In this paper, we describe a novel approach to the study of problem solving involving the detailed analysis of natural scanning eye movements during the “one-touch” Tower-of-London (TOL) task. We showed subjects a series of pictures depicting two arrangements of colored balls in pockets within the uper and lower halves of a computer display. The task was to plan (but not to execute)the shortest movement sequence required to rearrange the balls in one half of the display (the workspace)to match the arrangement in the opposite half (the Goalspace)and indicate the minimum number of moves required for problem solution. We report that subjects are more likely to look towards the Goalspace in the initial period after picture presentation, but bias gaze towards the Workspace during the middle of trials. Towards the end of a trial, subjects are once again more likely to fixate the Goalspace. This pattern is found regardless of whether the subjects solve problems by rearranging the balls in the lower or uper visual fields, demonstrating that this strategy correlates with discreate phases in problem solving. A second experiment showed that efficient planners direct their gaze selectively towards the problem critical balls in the workspace. In contrast, Individuals who make errors spend more time looking at irrelevant items and are strongly influenced by the movement strategy needed to solve the preceding problem. We conclude that efficient solution of the TOL requires the capacity to generate and flexibly shift between control sets, including those underlying ocular scanning. The role of working memory and the prefrontal cerebral cortex in the task are discussed.