The Strategy Specific Nature of Improvement:
            The Power Law Applies By Strategy Within Task

                           Peter F. Delaney
                     The Florida State University

                Lynne M. Reder and James J. Staszewski
                      Carnegie-Mellon University

                           Frank E. Ritter
                  University of Nottingham, England

                   Psychological Science, 9(1). pp. 1-8.
                            10 March 1996

                               Abstract

If strategy shifts speed up performance, learning curves
should show discontinuities where such shifts occur.  Relatively
smooth curves consistently appear in the literature, however.  To
explore this further we examined learning where their were multiple
known strategies.  We plotted aggregate power law learning curves for
the data from four mental arithmetic experiments and then plotted
similar learning curves separately for each subject and strategy.  We
then assessed the amount of unique variance accounted for by each
group of curves.  In the first three experiments, the regular power
law did not account for any unique variance, but the individual
strategy curves did.  In the final experiment, both accounted for
unique variance, although the amount accounted for by the regular
power law was small.  This can be attributed to insufficient
information about strategy use.  We conclude that improvement of
response time is better explained by practice on a strategy rather
than practice on a task, and that careful assessment of trial-by-trial
changes in strategy is necessary to adequately address the role of
practice in learning.