A new way of classifying tasks and problems used in studies with deaf people is presented. It is claimed that this classification system allows for the specification of tasks on which deaf people perform equally well as hearing people and of tasks on which they fail in comparison. The classification system is based on certain routines of steps which a subject is required to follow to show his/her understanding of the solution of nonverbal problems. These routines are identified by focusing on the very minimum of data needed by a person outside the experimental situation to decide whether a subject has shown his or her understanding of the solution of a problem. Deaf and hearing people perform equally well when required to show their understanding by referring to one of two such minimum data. The deaf perform worse when required to refer to both of two such data. It seems likely that the difficulties met by the deaf represent a fundamental communication problem which has its roots in the way the deaf are taught or trained.