Indexing strategic retrieval of colour information with event-related potentials

www.elsevier.com/locate/cogbrainres

Cognitive Brain Researc

Research Report

Indexing strategic retrieval of colour information with

event-related potentials

E.L. Wilding*, C.S. Fraser, J.E. Herron

School of Psychology, Cardiff University, Cardiff, CF10 3AT Wales, UK

Accepted 9 April 2005

Available online 31 May 2005

Abstract

Event-related potentials (ERPs) were acquired during two experiments in order to determine boundary conditions for when recollection of

colour information can be controlled strategically. In initial encoding phases, participants saw an equal number of words presented in red or

green. In subsequent retrieval phases, all words were shown in white. Participants were asked to endorse old words that had been shown at

encoding in one colour (targets), and to reject new test words as well as old words shown in the alternate colour (non-targets). Study and test

lists were longer in Experiment 1, and as a result, the accuracy of memory judgments was superior in Experiment 2. The left-parietal ERP

old/new effect—the electrophysiological signature of recollection—was reliable for targets in both experiments, and reliable for non-targets

in Experiment 1 only. These findings are consistent with the view that participants were able to restrict recollection to targets in Experiment 2,

while recollecting information about targets as well as non-targets in Experiment 1. The fact that this selective strategy was implemented in

Experiment 2 despite the close correspondence between the kinds of information associated with targets and non-targets indicates that

participants were able to exert considerable control over the conditions under which recollection of task-relevant information occurred.

D 2005 Elsevier B.V. All rights reserved.

Theme: Neural basis of behaviour

Topic: Learning and memory systems and functions

Keywords: Event-related potential; Colour information; Episodic memory; Recollection; Exclusion task

1. Introduction

Studies of retrieval from episodic memory in which

event-related potentials (ERPs) were recorded have led to

the identification of several modulations of scalp-recorded

neural activity that index distinct classes of retrieval process

(for reviews, see [16,43,53]). The most common contrast in

ERP studies of memory retrieval is between the electrical

activity that is evoked by old (previously studied) and new

items to which correct memory judgments have been made.

The differences revealed by these contrasts will be referred

to here as ERP old/new effects.

0926-6410/$ - see front matter D 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.cogbrainres.2005.04.012

* Corresponding author. Fax: +44 2920874858.

E-mail address: [email protected] (E.L. Wilding).

URL: http://www.cardiff.ac.uk/psych/home/wildinge/index.html

(E.L. Wilding).

There is a family of old/new effects, and the individual

effects are distinguishable on the basis of their time courses,

scalp distributions, and sensitivity to experimental variables

[11]. The old/new effect to which probably the most attention

has been paid is largest at parietal electrode locations,

typically larger over the left than the right hemisphere, and

comprises a relatively greater positivity for old than for new

test items to which correct judgments have been made [37].

This left-parietal ERP old/new effect onsets approximately

500 ms post-stimulus, and commonly lasts between 200 and

700 ms, the duration being influenced by task demands [10].

It is now widely accepted that the left-parietal ERP old/

new effect indexes recollection (for recent discussion, see

[5,15]). Recollection is defined here as successful recovery

of contextual information about a prior event [56], which

might include one or more of information sources such as

where or when the event occurred, the modality through

h 25 (2005) 19 – 32

E.L. Wilding et al. / Cognitive Brain Research 25 (2005) 19–3220

which information was perceived, and/or what actions were

completed during the event [28].

The data that support the link between the parietal effect

and recollection include the following: (1) patients with

deficits in recollection (as revealed by behavioural mea-

sures) exhibit attenuated or absent left-parietal ERP old/new

effects relative to appropriately matched controls [13,39,

46,47]; (2) the magnitude of the effect is correlated

positively with the amount of contextual information that

is retrieved, as inferred from the numbers of correct source

judgments that are made on retrieval tasks [50,51]; and (3)

the effect is larger for correct old judgments associated with

Remember (R) rather than with Know (K) judgments

[12,45]. R judgments are held to be based upon recollection

while K judgments are not [13].

