www.elsevier.com/locate/cogbrainres

Cognitive Brain Research

Research report

The late negative episodic memory effect: the effect of recapitulating

study details at test

David Friedman*, Yael M. Cycowicz, Michael Bersick

Cognitive Electrophysiology Laboratory, New York State Psychiatric Institute, Unit 6, 1051 Riverside Drive, New York City, NY 10032, United States

Accepted 18 October 2004

Available online 21 November 2004

Abstract

An hypothesis concerning mnemonic function suggests that perceptual details of previously experienced episodes are retrieved from the

cortices that initially processed that information during the encoding phase. Cycowicz et al. [Cycowicz, Y.M., Friedman, D. and Snodgrass,

J.G., Remembering the color of objects: an ERP investigation of source memory, Cereb Cortex, 11 (2001) 322–334.] have interpreted the

presence of a late negative episodic memory (EM) effect, maximal over parieto-occipital scalp, as a brain signature of the search for and/or

retrieval/evaluation of the specific perceptual source-specifying attributes (i.e., color) of pictures in the visual cortical regions that were

recruited during the encoding of that information. The present study assessed the validity of this hypothesis. Twelve participants studied

pictures outlined in red or green and were subsequently tested with inclusion (i.e., item; old or new regardless of color) and exclusion (i.e.,

source; same color, different color/new judgments) tasks. In both, old pictures were presented either in the same color as at study or in the

alternate color. A late negative, parieto-occipital EM effect was of much larger amplitude in the source compared to the item task. It was of

similar magnitude to correctly recognized pictures whose colors were identical at study and test relative to those whose colors changed, and

was not modulated by the success or failure of the source retrieval. These data run counter to the initial hypothesis that the late negative EM

effect reflects the search for and/or retrieval of specific perceptual attributes such as color. Rather, the late negative EM effect may reflect the

search for and/or retrieval/evaluation of more general source-specifying information in the cortical regions that initially processed the stimuli.

D 2004 Elsevier B.V. All rights reserved.

Theme: Neural basis of behavior

Topic: Learning and memory: physiology

Keywords: Item memory; Source memory; ERP episodic memory (EM) effect

1. Introduction

Investigators of memory function have often contrasted

two processes, familiarity and recollection, which are

thought to underlie recognition memory performance [18].

Whereas familiarity is relatively automatic and is hypothe-

sized to underlie the retrieval of item or content information

without the details, or context within which the event was

embedded, recollection is effortful and is required when

retrieving contextual information, such as the spatio-

0926-6410/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.cogbrainres.2004.10.005

* Corresponding author. Fax: +1 212 543 6002.

E-mail address: df12@columbia.edu (D. Friedman).

(where) temporal (when) attributes within which the initial

episode was encountered. The retrieval of contextual

attributes is labeled source memory.

Compared to simple, old/new recognition memory

paradigms, source memory paradigms, in addition to

requiring old/new memory judgments, also solicit judg-

ments concerning the context within which the original

episode was experienced. For example, during a study phase

participants might hear words presented in either a male or

female voice. During the subsequent test phase, subjects

would be asked to make old/new judgments to visually

presented words. Then, for any word judged old, they would

be asked to provide a source judgment concerning the

original presentation, was it presented in either the male or

23 (2005) 185–198

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198186

female voice? [35]. In this type of memory experiment,

simply judging whether the item is old or new can be

accomplished without reference to contextual details, i.e.,

this judgment can be based solely on whether the item

seems familiar or not.

However, remembering whether the item was presented,

for example, by a male or female voice, requires active

recollection of such contextual information. In these kinds

of designs, a recollection-based response has been oper-

ationalized as a correctly recognized old item (i.e., a hit)

whose source has also been correctly retrieved. A familiar-

ity-based recognition response has been operationalized as a

correctly recognized old item whose source has been

incorrectly attributed. Presumably, this indicates that this

latter type of recognition judgment was not accompanied by

contextual detail. That is, an item correctly recognized as

old, but receiving an incorrect source judgment is assumed

to be based only on familiarity; an item correctly recognized

as old attracting a correct source judgment is assumed to be

based on familiarity as well as recollection.

This source memory paradigm has been very often

employed in event-related brain potential (ERP) studies of

recognition memory. The aim of these investigations has

been to obtain brain activity signatures corresponding to

familiarity and recollection hypothesized to operate in two-

process theories of recognition memory [18]. Investigators

using this type of paradigm have uncovered a series of old/

new or episodic memory (EM) effects that appear to reflect

unique mnemonic functions [7,13,24,25]. Typically, cor-

rectly recognized old items elicit greater positivity than

correctly rejected new items. The EM effect is then defined

as the ERP difference between correctly recognized old

and correctly rejected new items. The most consistently

reported EM effects have been labeled the medial

prefrontal EM effect (active between about 300 and 500

ms), the left parietal EM effect (500–900 ms), and the right

prefrontal EM effect (800–2000 ms). Some consensus as to

the functional roles of each of these EM effects exists. The

medial prefrontal EM effect has been associated with the

familiarity component of recognition memory [2,14,19]

(but see [37] and [29]). This conclusion is based on the

findings that the medial prefrontal EM effect is of

equivalent magnitude in the ERPs associated with pre-

viously studied, correctly recognized old items whether

those items were given a brememberQ (retrieval of contextbased on recollection) or bknowQ (retrieval of content

based on familiarity) judgment [28], according to the

paradigm originally described by Tulving [30]. In the same

vein, the medial prefrontal EM effect is generally of

equivalent magnitude in the ERPs associated with old

items and unstudied lures that are highly similar to one

another and thus generate a large familiarity signal [2,22].

The subsequent parietal EM effect has been associated

with recollection, based on a large number of findings

indicating that this EM effect is larger in association with

items whose sources are correctly attributed compared to

those that are not [28,34,35]. The functional role of the

right prefrontal EM effect is currently controversial,

although some investigators have advanced the hypothesis

that it reflects some kind of executive control function,

such as monitoring the products of retrieval in the service

of modifying ongoing memory performance (e.g., Refs.

[25,36]).

Relatively few types of contexts or sources have been

used in these investigations, which have included gender of

voice, spatial location and list membership (i.e., temporal

context). Despite the obvious differences, these source

types have yielded fairly similar EM effects. By contrast

with these types of source information, Cycowicz et al. [5]

used line drawings of common objects that were painted in

either red or green during the study phase. During two

ensuing test phases, pictures were outlined in black and, in

the inclusion or item test, subjects had to judge simply

whether the picture was old or new, regardless of its outline

color in the study phase; during the exclusion or source

test, subjects had to judge whether the item was initially

painted in a target color (for example, red) during study

(hereafter referred to as Targets), or was painted in the

alternate color (green in this example) or was new (these

latter responses were assigned to the same response hand).

Previously studied pictures painted in the alternate color are

hereafter designated as Nontargets to distinguish them from

new pictures.

Unlike previous investigations of source memory, Cyco-

wicz et al. recorded a large-amplitude negative EM effect,

maximal over parieto-occipital scalp, which was markedly

larger in the source than the item task. Due to three factors,

(1) the source in this paradigm was color, (2) the negative

EM was coincident with mean reaction time (RT), and (3)

the negative EM effect showed a topographic focus over

parieto-occipital scalp, Cycowicz and co-workers suggested

that the late negative EM effect could have reflected the

search for and/or retrieval of the color in which the picture

was painted during the study phase in the cortical regions

that originally processed color information.

