We have explored the ability of artificial neural network technologies to g
enerate performance models of complex problem-solving tasks without the det
ailed a priori knowledge of the nature of the task. To test the generalizib
ility of this approach we applied this analysis to two diverse content doma
ins-high school genetics and clinical patient management. In both domains,
the artificial neural networks, using only the sequence of actions taken wh
ile performing the task, generated multiple classification groups defining
different levels of competence. The validity of these neural network perfor
mance groupings was further established by the good concordance of these cl
assifications with independently derived expert ratings. (C) 1999 Elsevier
Science Ltd. All rights reserved.