Connectomics has revolutionized our understanding of brain function by emphasizing the importance of neural networks and their topographical organization. Corticostriatal circuits, which play a critical role in cognition and emotion, follow a precise topographic architecture essential for integrating and processing cortical information within the basal ganglia. Disruptions to this connectivity are often implicated in neurodevelopmental and psychiatric disorders such as obsessive compulsive disorders, schizophrenia, epilepsy, and autism spectrum disorders. However, studying network disruptions in vivo presents significant challenges due to their intricate architecture and early developmental onset. To address this, we employed a brain-on-chip microfluidic platform to recreate a biologically relevant model of topographically organized corticostriatal networks. By mimicking the directional control of neuronal projections using Tesla valve-inspired microchannels, we demonstrate that genetic perturbations affecting neuronal excitability during development lead to selective alterations of local versus long-range network topology, resulting in the formation of new convergent nodes. This model offers critical insights into how early perturbations contribute to circuit-specific pathologies, providing a valuable tool for understanding neurodevelopmental disorders and advancing therapeutic strategies.