Variability in Self-Organizing Networks of Neurons: Between Chance and Design

Activity-dependent structural plasticity plays a pivotal role in the self-organization of neuronal networks. Emerging activity continuously feeds back into the growth process, regulating connectivity and activity homeostatically towards an optimal functional dynamic range. Transferred into cell culture, neurons follow the same homeostatic principles. They migrate and regrow neurites to establish synaptic connectivity and spontaneous network activity. Despite of homeostatic regulation, the resulting network structure and activity dynamics exhibit considerable variability across and within studies, even under seemingly constant experimental conditions. This increases the experimental effort to statistically validate findings and limits the standardization of neuronal cultures as a model system for neuronal development and plasticity. From another perspective, however, the variability across networks could be interpreted as a rich repertoire of self-organized functional configurations, resulting from systematic neuronal adaptation to different micro-environmental conditions. Predictions from activity-dependent growth models that simulate cell migration and neurite outgrowth allow to better understand the self-organization process in vitro and to interpret the impact of experimental variations. We argue that the influence of experimental conditions converges on two main parameters that determine the topology and thereby the functional properties of developing networks: neuronal motility and neuron density. Their interaction produces a wide range of different network architectures with varying degrees of modularity and synchronization [1, 2] and is described by a simple but biologically plausible growth model [3]. This perspective creates an opportunity to re-analyze and interpret existing results, exploit the richness of network properties, and facilitates more targeted experimental design.