While this brief summary provides arguably a compelling

case for a link between the parietal old/new effect and

recollection, some recent findings, at least at a first pass, jar

with this account. These findings have come from exclusion

tasks [23,24], in which ERPs were acquired time-locked to

test stimuli. In the study phases of exclusion tasks, items are

typically associated with one of two contexts. In subsequent

retrieval phases, items associated with one context are

designated as targets, while items from the other are

designated as non-targets. Participants are asked to respond

on one key to targets, and on another to non-targets as well as

to genuinely new test items. The exclusion task was

introduced as one component of the process-dissociation

procedure (PDP; [23–25]), and while the details of that

procedure are not important here, the critical point for present

purposes is that accurate target/non-target judgments have

been assumed to be based upon recollection of information

associated with targets and with non-targets [23].

If this assumption is correct, then correct target as well as

non-target judgments will be accompanied by reliable left-

parietal ERP old/new effects. In one recent study, however,

Herron and Rugg [20] reported a somewhat different pattern

of effects. They acquired ERPs during the retrieval phases

of two exclusion tasks. For both experiments, the first

encoding phase involved generating sentences incorporating

single words that were presented at study, while the second

encoding phase differed across the two experiments. In the

first experiment, participants rated words for pleasantness,

while in the second, participants were asked simply to read

words aloud. Targets were designated as words that were

rated for pleasantness in Experiment 1, and words that were

read aloud in Experiment 2. Thus, the non-targets were

associated with the same encoding task in both experiments,

and while reliable parietal ERP old/new effects were obtained

for targets in both experiments, for non-targets, a reliable left-

parietal effect was revealed in Experiment 2 only.

This finding is striking, because in both experiments,

the information associated with non-targets is likely to be

highly similar, since they were encoded using the same

task (sentence generation), and in keeping with this

observation, the accuracy of non-target judgments was

equivalent in the two experiments. To the extent that

participants were equally likely to have recollected non-

targets in both experiments, therefore, the absence of a

left-parietal ERP old/new effect in Experiment 1 might be

seen to call into question the link between the parietal ERP

old/new effect and recollection (see also [21]).

Whether these results challenge current views of the

functional significance of the left-parietal ERP old/new

effect, however, depends upon the assumption that partici-

pants did in fact recollect information about non-targets

equally often in the two experiments. Herron and Rugg

[20] questioned this assumption, noting that recollection of

non-targets is not always necessary for successful comple-

tion of the exclusion task. The reason for this is that the

failure to recollect information that would identify a test

item as a target is one basis upon which a correct response

to non-targets and new words can be made. Herron and

Rugg [20] suggested that implementation of this strategy

was the reason for the absence of the non-target old/new

effects in Experiment 1: participants were successful in

restricting recollection to information diagnostic of the

target status of an item and did not recollect information

about non-targets.

In extending this account to explain why non-targets

were associated with a reliable left-parietal ERP old/new

effect in Experiment 2, Herron and Rugg noted that the

reliability of the strategy described above decreases along

with the likelihood of recollecting information about targets.

They noted that target judgments in Experiment 2 were less

accurate than in Experiment 1 and proposed that because of

this, participants attempted to recollect information about

both targets and non-targets in order to ensure successful

completion of the task.

There are several implications of the findings of Herron

and Rugg, one of which is the way in which these data impact

on assumptions concerning the use of the exclusion task and

the process-dissociation procedure [20,24,57,58]. A second

implication, and the principal focus here, concerns the

boundary conditions for when a strategy of recollecting

information associated with targets only can be implemented.

The experiments that are presented here are concerned with

these boundary conditions, and are motivated by findings

suggesting that under some circumstances, even when the

accuracy of target judgments is relatively high, recollection

of information associated with non-targets occurs. The

critical data come from studies in which the target/non-target

distinction was based upon study voice (male or female;

[52,54]) or the colour in which pictures were presented at

study (red or green; [7–9]). In the case of both of these

manipulations, and in contrast to the findings of Herron and

Rugg described previously, the researchers have reported

reliable non-target ERP old/new effects (also see [19]).

