A striking feature of people engaged in problem solving activities outside the psychologist’s laboratory is that it exhibits a great deal of interactivity in a physical space populated with external symbols, artifacts, and, of course, other people. Yet, problem solving researchers often design experimental procedures in which interactivity is either limited or eliminated. We review three traditional areas of problem solving research and introduce new experimental methodologies wherein problems can only be solved by manipulating or restructuring a physical space. In all instances, problem solving is markedly superior than when observed in control two-dimensional non-interactive versions of these tasks. We suggest that the nature of the processes engaged in solving problems in distributed environments is different than in static environments and should encourage cognitive psychologists to revisit the process models elaborated to account for problem solving behaviour.