Understanding how the brain can store and process information is one of the biggest challenges we are currently facing in the field of neuroscience. It is possible to study the brain as a whole. However, such approaches tend to only recover neural activity from a sparse subset. To counteract this problem, we are using a bottom-up neuroscience paradigm, where neurons are cultured on top of a multi-electrode array (MEA). The MEA can be used to stimulate and record neural activity. To further reduce the complexity, we culture neurons inside of a polydimethylsiloxane (PDMS) microstructure, which ensures that the somas are confined to predefined locations (nodes) while at the same time guiding the growth of the neurites by physically constraining them. With such PDMS microstructures it is possible to create small circular neural networks consisting of 4 nodes, where the axons of one node predominately connect with other nodes in a clock-wise fashion. Hence, we can put constraints on the topology of such networks. It is possible to apply complex stimulation patterns to such a circular network through the MEA that consist of multiple electrical pulses. In this work we investigated the effect of 125 unique stimulation patterns on the network response. We found out that the network response to each stimulation pattern stays constant for multiple hours. Furthermore, the responses are unique for different stimulation patterns. Yet, similar stimuli elicit similar responses. We believe that given the relative ease of creating such circular networks and their corresponding spiking responses to a wide set of different stimulation patterns, the here proposed platform is a promising candidate for validating both simulated neuronal networks and tools that can reconstruct the topology of a network such as generalized linear models. In the future, we will be working on achieving both of these goals.