On the implications of users' prior knowledge for human-computer interaction

Abstract

This paper analyses the situation in which a beginning computer user tries to handle a computer system by only having had a brief period of instruction and manned with a manual. The situation is analysed as a problem solving situation, in which knowledge about how similar tasks are handled outside of the system plays a great role. It is suggested that the following situations will lead to slow learning: when the problem space is great, when necessary methods are difficult to access, when prior methods are inadequate and strong, when prior models are inadequate, and when the problem formulation is misleading. It is further suggested that the following may be learned in this situation: situation specific goal-condition-method rules, higher order rules, problem schemata, and causal explanations.

Empirical observation by means of think aloud protocols and registering of actual interactions are presented. These show that difficulties encountered by beginning users may be interpreted as suggested above. As to the learning content, the observations suggest that beginning users primarily learn situation specific goal-condition-method rules. They may furthermore redefine old or create new problem schemata. Higher order rules and causal explanations were not evident in the data collected.