This study examined the acquisition, retention, and transfer of a gross mot
or skill, namely, tossing, in 84 moderate-to-severely demented patients wit
h Alzheimer's disease (AD) and 72 healthy elderly controls. To identify opt
imal learning strategies, participants received 10 weeks of training under
1 of 5 practice conditions: constant, variable-parameter, variable-program,
variable-combined, or no training. Constant practice was the only training
condition that significantly enhanced learning and near transfer in AD pat
ients. In comparison, all 4 types of training facilitated acquisition and n
ear transfer in the healthy controls, with variable-combined practice being
the most beneficial. The superiority of the variable practice conditions i
n the healthy controls supports both Schmidt's (1975) variability of practi
ce hypothesis and contextual interference theory. The inability of AD patie
nts to benefit from variable forms of practice suggests that these impaired
individuals may have difficulty accessing and/or forming motor schemas.