One feature of legislative bargaining in naturally occurring settings is that the distribution of seats or voting weights often does not accurately reflect bargaining power. Game-theoretic predictions about payoffs and coalition formation are insensitive to nominal differences in vote distributions and instead only depend on pivotality. We conduct an experimental test of the classical Baron-Ferejohn model with five-player groups. Holding real power constant, we compare treatments with differences in nominal power. We find that initial effects of nominal differences become small or disappear with experience. Our results also point to the complexity of the environment as having a negative impact on the speed at which this transition takes place. Finally, and of particular importance as a methodological observation, giving subjects a pause accelerates learning. © 2019 Elsevier Inc.