In a study designed to distinguish between parallel and serial models
of performance on Sternberg's item recognition task, Krueger (1975) pr
esented subjects with a memory set (the ''added'' set) followed by a s
econd (''core'') set, a probe to the core set, and finally a probe to
the added set. Response latency (RT) to both the core- and added-set p
robes initially increased with the size of the extraneous set. Over su
ccessive blocks of trials, the effect of the size of the extraneous se
t diminished. In Experiment 1 of the present study, Krueger's basic pr
ocedures were replicated with results similar to those he reported for
early trials. Experiment 2 was a variation on the added-set procedure
. Rather than the memory sets' being drawn from different categories (
letters and digits), both sets consisted of digits. On half of the neg
ative trials for both the core and the added tasks, the probe was draw
n from the extraneous set, forcing the subjects to separate the two se
ts. Although RT on the core task did not increase with the size of the
added set, there was an increase in RT to the added set with increase
s in core-set size, a finding inconsistent with the serial search mode
l.