This paper is a commentary on three other papers presented at the 1983 American Educational Research Association Symposium on "The Trainability of Information Processing and Problem Solving Skills." The papers focus on certain aspects of the training of complex skills, such as reading, that can presumably be broken down into a variety of skills. In "Teaching Vocabulary-Building Skills: A Contextual Approach," R. J. Sternberg makes the case well for the importance of knowing how to use context effectively to infer the meanings of unfamiliar words and provides a systematic approach that is certainly worthy of experimental study. "Developing Skills by Training to Develop New Automatic Components," by W. Schneider, identifies a number of issues relating to the training of complex skills and challenges what the author views as some prevailing false assumptions regarding these issues. "A Componential Approach to Training Reading Skills," by J. R. Frederiksen et. al, states that learning some of the skills that are essential to reading (via computer-based games) can be made fun, and provides compelling evidence for his claim. All three papers have useful and provocative things to say about the trainability of information-processing and problem-solving skills and, in particular, the complex skill of reading. (Author/PN)