Abstract
Understanding how to use a computer system often requires knowledge of hidden events: things which happen as a result of user actions but which produce no immediate perceptible effect. How do users learn about these events? Will learners explain the mechanism in detail or only at the level at which they are able to use it? We extend Lewis' EXPL model of causal analysis, incorporating ideas from Miyake, Draper, and Dietterich, to give an account of learning about hidden events from examples. We present experimental evidence suggesting that violations of user expectations trigger a process in which hidden events are hypothesized and subsequently linked to user actions using schemata for general classes of situations which violate user expectations.
