The acquisition of a new motor skill is computationally demanding for our sensorimotor system. In a previous dual-task study, we found that the interference peaked early during acquisition - at different times for different secondary tasks (ST), which we interpreted as competition for different resources - and then converged towards a non-zero plateau (Eversheim and Bock 2000; SFN, Abst. 1990). The terminal plateau could reflect either a resource demand common to all ST, or structural interference between tasks. The present study was initiated to resolve this issue by manipulating structural interference in various ways. As in the previous study, subjects tracked with their index finger a visual target moving in a horizontal plane. To induce skill acquisition, visual feedback about finger position was up-down reversed. During 50% of tracking time, subjects performed with their other hand one of four different choice reaction time tasks (RTT): a) a manual four choice RTT; b) a manual two choice RTT; c) a ma...