During idea generation, designers might inadvertently restrict their exploration of the solution space by adhering either to their own initial ideas or to those developed by others. To investigate which of these sources of ideas is more fixating, we conducted an experimental study with Engineering students engaged in a series of ten computer-based structural design tasks. Participants were divided into two conditions, either developing their own solutions to the first five tasks or watching someone else do so. In the subsequent five tasks, those in the first condition more often adhered to the design approaches suited to the initial five tasks. This resulted in fewer design variations, greater costs and different subjective experiences of the tasks.