Subjects took an implicit memory test in which pairs of unrelated word
s were flashed briefly and the ability to identify the second word of
each pair was measured as a function of prior study. In young subjects
, identification was better when the second word had been studied comp
ared to when it had not been studied (word-specific priming). Identifi
cation was also better when the two words had been paired at study com
pared to when they had been studied separately (association-specific p
riming). In amnesic patients, word-specific priming was normal compare
d to that in age- and IQ-matched control subjects. Association-specifi
c priming was observed, but it was below normal levels. This outcome s
uggests that the amnesic dysfunction can encompass priming of new verb
al associations but spare priming of existing verbal information.