Recent positron emission tomography (PET) studies have identified neur
onal components of widespread novelty-assessment networks in the brain
. We propose that the efficacy of encoding on-line information into lo
ng-term memory depends on the novelty of the information as determined
by these networks, and report a test of this ''novelty/encoding'' hyp
othesis. Subjects studied a list of words. Half of the words were ''fa
miliar'' by virtue of their repeated presentation to the subjects befo
re the study of the critical list; the other half were novel, in that
they had not previously been encountered in the experiment. The result
s conformed to the prediction of the novelty/encoding hypothesis: accu
racy of explicit (episodic) recognition was higher for novel than for
familiar words.