One way to reconcile these findings is to assume that the

precision with which recollection can be controlled depends

upon the relationship between the information that either is

or is not to be recollected. That is, in the case of the voice

E.L. Wilding et al. / Cognitive Brain Research 25 (2005) 19–32 21

and colour studies described above, one explanation for the

presence of the non-target ERP old/new effects is that the

high degree of similarity between the target and non-target

information precluded the restriction of recollection to

information associated with targets only. In Experiment 1

of Herron and Rugg [20], meanwhile, the cognitive opera-

tions underlying sentence generation and ratings for

pleasantness were sufficiently distinct to permit recollection

to be restricted to the latter only.

While this explanation for the colour and voice data is

reasonable, the majority of the foregoing inferences about

the conditions under which recollection of non-targets

occurs in the colour and voice studies are based only on

informal observations of the correspondences between

electrophysiological as well as behavioural data across

studies. In the case of the colour and voice data, moreover,

the published non-target old/new effects have, at least on the

basis of visual inspection, been of smaller magnitude than

the target old/new effects. The reason that this is important

is that this pattern of findings raises the possibility that

larger non-target old/new effects (relative to target old/new

effects) might be obtained as target accuracy—hence, the

likelihood of target recollection—decreases. This pattern of

findings would be consistent with the view that at least

partial control of recollection of non-target material can be

exerted even when the information associated with targets

and with non-targets corresponds closely. There are, how-

ever, no studies to date in which the outcomes of the

appropriate contrasts—those between target and non-target

old/new effects for the same kinds of perceptual information

and separated according to levels of target accuracy—have

been reported.

The two experiments presented here were designed in

order to address this disparity, and in both experiments, the

colour of visually presented words was the contextual

information that was designated as being important for the

target/non-target distinction in subsequent retrieval phases.

The two experiments differed only in the number of study–

test cycles and the number of words per cycle. Experiment 1

contained fewer study–test blocks and more study and test

words per block than Experiment 2. This difference was

introduced in order to encourage more accurate performance

in Experiment 2, thereby permitting an analysis of the

left-parietal ERP old/new effects for the same forms of

information under conditions in which target accuracy

differed.

In addition, the lengths of study and test lists in the two

experiments were set at a level that ensured the accuracy of

target judgments was higher (Experiment 2) and lower

(Experiment 1) than in the majority of previous ERP studies

in which voice or colour has been the source of information

on which target/non-target judgments were to be made.

Non-target left parietal ERP old/new effects of smaller

amplitude in Experiment 2 than Experiment 1 (relative to

the relevant target effects in each case) would suggest that

participants can exert some degree of control over recollec-

tion of colour information even under conditions where

target recollection also depends on accessing colour

information. Equivalent non-target ERP old/new effects

across the two experiments would provide important

information concerning limits on the resolution with which

control over recollection can be exerted.

2. Methods

Due to the similarity of the two experiments, they are

described jointly.

2.1. Participants

Twenty participants (12 female, mean age = 21 years)

took part in Experiment 1. The data from two participants

(one male) were discarded due to insufficient trials in at

least one response category of interest because of excessive

EOG artefacts. Eighteen participants (eight female, mean

age = 22 years) took part in Experiment 2. No participant

took part in both experiments. All participants had normal or

corrected to normal vision and were right-handed, as

assessed by writing-hand. No participants reported suffering

from red/green colour blindness, all were native English

speakers, and each gave informed consent before complet-

ing the task. The data for Experiment 1 were collected prior

to the data for Experiment 2.

