The combined effect of homeostatic structural and inhibitory synaptic plasticity during the repair of balanced networks following deafferentation
Although a number of previous experimental and theoretical studies have investigated network reorganisation following deafferentation down to the level of synaptic elements , the mechanisms that are involved in this process are still not completely understood. We examined the dynamics of the repair mechanism by incorporating activity dependent homeostatic structural plasticity  into a spiking neural network model balanced by inhibitory synaptic plasticity . Results from our simulations suggest that the process of reconfiguration of lateral connectivity following sensory deprivation is extremely sensitive to the balance of excitation and inhibition (E-I) in the network. We find that while fast homeostatic inhibitory synaptic plasticity is able to re-establish the E-I balance in neurons outside the lesion projection zone (LPZ), it prevents them from transferring excitatory activity to the deprived neurons in the LPZ. On the other hand, uncontrolled disinhibition by suppression of homeostatic inhibitory synaptic plasticity initially allows deprived neurons to regain activity but fails to stabilise the network back to a functional balanced state. These observations are in accordance with findings that indicate that inhibition plays a critical role in network rewiring  seemingly by stimulating structural plasticity mechanisms seen during development . The sprouting of inhibitory axons outwards from the LPZ, opposite to excitatory axons has also been observed, possibly to re-inhibit neurons outside the LPZ . Therefore, we hypothesise that the ratio of excitation and inhibition must follow a specific trajectory in the different regions of the network to enable successful repair as has been observed in various studies. The model of structural plasticity implements the dynamics of synaptic elements as dependent on intrinsic properties of individual neurons only . The configuration of the network, by the formation and removal of synapses therefore depends solely on the numbers of various synaptic elements. Our current work extends this model by considering other factors that affect network rewiring, such as the activity dependent stability of synapses, and inhibition gradient guided axonal sprouting , to build a more faithful simulation of the underlying dynamics. This will enable us to study the effects of network reorganisation after deprivation on its computational functions, such as associative memory.