As mentioned, the peak latency of the negative EM effect

(between ~800 and 1000 ms) occurred at about the same

time as mean reaction time. The negative activity onset as

the preceding parietal EM effect was returning to baseline.

On this basis, Cycowicz et al. [5] advanced the hypothesis

that sufficient time would have elapsed for the late negative

EM effect to have reflected brain activity related to an

attempt to reinstate the initial image along with its

associated color during the source retrieval task [6]. Further,

in the original Cycowicz et al. study [5], the negative EM

effect was as large during successful as it was during

unsuccessful source retrieval. Hence, the breinstatementQinterpretation offered by Cycowicz and co-workers is

consistent with the presence of negative-going activity in

the ERPs associated with trials on which an incorrect source

decision had been made, as subjects would have had to

attempt to retrieve the conjunction of attributes (the picture

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198 187

and its color) regardless of the success of the retrieval

attempt.

Although other investigators had also observed similar

negative-going EM effects (review by Ref. [11]), Cycowicz

et al. [5] were, to our knowledge, the first to compare

directly this activity between item and source retrieval

tasks. Subsequent investigations have also assessed the

difference between item and source tasks. For example, in a

study by Johansson et al. [12], subjects viewed words

which were followed by the presentation of a rectangular

outline. In half the trials, a picture was presented within the

rectangle. In the remainder, a picture was not presented and

subjects were asked to imagine the object named by the

label and project it into the rectangular outline. For both

types of trials, subjects were asked to determine how well

the pictured (or imagined) object fit the verbal label. At test,

subjects performed two tasks, in which only the verbal

labels of previously studied and new items were intermixed.

One task was a simple, old/new recognition paradigm, the

other a source monitoring procedure. In the latter, partic-

ipants had to retrieve the action they performed during

study (either viewing the object or imagining it) in

association with the verbal label. By contrast with the old/

new recognition task, a late-onset, negative EM effect was

of much larger amplitude in the source task. It peaked

between 1000 and 1200 ms and was maximal over parieto-

occipital scalp, highly similar to the negative EM effect

recorded by Cycowicz et al. [5] (the scalp topography of the

Johansson et al. [12] negative EM effect can be seen in Fig.

2 of Johansson and Mecklinger [11]). Johansson et al. [12]

raised the possibility, consistent with the interpretation

offered by Cycowicz et al. [5], that the negativity reflected

the reinstatement of the original object (or imaged object)

along with the action that was performed (perceiving or

imagining).

Although lacking a comparison of item and source

tasks, Leynes et al. [16] also assessed source monitoring

by asking their subjects during study to perform an

action or to plan to perform the action in response to the

presentation of an action phrase. During test, the action

phrases (planned and performed) were re-presented

intermixed with new action phrases. Participants were

required to identify the phrase as performed, planned or

new. Leynes et al. recorded a large-amplitude, negative

EM effect that peaked between 1200 and 1800 ms and

displayed a parieto-occipital topography. Because both

planning to perform an action and actual performance of

an action phrase involve a high degree of visual

processing, the Leynes et al. data may also be interpreted

as consistent with the reinstatement of the representation

of the action and its contextual attribute, i.e., whether it

was planned or performed. In both the Johansson et al.

[12] and Leynes et al. [16] investigations, the parieto-

occipital topography of the negative EM effects is

consistent with these computations involving visual

cortical processing regions.

Johansson and Mecklinger [11] have reviewed these and

other studies in which late, negative EM effects have been

observed. The main conclusions from the review are that (1)

this negative activity can be recorded in old/new recognition

memory paradigms, provided baction monitoringQ is

required—typically engendered by difficult response

demands, as in false memory paradigms. In these tasks the

negativity is observed with a posterior scalp topography, but

only in reaction time (RT)-locked averages; (2) the late

negative EM effect occurs whether retrieval is successful or

not and (3) when attribute conjunctions may have to be

retrieved (as in source memory paradigms), the negativity is

observed in stimulus-locked averages. Johansson and

Mecklinger [11] proposed that, in source memory tasks,

the stimulus-locked, late negative EM effect could reflect

b. . . processes related to forming and holding a representa-

tion of a conjunction of attributes that specify the prior

episodeQ (p. 23), for example, the pictorial object and its

color as in previous investigations from this laboratory [3].

As stated earlier, both the original object and its associated

color might be reinstated for evaluation.

Based on the fact that the late negative EM effect is not

associated with successful retrieval, it is possible that,

similar to the functional role proposed for the right

prefrontal EM effect, the late negative EM effect might

reflect monitoring and/or evaluative operations that, due to

the perceptual nature of the source (e.g., color; visualizing

whether an object was imagined or viewed), take place in

posterior cortical regions associated with visual processing.

In previous investigations from this laboratory, pictures

were presented in red or green outline during the study

phase, but were presented in black outline during the test

phases. In the current investigation, by contrast, pictorial

objects were presented in red or green outline during the

encoding phases and then, during two test phases, were re-

presented either in the same color as during study or the

alternate color. During the item test, subjects were required

to judge the items as old or new regardless of the color in

which they were outlined during study and test. On the other

hand, during the source test subjects had to make Same (old

object, same color) or Different (old object, different color)/

New judgments, labeled, respectively, target old, nontarget

old and new items.

It was predicted, based on previous work from this [3,5]

and other [11] laboratories, that the amplitude of the

negative EM effect would not be modulated by whether or

not retrieval of source attributes was successful. Johansson

and Mecklinger [11] postulate that the late negative EM

effect is generated when attributes of the previously studied

item are not easily recovered upon presentation of the test

probe. In the current experimental design, this might occur

when the object presented at test is painted in a different

color than its studied counterpart, because the color cue does

not match the stored representation. On this view, one would

predict larger negative activity when the test cue and the

stored representation differ. Johansson and Mecklinger [11]

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198188

have also suggested that the magnitude of the posterior

negative EM effect increases when response demands are

complex or difficult. On this basis, one could predict that

this negative EM effect would be larger to nontargets in the

exclusion or source task, as these items require a complex

hand to response mapping, i.e., nontarget old and new items

are both assigned to the same response hand, whereas target

old items require a different response hand.

1 For complete details and rationale for the use of the proportions of old

and new items used here, see Cycowicz et al. [5].

2. Materials and methods

For ease of exposition, the inclusion or old/new

recognition task will be referred to as the bitemQ task,

because the retrieval of contextual information is not

explicitly required. By contrast, the exclusion task, which

requires retrieval of context, will be referred to as the

bsourceQ task. We use these terms as convenient labels, even

though on a proportion of bitem taskQ trials recollection may

have been involved and on a proportion of bsource taskQtrials a familiarity process was undoubtedly employed.

2.1. Subjects

Nineteen young adults were recruited for this study. The

data of 7 could not be used, 5 due to excessive eye

movements, 1 due to very poor performance, and 1 due to

technical difficulties during the experimental run. The data

of the remaining 12 young adults (9 female; mean age=23)

are the subject of this report. All subjects were native

English speakers, reported themselves to be in good health

and to have no major medical, neurological, or psychiatric

problems. All participants signed informed consent and

received payment for their participation.

