![](https://csdnimg.cn/release/download_crawler_static/15606630/bg4.jpg)
conditions (e.g. Aly, Knight, & Yonelinas, 2010; Bader,
Mecklinger, Hoppstädter, & Meyer, 2010; Quamme,
Yonelinas, & Norman, 2007; Rhodes & Donaldson,
2007, 2008; Yonelinas, Kroll, Dobbins, & Soltani,
1999). A common feature of these conditions is
that specific encoding procedures help integrate
to-be-associated items into a coherent whole
during study, a process that is referred to as unitisa-
tion (Graf & Schacter, 1989 ; Mayes, Montaldi, & Migo,
2007). Examples include the interactive imagery of
two words (e.g. violin–apple; Rhodes & Donaldson,
2008) or establishing a new conceptual meaning
by presenting a definition (e.g. smoke–apple: a
fruit maturing above flames) (e.g. Bader et al.,
2010). If, as expected, enactment similarly unitises
the action and the object, familiarity could be used
to reject recombined items after SPT but not
after VT.
The first study investigating the relative contri-
bution of familiarity and recollection to recognition
following SPT and VT was reported by Engelkamp
and Dehn (1997). Participants first made an old/
new decision in an item recognition test, and
after each decision, they judged whether they
only knew that the item was old or new (“know”)
or whether they instead remembered a specific
detail about an item’s prior presentation (“remem-
ber”) (Remember/Know judgments, R/K judgments
hereafter; Tulving, 1985). Engelkamp and Dehn
found an increase in R judgments after SPT com-
pared to VT. Because R judgments are considered
an indicator of recollection, those authors con-
cluded that the SPT effect in item recognition is
mainly due to enhanced recollection process after
SPT. However, this conclusion is somewhat proble-
matic because the feeling of “knowing that I did
it” was instructed to classify the items as remem-
bered (R). Such R judgments can involve relatively
coarse memory and do not require remembering
the binding of the action and the object. Therefore,
the enhanced proportion of R judgments cannot be
taken as evidence that the action-object relation is
recollected after SPT. Another problem is that the
R/K judgments describe the final status of the rec-
ognition process but not the process itself. At the
behavioural level, R and K judgments are exclusive
because they are two alternative forced-choice
decisions. However, the underlying mental states
(i.e. recollection and familiarity) are not exclusive.
Therefore, items that were judged as recollected
(R) might nevertheless vary in their experienced
familiarity.
Accordingly, regarding associative recognition,
even
if an accepted intact item and a rejected
recombined item (recall to reject) are both classified
as R, these two items might not have equal famili-
arity. If only memory for the individual components
of the action verb and the object noun influences
familiarity, the intact and recombined verb–object
pairs would be equally familiar. However, if enact-
ment encourages unitisation of the object and the
action, recombined items should appear less familiar
than intact items. We assume that the former
happens after VT encoding, whereas the latter
happens after SPT encoding. Therefore, familiarity
can contribute to associative recognition after SPT
but not after VT. The R/K judgments would not dis-
close this difference if the items were finally cor-
rectly recognised and recollected in both
conditions. We need another measure that is both
sensitive to familiarity and capable of measuring
memory processing in real time during retrieval. For-
tunately, electrophysiological correlates of recog-
nition memory provide such a possibility.
Event-related potentials (ERPs) are often used to
gain insight into memory processes during retrieval
(Curran, 2004; Parks & Yonelinas, 2007) by analysing
ERP old–new effects (see Wilding & Ranganath,
2011, for a review). The standard ERP old–new
effects are differences between the neural activity
elicited by correctly classified old (studied) test
items and the neural activity elicited by correctly
classified new (unstudied) test items. In this tra-
dition, two different ERP components are inter-
preted as correlates of familiarity or recollection
(for reviews, see Mecklinger, 2000; Rugg & Curran,
2007; Zimmer & Ecker, 2010). The first old–new
effect occurs approximately 300–500 ms after the
onset of the test item, typically at frontal sites (the
early frontal old–new effect, also called FN400),
which is considered to index familiarity. Although
some researchers have interpreted this component
as conceptual priming (e.g. Paller, Voss, & Boehm,
2007; Voss, Lucas, & Paller, 2012), we follow the
majority view and take the FN400 as a familiarity
index (see also Bridger, Bader, Kriukova, Unger, &
Mecklinger, 2012). The second old–new effect is
observed around 500–800 ms post-stimulus onset
at parietal sites (the late parietal old–new effect,
also called LPC), which is hypothesised to reflect
recollection.
Although no study has explored the contribution
of familiarity and recollection to associative recog-
nition of verb–object phrases, several previous
JOURNAL OF COGNITIVE PSYCHOLOGY 3