2.2. Materials

The stimuli comprised low-frequency words (MRC

psycholinguistic database: frequency 1–9/million); 240

words were used in Experiment 1 and 180 words were

used in Experiment 2. All words were open-class and

ranged between four and nine letters in length. Words were

presented visually on a monitor 1.2 m from the participant,

in red or green text in the study phases and white text in the

test phases. A black background was used during all study

and test phases. The stimuli subtended maximum visual

angles of 0.5- (vertical) and 2.2- (horizontal).

2.3. Design and procedure

Participants were fitted with an electrode cap before the

experiment (see below) and were seated in a sound-

attenuated room facing a monitor with their thumbs resting

on two response keys. There was a short practice session

before the experiment and short breaks between each

study–test cycle. There were 8 study–test cycles in Expe-

riment 1, and 12 cycles in Experiment 2. In Experiment 1,

the 240 words were divided randomly into 8 equal blocks,

creating 8 study–test cycles each containing 20 study words

and 30 test words (an equal number of old red, old green

and new words). In Experiment 2, the 180 words were

divided randomly into 12 equal blocks creating 12 study–

1 These and all subsequent analyses of variance incorporated the

Geisser–Greenhouse correction for non-sphericity where necessary [18].

Corrected degrees of freedom are shown where appropriate.


test cycles, each containing 10 study words and 15 test

words (again, an equal number of old red, old green, and

new words). In both experiments, the words in each block

were rotated so that, across participants, each word was

encountered as an old red word, an old green word, and a

new word. This manipulation resulted in the creation of

three complete task-lists in each experiment, and an equal

number of participants completed each list. The order of

word presentation at both study and test within each study–

test cycle of each list was determined randomly for each

participant.

Prior to commencing the experiments, participants were

informed that, in each study cycle, they would see some

words presented in red and some in green, and that they

were to make key presses with either their right or left hand

based upon the colour in which the words were presented.

They were instructed that, in each test phase, words

presented in one colour at study (targets) would require a

key press with one hand, while a key press with the other

hand would be required for words presented in the other

colour (non-targets) as well as for words that were new to

the experiment. Participants were also informed that target

designation would only be revealed at the start of each test

phase, and that no words would appear in more than one

study–test cycle.

In both experiments, each study trial began with an

asterisk (*), which was displayed for 500 ms and followed by

a blank screen (200 ms), after which the study word was

presented for 300 ms. Half of the study words were presented

in red/green in a random order. Participants indicated the

colour in which each word was presented, and the next trial

began 500 ms after the response. An interval of approxi-

mately 1 min intervened between each study and test phase,

during which participants were informed of the Ftarget_response requirements for the test phase.

Each test trial also began with an asterisk (500 ms

duration), and 200 ms then intervened before the presen-

tation of a test word. Each word was visible for 300 ms

and then the screen remained blank until 500 ms after the

response of the participant, at which time the asterisk

signalling the start of the next trial appeared. Participants

were asked to make a target or non-target/new judgement

via key press when each test word appeared. For each

participant, the first half of the study–test cycles (4 in

Experiment 1 and 6 in Experiment 2) had the same target

designation (red or green) and the remainder had the

alternate designation. Blocking of designation was

employed in order to reduce the number of times in

which the response requirements changed for participants

during the experiments. The order of target designation

was balanced across participants, as were the hands

required for the binary red/green judgment at study and

the binary target/non-target/new judgment at test. Partici-

pants were encouraged to respond while balancing speed

and accuracy equally. Test responses faster than 650 ms

were treated as errors.

2.4. Electrophysiological recording procedure

EEG was recorded from 25 silver/silver chloride electro-

des housed in an elastic cap. They were located at midline

(Fz, Cz, Pz) and left/right hemisphere locations (FP1/FP2,

F7/F8, F5/F6, F3/F4, T3/T4, C5/C6, C3/C4, T5/T6, P5/P6,

P3/P4, O1/O2). Additional electrodes were placed on the

mastoid processes, and in Experiment 2, a further electrode

was placed on the nose-tip. EEG was recorded continuously

at 166 Hz (6 ms per point) with Fz as the reference

electrode, and was re-referenced computationally off-line to

a linked mastoid reference into epochs of 1536 ms (256 data

points), each including a 102-ms pre-stimulus baseline

relative to which all post-stimulus amplitudes were mea-

sured. EOG was recorded from above and below the right

eye (VEOG) and from the outer canthi (HEOG). Trials

containing large EOG artefact were rejected, as were trials

containing A/D saturation or baseline drift exceeding T80AV. Other EOG blink artefacts were corrected using a linear

regression estimate [44].