2.2. Stimuli

The experimental stimuli were 312 unambiguous line

drawings of common objects that were divided into 6

lists of 52 pictures each, with lists carefully constructed

so that they were equated on category membership,

concept agreement, name agreement, familiarity, and

visual complexity according to previously published

norms [1,4,27]. Statistical analysis of the variables

characterizing the picture sets revealed no significant

differences among lists (PN0.10). An additional 52

pictures from the same normative sources, not used in

the experimental phases, were used for a practice session

and as fillers. The experiment was divided into six

phases. Each phase consisted of one study and two test

blocks, item and source. In each phase, one of the six

lists of pictures was used, with the order of list

presentation randomized across phases separately for each

subject. Of the 52 pictures in a list, 32 were randomly

assigned to the study block, while the remaining 20 were

assigned as foils to the test block.

2.3. Procedure1

The order of the item and source tests was counter-

balanced across subjects. Each picture was displayed for

500 ms with an ISI of 2000 ms. Study Block. Subjects

viewed 32 pictures, half in red and half in green, and were

asked to press one button for red and another button for

green pictures. Subjects were instructed to memorize the

pictures and their associated colors for a subsequent

memory test. Test Block—Item Recognition. Subjects

viewed 26 pictures, half in red and half in green, of which

12 were old (seen during study) and 14 were new. Of the 12

old items, half were presented in the same color as during

study, while the remainder was presented in the alternate

color. Subjects were asked to press one button for old and

another button for new pictures, regardless of the color in

which they were presented at study. Test Block—Source

Recognition. Subjects viewed 26 pictures, half in red and

half in green, of which 20 were old and 6 were new. Of the

20 old items, half were presented in the same color as during

study, while the remainder was presented in the alternate

color. Half of the new pictures were randomly assigned to

red and half to green. Subjects were asked to press one

button for old pictures presented in the same color as during

study (SAME), and another button for old pictures

presented in a different color (DIFF) during study. The Diff

button was also used for new pictures.

In order to ensure that all subjects knew which button to

press, cues were presented on the computer screen during

the entire block. Hence, during the study block, small

rectangles were presented below and to the right and left of

the to-be-remembered pictures. In one rectangle the word

bREDQ appeared in red and in the other the word bGREENQappeared in green. The left/right positions of these cues

reflected the hand assigned to each color. Similarly, during

the item test blocks, bOLDQ and bNEWQ cues in black

lettering appeared in rectangles below and to the right and

left of the pictures, again consistent with the assigned hand

of response. During the source test blocks, the cues

contained the word bSAMEQ and bDIFFQ in black letters.

During study, item and source blocks, subjects made choice,

speeded and accurate, respectively, bredQ/bgreen,Q boldQ/bnew,Q and bSAMEQ (old)/bDIFFQ (old painted in a different

color or new) decisions to each picture. The hands assigned

during study to bredQ and bgreenQ buttons, during the item

task to boldQ and bnew,Q and during the source task to

bSAMEQ and bDIFFQ were counterbalanced across subjects.

The horizontal visual angles ranged from 0.858 to 4.818, andthe vertical visual angles from 0.568 to 3.408, respectively,for the smallest and largest pictures. To counterbalance

order effects, in half of the test phases, the item task

preceded the source task and, in the other half, source

testing preceded item testing. Subjects were not informed

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198 189

prior to the study-block which test block would be

administered first. The sequence of stimuli was separately

randomized for each subject.

Hereafter, for the item task, old pictures presented in the

same color at study and test are referred to as bOld Same;Qold pictures presented in different colors at study and test are

referred to as bOld Diff.Q For the Source task, old pictures

presented in the same color at study and test (requiring a

response of bSameQ) are referred to as bTargetQ items; old

pictures presented in different colors at study and test

(requiring a response of bDifferentQ) are referred to as

bNontargetQ items. Correctly rejected new items are labeled

bNewQ (in the source test, these items also required a

response of bDifferentQ).

2.4. EEG recording

EEG (sintered Ag/AgCl electrodes; DC; 100 Hz upper

cutoff; 250 Hz digitization rate) was recorded continuously

with Synamp amplifiers (Neurosoft) using an Electrocap

with 62 locations (extended 10–20 system placements [23],

including left and right mastoids). All leads were referred to

nose tip. Vertical EOG was recorded bipolarly from

electrodes placed on the supraorbital and infraorbital ridges

of the right eye, and horizontal EOG was recorded bipolarly

from electrodes placed on the outer canthi of the two eyes.

Trials containing eye movement artifact were corrected off-

line using the procedure developed by Gratton et al. [8].

Trials were epoched off-line with 100 ms pre- and 1900 ms

post-stimulus periods.

2.5. Data analyses

ERPs were averaged to correctly recognized Old same and

Old diff and correctly rejected new pictures during the item

task and to correctly recognized Target and Nontarget old

trials and correctly rejected new items during the source task.

The SPSS V. 11.5 repeated measures ANOVA program was

Table 1

Behavioral data in the item (A) and source (B) tasks

(A)

Item task

%

Hits

RT

Hits

%

CR

RT

CR

%

FA

%

S

Mean 81.9 829.4 94.2 836.3 5.8 8

S.D. 7.4 152.2 4.5 153.9 4.5

(B)

Source task

%

Target

RT

Target

%

CR

RT

CR

% FA

New

%

T

Mean 73.1 956.2 94.6 841.6 5.0 6

S.D. 10.1 153.9 4.6 162.5 7.4 1

CR=correct rejection; FA=false alarm; Diff=different.

A target in the source task is a picture correctly recognized as having been present

correctly recognized during the test phase as having been presented in the alterna

Pr=measure of discrimination of old from new items; Br=measure of bias, both c

In the source task, Pr was computed in two ways: (1) by subtracting the false alarm

alarm rate to non-target items from the target old hit rate (z). Two Br indices wer

* Reliably different from zero as assessed by t-test.

used for all analyses. The Greenhouse–Geisser epsilon (e)correction [10] was used where appropriate. Uncorrected

degrees of freedom are reported along with the epsilon value;

the P values reflect the epsilon correction. Significant main

effects and interactions were followed-up, where appropriate,

with simple effects tests and/or post-hoc analyses using the

Tukey Honestly Significant Difference (HSD) test.

To be consistent with our previous work on the late

negative EM effect [5], and to capture any anterior/posterior

and/or left/right asymmetries, the main ANOVAs were

performed on the data recorded from 24 scalp sites along

lateral (left, midline, right) and anterior/posterior planes.

The 24 scalp sites included on the left, FP1, F3, FC3, C3,

CP3, P3, PO3, and O1; on the midline, FPz, FCz, Fz, Cz,

CPz, Pz, POz, and Oz; on the right, FP2, F4, FC4, C4, CP4,

P4, PO4, and O2. The ANOVAs always included the factors

of Laterality (left, midline, and right), and Anterior/Posterior

scalp location (Frontal Pole, Frontal, Fronto-Central, Cen-

tral, Centro-Parietal, Parietal, Parieto-Occipital, Occipital).

