A hypermnesic task was administered to 82 younger adults (ages 27-39),
63 middle-aged adults (ages 40-59), and 119 older adults (ages 60-87)
. Previous research suggests that relational encoding prevents loss of
items and item-specific encoding promotes item gains in a hypermnesic
task (Klein et al., 1989) and that there are age differences in relat
ional but not item-specific encoding (Luszcz et al., 1990). This infor
mation provided the basis for three predictions: (a) There are age dif
ferences in hypermnesia, (b) there are age differences in word losses
in a hypermnesic task, and (c) there are no age differences in word ga
ins in a hypermnesic task. In order to manipulate type of encoding, a
list of words with high association strength (to evoke relational enco
ding) and words with low association strength (to evoke item-specific
encoding) was constructed. The results of this investigation provide s
upport for the encoding manipulation and for all three predictions. In
addition, the nature of the age differences in word loss observed sug
gests that although older adults may be capable of relational encoding
, this form of encoding is not as effective at preventing word loss fo
r them as it is for younger adults.