Universal principles of topology governing both of structural and effective connectivity
Masanori Shimono (University of Tokyo), John Beggs (Indiana University)
Since the era of Hebb, the importance and mysterious role that neuronal ensembles play in has been a main concern of the neuroscience [Hebb, 1949]. Recently, much work using structural connectivity has revealed patterns of synaptic connections in neuron ensembles [Bock et al., 2011]. Structural connectivity information is extremely valuable, as it indicates pathways through which one neuron could possibly influence spiking in another. In contrast, effective connectivity aims to describe the pathways through which influence actually occurs. The concept of “effective connectivity” was initially described in regard to local neuronal networks [Aertsen et al., 1989]. However, almost all research on “effective connectivity” has been done in macroscopic dynamics recorded using fMRI, MEG, and EEG [Friston, 1994]. Furthermore, even out of the studies on microcircuits, almost no work has been done on effective connectivity in local cortical networks at the timescale of typical synaptic delays within the cortex (1-20 ms). This is unfortunate, as direct influence between neurons would be expected to occur at these time delays. Structural connectivity studies have shown that groups of 3-7 cortical neurons are more likely than chance to be synaptically connected to each other if they have synapses onto a common neighbor neuron [Perin et al., 2011]. This led to the question whether effective connectivity also shows this pattern.
In order to investigate these topics, we used a 512 electrode array system to record spontaneous activity in 9 slice cultures that included neocortex and portions of hippocampus. On average, we recorded over ~120 neurons from each culture for 1 hr or more. Although many metrics of effective connectivity have been proposed, we selected transfer entropy because several studies found it to compare favorably in accuracy to other metrics. In the comparison between the topological properties of structural neuronal networks and the topological properties of the reconstructed effective connectivities, we could find universal principles of topology governing both of structural and effective connectivity.
Preferred presentation format:
Large scale modeling