Averaged ERPs were formed for correct judgments at test

to target, non-target, and new words for each participant in

each experiment. Participants were excluded if they did not

contribute at least 16 artefact-free trials to each response

category [51,55]. The ERPs were collapsed across colour,

on the basis of null results for this factor in preliminary

analyses of both the behavioural and the electrophysiolog-

ical data (in keeping with the findings of Cycowicz et al.

[8]). On average, less than 20% of trials per participant per

response category were rejected due to artefact. In Exper-

iment 1, the mean numbers of trials per category were 39,

47, and 65 for correct responses to target, non-target, and

new words, respectively. The equivalent values in Experi-

ment 2 were 39, 40, and 50. The averaged ERPs for each

participant and for each category of interest in each

experiment were subjected to a 7-point binomially weighted

smoothing filter prior to analysis.

3. Results

3.1. Behavioural data

The probabilities of correct responses to red and green

words during the study phases of both experiments were

high and statistically equivalent, as assessed by ANOVA1

with between and within factors of experiment and colour,

respectively (Experiment 1: red = 0.98, green = 0.99;

Experiment 2: red = 0.99, green = 0.99). In two further

analyses, the reaction times for the colour judgments at

study were analysed separated according to whether they

Table 1

Probabilities of correct responses and reaction times (RT) to target, non-

target, and new words in Experiments 1 and 2 (SD in brackets)

Target Non-target New

Experiment 1

P (correct) 0.63 (0.14) 0.76 (0.07) 0.95 (0.05)

RT 1040 (279) 1126 (387) 886 (246)

Experiment 2

P (correct) 0.78 (0.06) 0.82 (0.06) 0.97 (0.02)

RT 977 (258) 1118 (338) 871 (279)


were designated as targets or non-targets in the subsequent

test phase(s). These analyses were restricted to the blocks

where participants had completed three previous blocks with

the same target designation (data from blocks 4 and 8 in

Experiment 1 and data averaged across blocks 4–6 and 10–

12 in Experiment 2). Separate analyses were conducted on

the data from each experiment, including the factors of

subsequent target designation and block (4 vs. 8 in Exp. 1,

4–6 vs. 10–12 in Exp. 2). No reliable effects were revealed

in either analysis. The mean study RTs for targets and non-

targets in Experiment 1 were 875 and 850 ms, respectively,

for block 4, and 865 and 840 ms for block 8. In Experiment

2, the values were 1070 and 1101 ms (blocks 4–6) and 1143

and 1113 ms (blocks 10–12)2. The reason for these

restricted block analyses was to determine whether there

was any evidence to support the view that, as the task

unfolded, participants treated study items differentially,

which might impact upon the patterns of ERP data that

were acquired in subsequent retrieval phases. We return to

this issue in the Discussion.

Table 1 shows the mean probabilities of, and reaction

times (RTs) for, correct responses to target, non-target, and

new words in Experiments 1 and 2. Correct responses to

targets and non-targets will be referred to as target hits and

non-target hits, respectively. The term old words will refer

to targets and non-targets jointly. In each experiment, the

probabilities of a target hit were reliably greater than the

probabilities of a target response to a non-target [Experi-

ment 1: t(17) = 6.85, P < 0.001; Experiment 2: t(17) =

14.89, P < 0.001] or a new word [Experiment 1: t(17) =

13.11, P < 0.001; Experiment 2: t(17) = 31.10, P < 0.001].

