In this paper the dynamic optimal toll design problem as a game of the Stackelberg type is investigated, with the road authority as leader and drivers on the road network as followers. The road authority sets dynamic traffic-flow dependent tolls on some links in order to minimize its objective function, while the drivers choose their routes and departure times so as to minimize their own perceived travel costs, which include a travel time component, tolls, and penalties for deviation from their preferred arrival and departure times. The drivers' behavior is modeled using a dynamic user equilibrium model. We define the problem in a general form so that a wide class of objective functions for the road authority and a wide class of the dynamic traffic equilibria can be employed. We also discuss the problem properties. The problem to find optimal traffic-flow dependent tolls in a general setting introduced in this paper is NP-hard. One of the ways of tackling such problems is to use advanced heuristic methods. In this paper a neurosimulation-based approach is proposed, using the neurosimulator FAUN 1.0. The proposed solution method is illustrated on case studies with the so-called Chen network.