes were assumed
symmetric and bidirectional, so the information could be
transmitted from A to B as well as from B to A.
When this homogeneity was broken by increasing the oscillation
frequency of one of the nodes (namely creating a
high-frequency node, HFN), the response of the network
drastically changed. It was observed that a small perturbation
superimposed on the HFN propagated over the entire network and
could be detected at all other nodes. In contrast, this
propagation did not happen if the perturbation was
superimposed to any other node oscillating at the average
frequency. The situation got even more interesting when
superimposing a periodic (or aperiodic) signal to one of
the nodes of the network. It was found that the node that
oscillated at the high frequency influenced the direction of
propagation of the signal, changing the effective connectivity
of the entire network. Moreover, if two different signals were
applied simultaneously, one at the HFN and the other at any
other node of the network, the one applied at the HFN
propagated through the whole network while the other signal
could only propagate away from the position of the HFN.
It turns out that the global dynamics of the neural network can be strongly influenced by the internal dynamics of one of the nodes, without any structural change of the network. This finding has consequences for the total effective connectivity of the network: for example, the HFN node is able to transmit information to other distant nodes even without direct connections. In addition, it can determine a direction of transmission of information even in the presence of bidirectional and symmetric connections. The "director" node can govern the functional connectivity of the entire network only by an increase its oscillating frequency. The work concludes that any node of the network can play the role of "director" if its oscillation frequency is increased, even if it exhibits lower connectivity than others. Interestingly, similar results were obtained in a network that mimics a real cortical area.
The published study is an approach to understand how the brain’s information routing can change, and how the local dynamics of a set of neurons can affect the overall dynamics of the total network. The researchers speculate that the results might be generalized to more complex networks.
Pariz, A.; Esfahani Z. G.; Parsi, S. S.; Valizadeh, A.; Canals, S.; Mirasso, C. R. High frequency neurons determine effective connectivity in neuronal networks. Neuroimage, 2017. DOI: 10.1016/j.neuroimage.2017.11.014