The analysis of the accuracy data [factors of experiment

(between) and response category (within)] revealed a main

effect of experiment [F(1,34) = 9.24, P < 0.01] and an

interaction between this factor and category [F(1.6,55.4) =

4.31, P < 0.05]. Post hoc analyses (Newman–Keuls)

revealed that only target accuracy was superior in Experi-

ment 2 to Experiment 1. RT varied according to response

category only [F(1.8,62.4) = 38.40, P < 0.001], and post

hoc analyses revealed that RTs for correct judgments to new

words were reliably faster than those for targets, which were

in turn reliably faster than those for non-targets.

3.2. Results: ERPs

The ERPs from a 3 � 3 array of 9 electrode sites for

Experiments 1 and 2 are shown in Fig. 1. The figure

shows the ERPs associated with correct judgments to

target, non-target, and new words. The fact that this

selection of electrode locations permits inspection of the

principal divergences between response categories is

attested to by the data shown in Fig. 2, which comprises

2 We are grateful to an anonymous referee for suggesting this specific

analysis.

the scalp distributions of the ERP old/new effects for

targets and for non-targets in the two experiments over 4

time windows: 300–500, 500–700, 700–900, and 900–

1400 ms. These scalp maps are derived from the difference

scores obtained by subtracting the ERPs evoked by new

words from those evoked by targets and non-targets,

respectively. Fig. 3 shows the same data as for the lower

half of Fig. 1, with the exception that the data is referenced

to nose-tip. We return to the importance of this figure in

the Discussion.

3.2.1. Analysis of left-parietal ERP old/new effects

The initial directed analysis of the ERPs was conducted

in order to determine the relationships between the left-

parietal ERP old/new effects that were obtained in the two

experiments. This analysis was restricted to the P5 electrode

location, and included the factors of experiment and

response category (correct responses to target, non-target,

and new test words). The analysis revealed an interaction

between these two factors [F(1.8,60.9) = 5.02, P < 0.025]

and was followed up by separate analyses of the target and

non-target ERP old/new effects in each experiment. These

revealed that the ERPs evoked by old words were reliably

more positive than those evoked by new words with the

exception of the non-target vs. new contrast in Experiment 2

[Exp. 1: targets F(1,17) = 48.08, P < 0.001, non-targets

F(1,17) = 14.42, P < 0.01; Exp. 2: targets F(1,17) = 24.40,

P < 0.001, non-targets F < 1]. Target old/new effects were

also reliably larger than non-target effects in both experi-

ments [Experiment 1: F(1,17) = 17.84, P < 0.01;

Experiment 2: F(1,17) = 32.39, P < 0.001].

3.2.2. Global analyses

The directed analyses were followed by a series of

analyses conducted using data from the 9 electrodes shown

in Fig. 1 (F5, Fz, F6, C5, Cz, C6, P5, Pz, P6). The analyses

incorporated the factors of location in the anterior/posterior

(AP: anterior, central, posterior) and left–right (LR: left-

hemisphere, midline, right-hemisphere) planes. They were

conducted separately for each experiment in order to

determine the correspondence between the ERP old/new

effects in these experiments and those reported in previous

studies, as well as to license subsequent between-experi-

ment analyses [42,54]. They were conducted employing

Fig. 1. Grand average ERPs associated with correct judgments to target, non-target, and new words in Experiments 1 (upper panel) and 2. The data are shown

for nine sites at anterior (from left to right: F5, Fz, F6), central (C5, Cz, C6), and posterior (P5, Pz, P6) scalp locations.


data from the 300–500, 500–700, 700–900, and 900–1400

ms time windows, selected in order to encompass the

epochs in which ERP old/new effects have been reported

previously (e.g., [54]), as well as to capture the differences

between experiments that are evident in Figs. 1 and 2.