However, unless they interacted with the variables of

interest (e.g., item vs. source tasks), the main effects or

interactions of Laterality and Anterior/Posterior location are

not interpreted or reported as, by themselves, they do not

reflect memory-related differences.

3. Results

3.1. Behavioral data

Table 1 presents the behavioral data from the item (A)

and source (B) tasks.

3.1.1. Reaction time (RT)

RTs in the item and source tasks were compared in an

ANOVA that assessed the effects of Task (item, source) and

Study/Test color pairing (same, different). RTs were longer

in the source (1021 ms) than in the item (831 ms) task

ame

RT

Same

%

Diff

RT

Diff

Pr Br

4.5 812.0 79.2 849.0 0.75* 0.26

9.8 148.7 8.0 166.0 0.08 0.14

Non-

arget Diff

RT

Nontarget

Pr# Br# Prz Brz

1.9 1086.7 0.67* 0.17 0.35* 0.58

1.3 207.7 0.12 0.11 0.13 0.12

ed in the same color during study. A nontarget in the source task is a picture

te color during study.

omputed according to Snodgrass and Corwin [26].

rate to new items from the target hit rate (#); and (2) by subtracting the false

e also computed using the new and non-target false alarm rates.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198190

(F(1,11)=48.43, Pb0.0001), and Old same/Target items

were responded to faster (884 ms) than Old Diff/Nontarget

items (968 ms; F(1,11)=16.44, Pb0.002). However, Task

and Study/Test pairing interacted ( F (1,11)=17.06,

Pb0.002). Post-hoc testing indicated that the difference

between Old same and Old diff in the item task failed to

reach significance (although in the expected direction),

whereas the difference between Target and Nontarget RTs

was reliable in the source task.

A repeated measures ANOVA with the factors of Task

(item, source) and Old/New (target RTs were used for the

source task) revealed that RTs in the source test were longer

than in the item task (F(1,11)=20.60, Pb0.001) and that RTs

to new items were faster than those to old items

(F(1,11)=6.97, Pb0.02). However, the two main effects

interacted (F(1,11)=43.90, Pb0.0001). Post-hoc testing

indicated that old and new RTs did not differ significantly

for the item task, while target RTs were reliably slower than

their new counterparts in the source task.

3.1.2. Percent correct and sensitivity

This analysis assessed the effect of Task (item, source)

and Study/Test color pairing (same, different) on the

percentage of correct responses. The percentage of correct

responses was larger in the item than the source task

(F(1,11)=32.10, Pb0.0001). For both tasks, the percentage

of correct responses to Old same/Target items was reliably

greater than Old diff/Nontarget items (F(1,11)=5.50,

Pb0.04; interaction F(1,11)=1.90, PN0.10).

Estimates of the subjects’ ability to discriminate old from

new items (Pr) along with response bias (Br) were

computed according to the methods described by Snodgrass

and Corwin [26], and are shown in Table 1. In the source

task, these indices were computed in two ways: using the

false alarm rate associated with new and nontarget items.

The 3 Pr values listed in Table 1 were all reliably different

from zero, as assessed by t-tests (item Pr t(11)=33.8,

Pb0.0001; source Pr based on the new FA rate t(11)=19.6,

Pb0.0001; source Pr based on the nontarget FA rate

t(11)=9.3, Pb0.0001). Using the false alarm rate to new

items, Pr was larger in the item than the source task

(F(1,11)=5.40, Pb0.04), and responding was more liberal

in the source than the item task (F(1,11)=6.11, Pb0.03).

Using the false alarm rate to nontarget items, Pr was again

larger in the item than the source task (F(1,11)=98.70,

Pb0.0001), and responding was still more liberal in the

source task (F(1,11)=55.03, Pb0.0001). A comparison of

the two Pr values in the source task revealed greater

sensitivity in detecting targets relative to new items

compared to detecting targets relative to nontargets

(F(1,11)=93.02, Pb0.00001).

In sum, accuracy was lower and RTs were longer in the

source task. Regardless of which false alarm rate was used,

participants were able to discriminate among the three

classes of stimuli in the source task. Changing the picture’s

outline color between study and test led to lower accuracy

and longer RTs in both item and source tasks although, for

RTs, this effect was reliable only in the source task.

3.2. ERP Data

Fig. 1 depicts the grand mean ERP data for both the item

and source tasks, categorized according to the conjunction

of the color in which the item was presented at study and

test. For the item task, the ERPs associated with correctly

rejected new and correctly recognized Old same and Old

diff items are depicted in the left panel of Fig. 1. Similarly,

for the source task, in the right panel of Fig. 1, the ERPs

associated with correctly recognized Target and Nontarget

trials are depicted along with the ERPs elicited by correctly

rejected new items. The mean number (FS.D.) of trials

comprising each of the averages depicted in Fig. 1 are

presented in Table 2.

Fig. 1 indicates that the waveforms from both the item

and source tasks are characterized by parietal EM effects

(maximal at ~500–600 ms) that do not appear to differ in

magnitude. Second, the presence of late negative activity

appears to reduce the magnitude of the parietal EM effect,

resulting in a large magnitude late positivity (~700 ms)

associated with correctly rejected new items in both tasks.

Third, while a late negative EM effect (between about 800

and 1000 ms) appears to be present in the ERPs of the item

task, this activity is much larger in the source task.

For the item task, Fig. 1 suggests that neither the parietal

nor the late negative EM effects differs according to whether

the test picture was painted in the same or the alternate color

as at study. By contrast, there is a suggestion in the source

task data that, for the parietal EM effect, pictures painted in

the same color as at study elicit a somewhat larger EM effect

than those painted in the alternate color. In similar fashion to

the item task, the late negative EM effect does not appear to

differ according to the conjunction of study and test color

pairing. The mean RT marks at the top of the figure

(depicted below the FPz scalp site) demonstrate that, for

both the item and source tasks, RT activity occurs during the

latency window of the late negative EM effects for the two

classes of old items. There is also the suggestion of a right

prefrontal EM effect in the source task data only at the FP2

electrode site, which is absent at the FP1 site. At this

location, targets appear to elicit more positive-going activity

than either new or nontarget items, beginning quite early

(~300–400 ms) and lasting until the end of the recording

epoch.

Fig. 2 depicts, from a reduced number of scalp sites, the

old–new difference waveforms for the item and source tasks

computed using the data depicted in Fig. 1. It is clear that

the parietal EM effects elicited in both tasks are highly

similar. Both item and source task waveforms are charac-

terized by late negative EM effects (~800–1000 ms), which

are larger in the source task.

