During the summer of 1986 a conference funded by the National Science Foundation (NSF) was organized to assess the current state of cognitive research on the psychology of physics problem solving, and to examine the needs of physics instructors and instructional designers that must be addressed by a psychological theory of physics problem solving. This paper outlines a consensus model of the novice physics problem solver that was identified at the conference. It describes a general perspective and a frame of reference about physics problem solving, and indicates some of the disagreements. The paper focuses primarily on the observation that novices often have multiple interpretations of physics concepts. Both theoretical and practical implications of this observation are discussed. It is suggested that this framework be used to interpret many efforts to design effective physics instruction, particularly those concerned with problem solving performance and learning. (TW)