To date, our understanding of how the human brain retrieves, processes, and stores information remains limited. A specific challenge lies in unravelling how local plasticity mechanisms shape the transfer of information between neurons. Traditional approaches to studying local plasticity in vitro, such as patch clamp or calcium imaging, are often limited by low throughput or the ability to record only from a subset of neurons within intricate networks. In response to these challenges, we adopt a bottom-up approach to engineer low-density neuronal networks with defined architectures. Human induced pluripotent stem cell (hiPSC)-derived neurons are cultured within custom-designed polydimethylsiloxane (PDMS) microstructures positioned on microelectrode arrays (MEAs). The PDMS microstructures allow us to restrict the connectivity in the networks, resulting in reduced complexity, while the MEAs enable the recording and modulation of spiking activity with high temporal resolution throughout the entire network. Our methodology achieves spatial and functional isolation of networks consisting of as little as 1-3 neurons, while maintaining consistent viability and spiking behaviour. We show that we can modulate the network activity using stimulation protocols that could potentially evoke changes in the information transfer. Consequently, this platform could facilitate the investigation of plasticity rules in isolated neuronal networks with reduced complexity, ensuring high reproducibility and throughput. Looking ahead, our objective is to include dopaminergic neurons in distinct compartments of the microstructures, and to vary the network architectures. These investigations may provide insights into the impact of neuromodulators such as dopamine on plasticity and the necessary levels of complexity for learning.