Two experiments examined the retrieval process from secondary memory. It is
argued that the difference in reaction time between the primary and second
ary memory conditions in a memory scanning task represents the time taken t
o retrieve information units from secondary memory. Previous work by Conway
and Engle (1994) suggested that this process is independent of the amount
of retrieved information. We replicated their finding, that large sets are
activated into primary memory as fast as small sets. Variation in the trial
presentation form and the length of the delays allowed an estimate of the
duration of the retrieval process from secondary memory at approximately 20
0 msec. It is argued that the process of retrieval is a bottom-up, automati
c process, which is triggered by the appearance of the set cue.