The statistical analyses of effects identified in Fig. 1

were performed on a series of averaged voltages

Fig. 1. Grand mean ERPs associated with correctly recognized Old same and Old diff trials and correctly rejected new items in the item task (left panel). For the

source task (right panel), the target (same color) and nontarget (different color) ERPs are depicted, along with the ERPs associated with correctly rejected new

items. Arrows mark stimulus onset with time lines every 500 ms. For the item task, solid lines indicate the ERPs to correctly recognized old same (painted in

the same color at study and test), dashed lines represent the ERPs associated with correctly recognized old diff items (painted in different colors at study and

test), and the dotted lines represent correctly rejected new items. For the source task, solid lines represent the target ERPs, dashed lines the nontarget ERPs, and

dotted lines the ERPs associated with correctly rejected new items. Vertical bars below the FPz scalp site mark mean RT, with horizontal bars indicating the

S.D. of the RT distribution. DIFF=different.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198 191

beginning at 300 ms and ending at 1000 ms post-

stimulus. The ANOVAs were performed separately for

averaged voltages encompassing an early region of the

waveform (often identified with familiarity), the parietal,

late negative and right prefrontal EM effects, i.e., 300–

500, 500–700, 800–1000 and 1200–1400 ms. The first

set of ANOVAs contrasted the EM effects between the

item and source tasks based on the data depicted in Fig.

1. The second series of ANOVAs was concerned with

determining whether the late negative EM effects

Table 2

Mean number (FS.D.) of the numbers of trials comprising the ERP

averages

Task Old same/

Target

Old diff/

Nontarget

New Target

incorrect

Nontarget

incorrect

Item

task

29 (4) 27 (2.5) 75 (6)

Source

task

41 (7) 35 (7) 32 (2.5) 15 (2.5) 21 (3.2)

associated with correct and incorrect source judgments

differed in magnitude.

3.2.1. Item versus source tasks (Same color, different color,

new pictures)

The first ANOVA involved the within-subjects factors of

Task (item, source), Condition (same color, different color,

new), Laterality (three levels), and Anterior Posterior

dimension (eight levels). For the 300–500 ms window, the

Task main effect was not reliable (Fb1), but the Condition

main effect was significant (F(2,22)=3.62, Pb0.04). Post-

hoc testing showed that, across item and source tasks, old

same/Target items and old diff/Nontarget items elicited

greater positivity than new items, indicating robust EM

effects. However, the amplitude of the 300–500 ms averaged

voltage did not differ between Old same/Target and Old diff/

nontarget items. The Task by Condition by Laterality

interaction was reliable (F(4,44)=3.10, Pb0.05, e=0.56).As revealed by post-hoc testing, this interaction resulted

because Target and Nontarget items both produced larger

amplitudes than their new counterparts, but did not differ

Fig. 2. Grand mean difference waveforms (old minus new) based on the data depicted in Fig. 1. Arrows mark stimulus onset with time lines every 500 ms.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198192

from each other; for target items (same color) this effect was

robust at left, midline and right sites, whereas for nontarget

items (different color), this effect was only reliable at left-

sided locations. By contrast, in the item task, only Old diff

items produced reliably larger amplitudes than new items at

midline and right scalp locations.

For the 500–700 ms region, only the Task by

Condition by Anterior/Posterior interaction was significant

(F(14,154)=4.33, Pb0.01, e=0.21). As revealed by post-

hoc testing, none of the differences among the three

event types for the item task was reliable. However, for

the source task, the averaged voltages associated with

Target items were significantly larger than those to new

items at fronto-polar sites, whereas the reverse was the

case at parietal, parieto-occipital and occipital scalp

regions. New items were larger than nontarget items,

but only at parietal, parieto-occipital and occipital scalp

regions. These effects are most likely due to the overlap

of the large positivity associated with new items and the

negative-going activity associated with target and non-

target old items (Figs. 1 and 2).

In the region of the late negative EM effect (800–1000

ms), the Condition main effect was significant

(F(2,22)=31.95, Pb0.0001, e=0.95). Post-hoc testing

indicated that the ERPs associated with Old same/Targets

and Old diff/Nontargets did not differ, but both were

larger (i.e., more negative) than the corresponding new

items, revealing robust EM effects. However, this main

effect was modified by the Task by Condition by

Anterior/Posterior interaction (F(14,154)=5.27, Pb0.003,

e=0.24). As indicated by post-hoc assessment, at anterior

scalp sites, Old same/Targets and Old diff/Nontargets did

not differ from each other for either the item or source

tasks. However, at posterior scalp sites, Old same/Targets

and Old diff/Nontargets differed from their new counter-

parts for both item and source tasks. In addition, with the

exception of new items, which did not differ between

item and source tasks, source task amplitudes were more

negative than their item task counterparts at centro-

parietal and parieto-occipital scalp regions.

An additional analysis was performed only on the data

recorded from the fronto-polar electrode sites to deter-

mine whether right prefrontal activity was present and, if

so, to assess whether it differed between item and source

tasks. The fronto-polar locations were those where the

data were least overlapped by late negative activity (Figs.

1 and 2). The Task (item, source) by Condition (same

color, different color, new) by Electrode Location (FP1,

FPz, and FP2) ANOVA revealed a Task by Condition by

Electrode Location interaction (F(4,44)=3.86, Pb0.02,

e=0.75). Assessment via post-hoc testing indicated that,

for the item task, Old diff items showed greater negative-

going activity than new items, but only at the FP1 scalp

location. By contrast, for the source task, Targets were

associated with greater positive activity than nontargets

and new items. These differences, however, were only

reliable at the FP2 scalp site. These latter results appear

to indicate the presence of a right prefrontal EM effect,

but only in the source task, as depicted in Fig. 3.

3.3. Correct versus incorrect source judgments

Fig. 4 depicts, at six locations representing anterior

and posterior scalp, the source task difference mean ERPs

associated with old items whose sources were and were

not correctly attributed (Target correct-new, Target incor-

rect-new, Nontarget correct-new, Nontarget incorrect-

new). The mean (FS.D.) of the number of sweeps

associated with the incorrect waveforms are presented in

Table 2. The data in Fig. 4 indicate that the late negative

EM effect (and, to some extent, the parietal EM effect)

show similar magnitudes whether or not the source had

Fig. 3. Grand mean averaged voltages at the FP1, FPz, and FP2 scalp sites computed on the ERPs associated with same study-test color pairings (Old same/

Target), different study-test color pairings (Old diff/Nontarget) and new items in both item and source tasks. Asterisks indicate the electrodes at which the

conditions within the brackets differed reliably. DIFF=different.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198 193

been judged correctly. By contrast, there is a suggestion

that, for targets, correct judgments lead to larger

amplitudes of the activity between 300 and 700 ms,

whereas for nontargets, incorrect judgments lead to larger

amplitudes than correct judgments.

To determine statistically if the magnitudes of the

parietal and late negative EM effects differed reliably for

successful compared to unsuccessful source retrieval, the

300–500, 500–700, 800–1000, and 1200–1400 ms

regions were analyzed in Target/Nontarget by Correctness

(correct, incorrect source) by Laterality by Anterior/

Posterior ANOVAs using the difference waveforms (Fig.

4). For all four averaged voltage windows, none of the

main or interaction effects involving the Target/Nontarget

or Correctness factors was reliable (Fsb2.50, PsN0.10).

