THE INFLUENCE OF THE USER-INTERFACE ON SOLVING WELL-DEFINED AND ILL-DEFINED PROBLEMS

It is well documented that people can learn in two different modes. An explicit mode is recognized by a conscious and selective attention to wards the problem, while in the implicit mode, the learning is more un conscious and trial and error based. According to Hayes and Boradbent the success of employing each of these modes depends on characteristic s of the problem. By applying the theory to human-computer interaction the users of conversational interfaces are predicted to be induced to an explicit learning mode, while users of direct manipulation interfa ces would be induced to an implicit mode. The ability of the user inte rface to induce a learning mode makes it important to identify problem s which can be optimally solved with each of the interfaces. Two exper iments were designed to test if problems characterized as well- or ill -defined can predict which of the interfaces would be more successful. Well-defined problems were expected to be better solved by a conversa tional interface, and ill-defined problems with direct manipulation in terfaces. The experiments demonstrate that the interfaces can induce t he two different learning modes as predicted. The results only partly support that problem definition can predict which interface would be m ore successful. The problem definition concept is therefore not recomm ended as a criterion to select interface. (C) 1996 Academic Press Limi ted