Explanation and Learning in Procedural Skills.
Final rept. 1 Sep 85-31 Dec 88,
COLORADO UNIV AT BOULDER DEPT OF COMPUTER SCIENCE
Pagination or Media Count:
This report summarizes the findings of an investigation into the role of explanations in learning procedures. Experimental and theoretical results from studies of the analysis of examples and generalization methods, and issues remaining open, are presented. Computer operations are often learned from demonstrations. Effective learning requires not just seeing and remembering what was done but being able to modify what was done to meet the requirements of new tasks. The EXPL model provides an account of how causal analysis could be used to support this kind of generalization. We investigate learning from video demonstration. Learning to use a computer system is difficult, and existing training methods are costly and often ineffective. The results of a study of learning to use a spreadsheet from a 15 minute video demonstration is reported. Learners who viewed the video performed less well than other learners who worked through a hands-on tutorial manual showing the same operations. Easy learning of a user interface depends in part on users being able to generalize successfully about it. Philosophical doctrine, and some recent work in human-computer interaction, argues that casual analysis of interactions can support generalization. We propose a rigorous causal analysis theory, and show how it accounts for two robust generalizations, using certain general assumptions. AW
- Personnel Management and Labor Relations