Fig. 4. Grand mean difference waveforms in the Source Task associated with correc

every 500 ms. Solid lines indicate the ERPs associated with correct target (left pan

the ERPs associated with incorrect target (left panel) and incorrect nontarget (rig

3.4. Comparison of nose- and mastoid-referenced data

Fig. 5 depicts the Fig. 1 data re-referred to a linked

mastoid reference. The major difference between the

nose- and mastoid-referred data appears to be the

presence of a right prefrontal EM effect, which is larger

in the source compared to the item task, especially for

the ERPs associated with Target trials. As pointed out

earlier, re-inspection of Fig. 1 also demonstrates the

presence of this right prefrontal EM, although it does not

appear as prominent as in the mastoid-referenced ERP

waveforms. Fig. 6 illustrates a more direct comparison of

the two sets of data using the difference means at

prefrontal and centro-parietal electrode sites. Overall, the

mastoid-referred data are characterized by more positive-

t and incorrect source decisions. Arrows mark stimulus onset with time lines

el) and correct nontarget (right panel) old decisions. Dashed lines represent

ht panel) decisions.

Fig. 5. Grand mean ERP data from Fig. 1 re-referenced to linked mastoids. Time lines, markers and line types are the same as in Fig. 1. DIFF=different.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198194

going activity over prefrontal scalp sites and less

negative-going activity at posterior scalp sites. However,

the Task and Condition effects do not appear to differ

markedly in the two data sets.

3.5. Summary of ERP findings

To summarize, the late negative EM effect was

larger in the source task, whereas there were very

slight differences in the parietal EM effect between the

item and source retrieval conditions, consistent with

previous findings [5]. The ERPs associated with same-

colored or differently colored old pictures did not

differ reliably for either task for any of the measured

regions of the waveforms. In the source task, both the

parietal and late negative EM effects were larger to

old than new items, thus demonstrating robust EM

effects. Some evidence for the presence of a right

prefrontal EM effect was obtained and this activity was

only observed in the source task. The late negative

EM effect did not differ between trials associated with

correct versus incorrect source judgments, consistent

with previous investigations from this laboratory [5].

This latter finding suggests that the negative EM effect

does not reflect successful retrieval.

4. Discussion

The results of the current study replicate well two

previous findings: (1) greater magnitude late negative

activity in the source compared to the item task; and (2)

very little difference between tasks in the magnitude of the

parietal EM effect. The current data suggest that the

negative EM effect reflects the search for and/or retrieval/

evaluation of source-specifying information, although what

is searched for, retrieved and/or evaluated does not

necessarily have to be specific, color features (see below).

Nonetheless, the parieto-occipital scalp distribution of the

negative EM effect suggests that the computations involved

may be modality-specific (i.e., localized to visual cortical

regions). This interpretation is consistent with the fact that

the parieto-occipital EM effect was present even in those

ERPs that were associated with trials on which an incorrect

source decision had been made (i.e., target misses and

nontarget false alarms—see Fig. 4), joining the results of a

previous study [5]. That is, as for the right prefrontal EM

effect observed in other investigations [7], it is unlikely that

the late negative EM effect reflects successful source

retrieval [11,17].

It is also possible that, in situations in which perceptual

details do not change between study and test, recollection

Fig. 6. Grand mean difference waveforms for the nose-referred (top two rows) and mastoid-referred (bottom two rows) data. The item task ERPs are shown on

the left and the source task ERPs on the right. Solid lines represent the ERPs associated with correctly recognized old items that were painted in the same color

at study and test (Old same/Target). Dashed line ERPs represent correctly recognized old items that were painted in different colors at study and test (Old diff/

Nontarget). Arrows and time lines the same as in Fig. 5.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198 195

(i.e., recovery of source attributes) is not the only basis for

making source judgments. Perceptual bfluency,Q which is

thought to influence implicit or indirect memory perform-

ance (e.g., repetition priming), might also play a role [9].

Performance on indirect tests of memory is usually

enhanced significantly when perceptual features do not

change from study to test because of bfluent reprocessingQ ofthe perceptual attributes of stimuli that have been recently

presented [31]. For example, Kelley et al. [15] assessed

memory for modality directly (old/new recognition) and

indirectly (perceptual identification). Subjects read or heard

words in mixed encoding lists and were then tested only

with visually presented old and new items. In experiment 1,

during perceptual identification and recognition memory

tests (did you read or hear the word or is it new?), subjects

were reliably more accurate when items were presented in

the same modality (i.e., read) as at study. The dependence

between the indirect and direct tests of memory in their

experiment 1 led Kelley et al. [15] to conclude that the

bperceptual fluencyQ between items studied and tested in the

same format was a source of information which subjects

were able to use (whether consciously or not) in order to

make judgments of modality in the direct recognition task.

Similarly, in the current experimental situation, it could be

hypothesized that, in the case where study and test

perceptual details match (i.e., target old items), subjects’

source judgments could have been based on an indirect

repetition priming mechanism (through fluency) and not

only on recollection of contextual information. As in Kelley

et al. [15], this hypothesis receives some support from the

finding that in both the item recognition and source retrieval

tasks, participants were more accurate when the object’s

studied color and tested color matched compared to when

they mismatched. On the other hand, the ERP data do not

appear to support this type of indirect mechanism. There

were no reliable differences for any of the measured regions

of the waveforms between identically and differently

colored old pictures either in the item or source tasks.

Hence, although we cannot rule out the possibility of a

bfluency-basedQ influence, the current data are equivocal

with respect to this interpretation.

We have previously hypothesized that the negative EM

effect reflects a modality-specific search for and/or

retrieval/evaluation of the recovered attributes. This may

be a viable interpretation, but must be considered

tentative, as experiments that test this notion directly

have not yet been performed. Nonetheless, some support

for this hypothesis comes from recent studies by

Johansson et al. [12] and Leynes et al. [16], whose

studies were described in the Introduction. In both

investigations, source retrieval demands required partic-

ipants to perform a large amount of visual processing

(i.e., retrieve whether the object was viewed or imagined

in Johansson et al.; planned or performed in Leynes et

al.). The posterior topography of the negative EM effects

in those studies is consistent with the scalp recorded data

Fig. 7. Grand mean RT-locked ERPs in the source task for correctly

recognized target (solid lines), nontarget items (long dashed lines), and

correctly rejected new items (dotted lines). The averages were computed for

200 ms pre- and 1000 ms post-RT intervals at eight midline electrode sites.

Arrow marks mean reaction time, with time lines every 500 ms.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198196

reflecting the output of generators in visual cortical

regions.

As mentioned earlier, Johansson and Mecklinger [11]

have suggested that the late negative EM effect may not be a

unitary phenomenon but, depending on task demands, may

be manifested by two different components. The first,

present in response-locked averages, is thought to reflect

baction monitoringQ in situations where there is a good deal

of response conflict as, for example, when participants must

choose between highly similar, non-studied, blureQ items

and genuinely old, previously studied events [21]. Hence,

this negative EM effect could occur in relatively simple yes/

no recognition memory paradigms that engender some

degree of response conflict. The RT-locked component has

a posterior distribution similar to its stimulus-elicited

counterpart. The second negative EM component, present

in stimulus-locked averages, is thought to reflect, as stated

earlier, the search for and/or retrieval/evaluation of attribute

conjunctions (e.g., the object and its color as in the current

experiment). This EM effect appears to be synonymous with

the late negative EM effect that is the subject of the current

investigation. It has a parieto-occipital scalp focus.