For each experiment, an initial analysis was conducted

incorporating data from all three response categories as well

as the factor of epoch, in addition to the site factors (AP and

LR) described above. Both analyses revealed interactions

between epoch (EP) and response category (RC), including

in each case a four-way interaction between these two

factors, AP and LR [Exp. 1: F(7.4,126.0) = 5.08, P < 0.001;

Exp. 2: F(6.0,101.9) = 2.20, P < 0.05]. For each expe-

riment, these analyses were followed up by paired contrasts

for the data from successive epochs. All other factors were

as described above, and the outcomes of these analyses

(restricted to interactions involving epoch and response

category) are shown in Table 2, where it can be seen that

there are either reliable or marginal interactions between the

relevant factors for each of the paired epoch contrasts.

The purpose of the analyses that included epoch as a

factor (see Table 2) was to establish that the epochs

employed in this analysis section are the ones that likely

separate the ERP data into periods during which reliably

different patterns of ERP old/new effects occurred. In order

to establish the profile of the old/new effects within each

epoch, these initial analyses were followed by separate

analyses within each epoch. The analyses comprised all

possible paired contrasts of the ERPs associated with correct

judgments to targets, non-targets, and new test words (see

Tables 3 and 4: main effects or interaction terms that do not

include the factors of response category are not shown as

they are not the outcomes of principal interest here). The

following description of the outcomes of these analyses is

restricted to the highest-order interactions that were obtained

in each case.

Fig. 2. Scalp distributions of the ERP old/new effects for targets and for non-targets in Experiments 1 and 2. The data are shown for four post-stimulus epochs:

300–500, 500–700, 700–900, and 900–1400 ms. The maps were computed from the scores obtained by subtracting mean amplitudes from the ERPs evoked

by new words from those evoked by targets and non-targets. The paired values below each map denote the maxima and minima of the amplitude differences

between response categories, which can be interpreted relative to the bar on the left-hand side of the figure.


3.2.3. Experiment 1 — see Table 3

For all epochs, the contrasts involving new words

revealed interactions between response category, AP, and

LR. For the 300–500 and 500–700 ms epochs, these

interactions reflect the vertex maximum of the positive-

going ERP old/new effects, and the left-lateralisation of the

Fig. 3. Grand average ERPs associated with correct judgments to target, non-target

as for Fig. 1.

effects at posterior electrodes, as Fig. 2 illustrates. The same

posterior lateralisation is evident in the 700–900 ms epoch,

while at frontal locations over this time window, the small

positive-going old/new effect is distributed relatively more

evenly across the two hemispheres. In the 900–1400 ms

epoch, the right-lateralised ERP old/new effects at anterior

, and new words in Experiment 2, referenced to nose-tip. Electrode locations

Table 2

Results of the paired epoch contrasts between the ERPs evoked by correct judgments to target, non-target, and new words in Experiments 1 and 2

Effect df 300–500 vs. 500–700 500–700 vs. 700–900 700–900 vs. 900–1400

F E F E F E

Experiment 1

EP � RC 2,34 7.79** 0.89 13.75*** 0.94 7.70** 0.78

EP � RC � LR 4,68 3.24* 0.72 12.27*** 0.55 4.13* 0.61

EP � RC � AP 4,68 3.53* 0.54 3.56* 0.44 12.66*** 0.50

EP � RC � AP � LR 8,136 5.42*** 0.45 7.65*** 0.63 2.18& 0.52

Experiment 2

EP � RC 2,34 ns 11.11*** 0.82 ns

EP � RC � LR 4,68 ns 11.68*** 0.70 ns

EP � RC � AP 4,68 ns ns 4.56* 0.51

EP � RC � AP � LR 8,136 2.58& (P = 0.05) 0.45 ns ns

EP = epoch, RC = response category, AP = anterior/central/posterior, LR = left/midline/right. &P < 0.1, *P < 0.05, **P < 0.01, ***P < 0.001, E = epsilon value.


locations are accompanied by a negative-going old/new

effect with a posterior midline (Pz) maximum.

