The dominant theory of transfer of training is a theory of ''common el
ements'' based on Anderson's ACT theory of skill acquisition (Singley
& Anderson, 1989). In this theory, the knowledge acquired while learn
ing a skill is encapsulated in procedures called production rules. Tra
nsfer between tasks is predicted to occur to the extent that the two t
asks share production rules or ''common elements.'' This leads to a pr
inciple of ''use specificity of knowledge'' which makes the strong sta
tement that knowledge acquired in the practice of one subskill (such a
s writing a computer program) will not transfer to performance in a re
lated subskill (such as understanding a computer program), even throug
h the two subskills rest on a shared declarative knowledge base (such
as definitions of programming language instructions) (McKendree & Ande
rson, 1987). Our research provides a test of the ACT predictions of t
ransfer and the use-specificity principle, when considering transfer b
etween two subtasks within the acquisition of computer programming ski
ll. First we provide detailed a priori transfer predictions based on a
task analysis and production system simulation model of two programmi
ng subtasks: the evaluation and generation of LISP instructions. Next,
we present results from an empirical study of training and transfer b
etween these two subtasks. Comparisons between empirical results and s
imulation predictions reveal that there is substantially more transfer
between subtasks than was predicted. In a final study we provide evid
ence that these results are due to the elaboration of declarative know
ledge. We conclude that the emphasis on procedural transfer currently
dominating the skill acquisition literature overlooks important source
s of transfer and overestimates the extent to which knowledge is use s
pecific. (C) 1995 Academic Press, Inc.