Based on the hypothesis suggested by Johansson and

Mecklinger [11], response-locked averaging in the current

situation should yield a larger amplitude negative compo-

nent in the nontarget compared to the target ERPs. This is

so because, in the case of nontargets, one hand is mapped

onto two responses (old items in the alternate color and

new items), whereas the other hand is associated with only

a single response, that of the target (same color). This may

have engendered a greater amount of response conflict in

the former situation, requiring greater action monitoring.

However, when RT-locked averages were computed

(Fig. 7), negative activity focused over parieto-occipital

scalp was not present in the ERPs associated with either

targets or nontargets.

Nessler and Mecklinger [21] observed negative EM

effects in both stimulus- and response-locked averages

during a false memory paradigm, in which high levels of

response conflict may have been due to the large familiarity

signals generated by bnewQ items that were related

semantically to previously studied words. Hence, the degree

of response conflict generated in the current situation may

not have been as marked as that in the Nessler and

Mecklinger study [21]. Therefore, to the extent that the

negative EM effect in RT-locked averages reflects response-

related conflict [11], the data reported here suggest that

these types of processes do not appear to modulate the ERP

waveforms. Alternatively, given the paucity of data that can

speak to this issue, it is also possible that the negative EM

effect observed in RT-locked averages reflects other, as yet

undiscovered, aspects of cognition.

By contrast with previous investigations from this

laboratory [4,6], the negative activity was fairly substantial

in the item recognition task, although reliably smaller than

in the source task. Further, unlike the data from previous

studies in this series, the late negative activity associated

with correctly recognized old objects was significantly

larger than that to new objects in the item recognition task.

Although care was taken to randomize the presentation of

item and source tests, once subjects had participated in a

source test, some of the strategies used to perform in that

condition may have carried over to the item recognition

task. By this account, on seeing an object painted in either

red or green, rather than simply responding old or new,

participants may have attempted to think back to the original

episode to determine if the object and its color recapitulated

the study phase event. Hence, if the hypothesis advanced

here is valid, a search for and/or evaluation of the bsourceQmay have led to the presence of this negative activity even

in the item recognition task, which should not have required

any contextual retrieval.

As described throughout this discussion section, there are

at least two potential explanations for the presence of the

late negative EM effect in the current data: (1) search for

and/or retrieval/evaluation of specific, color information in

modality-specific cortical areas, and (2) retrieval of bboundQinformation. Perhaps the strongest evidence for the first

possibility is the finding that, in all of our previous

investigations in which the negative EM effect was

recorded, color was the diagnostic source attribute. More-

over, in those studies, the negative EM effect was markedly

larger in source compared to item memory blocks. In

addition, the scalp topography of this EM effect has been

observed consistently to be focused over parieto-occipital

scalp, concordant with modality-specific processing. Con-

trary to this argument, however, are the observations of

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198 197

putatively similar late, parieto-occipital, negative EM effects

associated with non-pictorial stimuli whose source attributes

were strikingly different (and not necessarily bvisualQ)compared to the one used here, e.g., temporal list member-

ship, gender of voice and encoding task [32,33]. Similarly,

the finding that the amplitude of the negative EM effect did

not differ between the ERPs associated with colors that were

or were not identical between study and test appears to run

counter to the hypothesis that the negative EM effect reflects

the search for and/or retrieval of specific, perceptual

attributes (i.e., color). Rather, it may reflect the search for

and/or retrieval/evaluation of more general, source-specify-

ing information. Moreover, the search and/or retrieval/

evaluation might not necessarily occur for attributes

specified by the experimenter, i.e., bdiagnosticQ [20]. Thatis, for example, non-diagnostic, idiosyncratic attributes

might also be retrieved and evaluated.

However, as mentioned, the attributes retrieved would not

necessarily have to be the correct ones, as indicated by a lack

of amplitude difference between the ERPs associated with

correct and incorrect source judgments in the stimulus-locked

data. Presumably, whether or not these attributes matched, a

search and/or retrieval/evaluation would still be required to

support ongoing memory performance. However, the finding

that the correct and incorrect waveforms did not differ in

magnitude should be viewed with caution, as the trial counts

associated with the incorrect waveforms were low and may

have impacted negatively the power of this analysis.

The possibility that the negative EM effect could reflect

the retrieval of source-specifying attributes might be sup-

ported further by the temporal relation between the onset of

the negative EM effect (~500 ms) and mean reaction time

(~1000 ms). That is, sufficient time would have elapsed

between the onset of the negative EM effect and RTso that the

negative EM effect could be the brain event reflecting the

subject’s decision as to whether a target or nontarget had been

retrieved. Hence, the negative EM effect could have been

causally related to the reaction time response.

The second possibility, suggested by Johansson and

Mecklinger [11], is the retrieval of bound attributes when

the recovery of those attributes is difficult or continues to

require evaluation. In the current design, these bboundQfeatures might be the picture itself and the color in which it

was painted during the initial learning episode. Because the

design used here did not require the overt retrieval of

attribute conjunctions, the current data cannot speak directly

to this issue. On the other hand, we have previously

suggested that one strategy for retrieving the color of a

previously presented item might be to image the conjunction

of the object and its associated color to determine if a match

had occurred [5]. As suggested in the Introduction, it might

be more difficult to retrieve this information when the test

object’s color does not match its studied counterpart. In the

Cycowicz et al. [5] investigation, all test items were

presented in black outline; hence, no color cue was available

with which the participant could have attempted to match

the stored memory trace. In the current study, by contrast,

objects were presented either in the same or alternate color

as they were painted in the study phase, thereby providing,

at least in the case of a target, an additional cue for matching

the test item with a stored representation. However, the late,

negative EM effect was not larger in the condition that

ostensibly engendered greater retrieval and/or evaluative

demands, even though that condition resulted in lower

performance. These findings appear to run counter to the

suggestion made by Johansson and Mecklinger [11], that the

late negative EM effect could reflect the retrieval of attribute

conjunctions when they are difficult to recover or neces-

sitate continued evaluation. However, these differing inter-

pretations remain highly speculative in the absence of

experiments that are designed specifically to test them.

Johansson and Mecklinger [11] have further suggested

that in order for the attribute conjunctions to be formed

and stored, a sensory specific search function would most

likely be needed, the products of which would be

bboundQ to the recognized item. As previously discussed,

we have also suggested that the negative EM effect could

reflect a search for and/or retrieval/evaluation of the

attributes in modality-specific cortical regions that pro-

cessed the information during the study phase [3,5].

However, the modality specificity of the late negative EM

effect has, to our knowledge, never been assessed directly.

Such research is sorely needed before this hypothesis can

be entertained further.

Acknowledgments

The authors thank Mr. Charles L. Brown, III for

computer programming and technical assistance, Ms.

Letecia Latif for subject recruitment, and the volunteers

for generously giving their time. This study was supported

in part by grant HD14959 from NICHD, and by the New

York State Department of Mental Hygiene.

References

[1] S. Berman, D. Friedman, M. Hamberger, J.G. Snodgrass, Devel-

opmental picture norms: relationships between name agreement,

familiarity and visual complexity for child and adult ratings of two

sets of line drawings, Behavior Research Methods, Instruments, &

Computers 21371–382 (1989) 371–382.

[2] T. Curran, Brain potentials of recollection and familiarity, Memory &

Cognition 28 (2000) 923–938.

