In this thesis, spiking neural networks are studied in two contexts -- one in a functional role where the network is trained with reinforcement learning, and second in a descriptive context studying the emergent self-organised criticality (SOC) property of the network. Recent advances in neural networks incorporating reinforcement learning, especially deep reinforcement learning, have set new expectations for the future of this field. However, these new advances have not been implemented yet with spiking neural networks for high dimensional tasks. In this thesis, we try to achieve this level of performance on such a task. Specifically, we train a spiking neural network to play Pong - a simple video game - with reinforcement learning. After training, the spiking neural network is able to play Pong at a basic level and score against its opponent occasionally. In the second part, we study and implement the setup from the paper "Synaptic Plasticity Enables Adaptive Self-Tuning Critical Networks". Particularly, the goal is to reproduce the result that a spiking neural network consisting of integrate and fire neurons and synapses with spiking timing-dependent plasticity and short-term plasticity dynamics exhibits self-organising criticality, such that the network operates in the critical regime after the self-organisation. Criticality is characterised by the scaling factor of the distribution of the size of neuronal avalanches in the network, the distribution's scaling factor of the Detrended Fluctuation Analysis (DFA) and the branching ratio.