A complete simulation of the selection process can be constructed using a population of self-replicating finite-state automata. The entire population is challenged with a repeating sequence of inputs, and those individuals that are best able to recognize the input sequence are allowed to replicate most rapidly. Replication proceeds with imperfect fidelity, so that under the constraint of constant total population size, a quasispecies distribution of error copies is obtained. The operation of this simulation provides an essential representation of an evolving system. When the input sequence is altered, the structure of the existing population is destabilized, and a new quasispecies distribution emerges. The ability of the system to respond to changes in the input and the structure of the quasispecies distribution are shown to be critically dependent on the fidelity of replication.