[3] Y.M. Cycowicz, D. Friedman, Source memory for the color of

pictures: event-related brain potentials (ERPs) reveal sensory-specific

retrieval-related activity, Psychophysiology 40 (2003) 455–464.

[4] Y.M. Cycowicz, D. Friedman, M. Rothstein, J.G. Snodgrass, Picture

naming by young children: norms for name agreement, familiarity,

and visual complexity, Journal of Experimental Child Psychology 65

(1997) 171–237.

[5] Y.M. Cycowicz, D. Friedman, J.G. Snodgrass, Remembering the color

of objects: an ERP investigation of source memory, Cerebral Cortex

11 (2001) 322–334.

D. Friedman et al. / Cognitive Brain Research 23 (2005) 185–198198

[6] M.J. Farah, F. Peronnet, M.A. Gonon, M.H. Giard, Electrophysio-

logical evidence for a shared representational medium for visual

images and visual percepts, Journal of Experimental Psychology.

General 117 (3) (1988) 248–257.

[7] D. Friedman, R. Johnson, Event-related potential (ERP) studies of

memory encoding and retrieval: a selective review, Microscopy

Research and Technique 51 (2000) 6–28.

[8] G. Gratton, M.G.H. Coles, E. Donchin, A new method for off-line

removal of ocular artifact, Electroencephalography and Clinical

Neurophysiology 55 (1983) 468–484.

[9] L.L. Jacoby, M. Dallas, On the relationship between autobiographical

memory and perceptual learning, Journal of Experimental Psychology.

General 110 (1981) 306–340.

[10] J.R. Jennings, C.C. Wood, The e-adjustment procedure for repeated

measures analyses of variance, Psychophysiology 13 (1976) 277–278.

[11] M. Johansson, A. Mecklinger, The late posterior negativity in ERP

studies of episodic memory: a selective review and a functional

account, Biological Psychology 64 (2003) 91–117.

[12] M. Johansson, G. Stenberg, M. Lindgren, I. Rosen, Memory for

perceived and imagined pictures—an event-related potential study,

Neuropsychologia 40 (2002) 986–1002.

[13] R. Johnson, Event-related potential insights into the neurobiology of

memory systems, in: F. Boller, J. Grafman (Eds.), Handbook of

Neuropsychology, vol. 10, Elsevier, Amsterdam, 1995, pp. 135–163.

[14] R. Johnson Jr., K. Kreiter, B. Russo, J. Zhu, A spatio-temporal

analysis of recognition-related event-related brain potentials, Interna-

tional Journal of Psychophysiology 29 (1998) 83–104.

[15] C.M. Kelley, L.L. Jacoby, A. Hollingshead, Direct versus indirect tests

of memory for source: judgements of modality, Journal of Experimental

Psychology. Learning, Memory, and Cognition 15 (1989) 1101–1108.

[16] P. Leynes, M. Bink, Did I do that? An ERP study of memory for

performed and planned actions, International Journal of Psychophy-

siology 45 (2002) 197.

[17] J. Li, A.M. Morcom, M.D. Rugg, The effects of age on the neural

correlates of successful episodic retrieval: an ERP study, Cognitive

and Affective Behavioral Neuroscience 4 (3) 279–293.

[18] G. Mandler, Recognizing: the judgement of previous occurrence,

Psychological Review 87 (1980) 252–271.

[19] A. Mecklinger, Interfacing mind and brain: a neurocognitive model of

recognition memory, Psychophysiology 37 (2000) 565–582.

[20] N.W. Mulligan, E. Hirshman, Measuring the bases of recognition

memory: an investigation of the process-dissociation framework,

Journal of Experimental Psychology. Learning, Memory, and Cogni-

tion 23 (1997) 280–304.

[21] D. Nessler, A. Mecklinger, ERP correlates of true and false

recognition after different retention delays: stimulus- and response-

related processes, Psychophysiology 40 (2003) 146–159.

[22] D. Nessler, A. Mecklinger, T.B. Penney, Event related brain potentials

and illusory memories: the effects of differential encoding, Cognitive

Brain Research 10 (2001) 283–301.

[23] M.R. Nuwer, D. Lehmann, F. Lopes da Silva, S. Matsuoka, M.

Sutherling, J.F. Vibert, IFCN guidelines for topographic and

frequency analysis of EEGs and EPs. Report of an IFCN

committee. International Federation of Clinical Neurophysiology,

Electroencephalography and Clinical Neurophysiology 91 (1994)

1–5.

[24] K.A. Paller, Neurocognitive foundations of human memory, in: D.L.

Medin (Ed.), The Psychology of Learning and Motivation, vol. 40,

Academic Press, San Diego, 2001, pp. 121–145.

[25] M.D. Rugg, E.L. Wilding, Retrieval processing and episodic memory,

Trends in Cognitive Sciences 4 (2000) 108–115.

[26] J.G. Snodgrass, J. Corwin, Pragmatics of measuring recognition

memory: applications to dementia and amnesia, Journal of Exper-

imental Psychology. General 117 (1988) 34–50.

[27] J.G. Snodgrass, M. Vanderwart, A standardized set of 260 pictures:

norms for name agreement, image agreement, familiarity and visual

complexity, Journal of Experimental Psychology. Human Learning

and Memory 6 (1980) 174–215.

[28] C.T. Trott, D. Friedman, W. Ritter, M. Fabiani, J.G. Snodgrass,

Episodic priming and memory for temporal source: event-related

potentials reveal age-related differences in prefrontal functioning,

Psychology and Aging 14 (1999) 390–413.

[29] D. Tsivilis, L.J. Otten, M.D. Rugg, Context effects on the neural

correlates of recognition memory. An electrophysiological study,

Neuron 31 (2001) 497–505.

[30] E. Tulving, Memory and consciousness, Canadian Psychologist 26

(1985) 1–12.

[31] A.D. Wagner, J.D. Gabrieli, On the relationship between

recognition familiarity and perceptual fluency: evidence for

distinct mnemonic processes, Acta Psychologica 98 (1998)

211–230.

[32] D.J. Wegesin, D. Friedman, N. Varughese, Y. Stern, Age-related

changes in source memory retrieval: an ERP replication and

extension, Cognitive Brain Research 13 (2002) 323–338.

[33] E.L. Wilding, Separating retrieval strategies from retrieval success: an

event-related potential study of source memory, Neuropsychologia 37

(1999) 441–454.

[34] E.L. Wilding, In what way does the parietal ERP old/new effect index

recollection? International Journal of Psychophysiology 35 (2000)

81–87.

[35] E.L. Wilding, M.D. Rugg, An event-related potential study of

recognition memory with and without retrieval of source, Brain 119

(1996) 889–905.

[36] E.L. Wilding, H. Sharpe, Episodic memory encoding and retrieval:

recent insights from event-related postentials, in: A. Zani, A.M.

Proverbio (Eds.), The Cognitive Electrophysiology of Mind and

Brain, Academic Press, New York, 2002, pp. 169–196.

[37] G. Yovel, K.A. Paller, The neural basis of the butcher-on-the-bus

phenomenon: when a face seems familiar but is not remembered,

NeuroImage 21 (2004) 789–800.