Is dexterity at pressing a buzzer on cue or fitting a peg into the proper hole a true measure of efficiency? The author of this paper thinks not. In the real world, he states, man's efficiency in coping with his environment depends not on how well he performs isolated tasks, but on how well he can co-ordinate several different functions in order to solve the problems of daily life. One way to study man as a problem-solver is to construct an artificial environment and examine the strategies used by human subjects in order to survive in this environment. N-person games are examples of this approach. The author suggests an alternative: begin with the environment, and attempt to design a subject with the minimal optimal qualities to function effectively in this environment. Suppose there were a planet where just two resources were in abundance, uranium and a nutritious fungus. How would one design a robot to go there to mine the uranium, using the fungus as his source of energy? The “fungus-eater” must also orient himself on the planet, the surface of which is paved with black and white pebbles randomly distributed. What are the minimum abilities which the robot must have in order to survive and to perform his task effectively?