The Cognitive Network Controller (CNC) defines a neuromorphic architecture where a spiking neural network can both encode network performance observations and select the optimal actions (e.g., routes) for the context of those observations. Because of these features, the CNC can quickly adapt to changes in the operational environment to either maintain or improve selected performance metrics. This behavior can be attractive for a space networking scenario with orbiting and ground-based assets that are either stationary or manned, bringing an elevated level of autonomy in network communication decisions. Using the SCaN testbed as a laboratory facility in orbit, we evaluated the adaptation abilities of the CNC applied to a space network routing application. Towards this end, the CNC design and the related neuromorphic processor were implemented in software and deployed on the flight computer of the SCaN testbed, and then applied to route bundles to a ground station over parallel links. This work likely constitutes the earliest demonstration of a space application for neuromorphic computing and a basic validation of the online adaptation capabilities of the CNC.