|Ph.D Student||Levy Ofri|
|Subject||Functionality in Modular Neural Networks|
|Department||Department of Medicine||Supervisor||PROF. Shimon Marom|
|Full Thesis text|
Brain architecture is inherently modular, being composed of sparsely connected local networks that are embedded in networks of networks across multiple scales. Modularity is believed to be essential for the enhancement of the networks functional capacities, by enabling a trade-off between functional differentiation - the capacity of each module to act independently, and integration - by which enhancement of functionality is achieved by lumping the different modules together.
In this work we have used a simple modular in-vitro design to study impacts of modular organization on the network functional capacities, focusing on the function of representation - a key function that neural networks are believed to realize. We cultured networks of cortical neurons obtained from newborn rats in-vitro on substrate-integrated multi-electrode arrays. We enforced the network to develop two well-defined modules of neural populations that are coupled by a narrow canal which allows only sparse connectivity between the modules, as demonstrated by fluorescent labeling. We monitored and characterized the neural activity in the modular design, and examined the capacity of each module to individually classify (i.e. represent) spatially distinct electrical stimuli and propagate input-specific activity features to their downstream coupled counterpart.
We show that neural activity dynamics in the studied in-vitro modular design agree with typical dynamics in modular topologies which involve fast intra-modular and slow inter-modular processes. Furthermore, we show that while each of the coupled modules maintains its autonomous functionality, a significant enhancement of classification capacity is observed when the activity evoked in one module is propagated to a downstream coupled module, in which responses to different input sources become more separated ("decorrelated"). We interpret these results in terms of a relative decorrelation effect imposed by weak coupling between modules.