Mapping the substrate for brain stimulation reward by means of current-number trade-off functions