Int. J. Man-Machine Studies (1990) 32, 385-398 Do they know what they're doing? An evaluation of word-processor users' implicit and explicit task-relevant knowledge, and its role in self-directed learning PAMELA BRIGGSt MRC/ESRC Social and Applied Psychology Unit, University of Sheffield, Sheffield, $10 2TN, UK (Received 10 March 1988 and in revised form 10 February 1989) Many people teach themselves how to use w0rd-processing systems, but how successful are they in their endeavor? This study investigates a number of theoretical and practical issues associated with self-directed learning. Users of differing experience were asked to perform a simple task, using an unfamiliar word- processing system. However, they were given no information about the new system, prior to task commencement, save information they explicitly requested. An analysis of users' questions revealed that only tile most experienced had a suitable mental task description available to them. Others relied upon visible components of the task to cue their questioning strategy in a manner which suggested reliance upon a recognition, rather than a recall strategy. A clear dissociation was noted between users' procedural knowledge of a task, reflected in their performance ability; and their metaknowledge of the task, i.e. their awareness of what procedural knowledge would be required in order to complete the task. The implications of these findings for the design of user support systems, and for user modelling are discussed. Introduction Studies which try to chart the process of self-directed learning are interesting for both practical and theoretical reasons. In practical terms, they are relevant to the design of an interface, manual or training package. Increasingly, micro-computers are being marketed with an "easy to learn" label. Market forces are such that systems which need least support in terms of training, manuals, etc. will be most competitive. It is therefore surprising that, while self-directed learning is implicitly recognised as the means by which most casual users (and many professional users) learn about systems, it has been given relatively little attention within studies of human-computer interaction. The studies which do exist have shown that computer users are not very good at self-directed learning (Carroll & Rosson, 1987; Briggs, 1988). To date, three major factors have been identified as contributing to this. First, self-taught users are said to be beset by a "motivational paradox" (Carroll & Rosson, 1987) in which the goal to complete work conflicts with the goal of learning more about the system, with the result that users tend to learn only a small subset of the commands available to them, and make poor use of manuals, which t Now at the Department of Psychology,City of London Polytechnic. 385 0020-7373/90/040385+ 14503.00/0 ~ 1990 Academic Press Limited