A time window is defined as the interval between the occurrence of an
initial event and the upper limit of the period within which new infor
mation can be integrated with the memory representation of that event.
The validity of the time window construct was documented in a prior r
eport involving a categorization paradigm and in which a new exemplar
was integrated into a previously established category. Presently, we a
sked if there is a time window within which information can be integra
ted across successive training sessions and, if so, what are its impli
cations for long-term retention. In two studies with 3-month-olds, we
found that (a) there is a delay between sessions after which the effec
ts of successive sessions no longer cumulate, (b) when the second sess
ion occurs nearer the end of the time window, retention is greater, an
d (c) a memory can be reactivated only if the second session occurs wi
thin the time window. These data further validate the time window cons
truct and illustrate that the timing of a succeeding event determines
whether it will be integrated with what came before or will be treated
as unique. This result has major implications for the development of
an individual's knowledge base.