The direct contrast between targets and non-targets re-

vealed that the ERPs evoked by targets are more positive-

going from 500 to 900 ms, with this positivity being most

pronounced at the midline for the 500–700 ms epoch,

reflected in a RC � LR interaction. The RC � AP � LR

interaction over the 900–1400 ms epoch reflects the fact that

the greater relative positivity for targets is larger at midline

and right than at left anterior locations, while at Pz, the ERPs

evoked by non-targets are relatively more positive-going.

3.2.4. Experiment 2 — see Table 4

For the target vs. new contrast, the interactions between

RC, AP, and LR from 500 to 900 ms and the marginal

interaction in the 900–1400 ms epoch come about for

similar reasons to those described above for Experiment 1

for the contrasts involving new words. In the 300–500 ms

Table 3

Results of the paired contrasts between ERPs evoked by correct judgments to targe

700–900, and 900–1400 ms epochs

Effect df 300–500 500–700

F E F

Target vs. new

RC 1,17 35.93*** 42.26***

RC � AP 2,34 ns ns

RC � LR 2,34 4.78* 0.88 9.92**

RC � AP � LR 4,68 4.26* 0.69 9.48***

Non-target vs. new

RC 1,17 31.15*** 10.49**


RC � LR 2,34 8.85** 0.86 5.23*

RC � AP � LR 4,68 4.36* 0.57 10.64***

Target vs. non-target

RC 1,17 ns 23.68***


RC � LR 2,34 ns 3.86*

RC � AP � LR 4,68 ns ns

Abbreviations and meanings of symbols are as for Table 2.

epoch, the same three-way interaction reflects the vertex

maximum of the old/new effect. For the non-target versus

new contrast, there were no reliable ERP old/new effects

between 500 and 900 ms. The analyses in the early (300–

500 ms) epoch revealed a main effect of response category

only—despite the apparent anterior maximum of the ERP

old/new effect—and reflects the relatively greater positivity

for non-targets. For the 900–1400 ms epoch, the RC�AP�LR interaction revealed in the non-target vs. new contrast

reflects primarily the fact that the differences between the

ERPs are restricted to midline central and posterior scalp

locations, where the ERPs evoked by new test words are

more positive-going than those evoked by non-targets.

For the target vs. non-target contrast, the RC � AP and

RC � LR interactions for the 300–500 ms epoch reflect the

fact that the greater positivity for targets is smaller at

anterior than at central and posterior locations, and larger at

the midline than at left and right hemisphere locations. For

t, non-target, and new words in Experiment 1 over the 300–500, 500–700

700–900 900–1400

E F E F E

12.80** ns

ns 23.22*** 0.62

0.89 2.97& 0.96 10.26*** 0.99

0.72 6.61** 0.69 8.29*** 0.71

5.62* ns

ns 10.02** 0.68

0.97 3.49* 0.88 4.21* 0.97

0.75 5.35** 0.78 3.20* 0.73

5.21* ns

ns 7.43** 0.64

0.92 ns ns

2.47& 0.63 2.84** 0.75

,

Table 4

Results of the paired contrasts between ERPs evoked by correct judgments to target, non-target, and new words in Experiment 2

Effect df 300–500 500–700 700–900 900–1400

F E F E F E F E

Target vs. new

RC 1,17 51.31*** 14.98** ns ns

RC � AP 2,34 3.04& 0.59 ns ns 4.92* 0.71

RC � LR 2,34 13.37** 0.64 3.15& 0.84 ns 4.03* 0.90

RC � AP � LR 4,68 2.85* 0.70 4.78** 0.66 3.15* 0.74 2.24& 0.73

Non-target vs. new

RF 1,17 13.51** ns ns 4.86*

RC � AP 2,34 ns ns ns 7.49** 0.74

RC � LR 2,34 3.18& 0.91 ns 3.34& 0.99 7.10** 0.92

RC � AP � LR 4,68 ns 2.42& 0.82 ns 3.54** 0.75

Target vs. non-target

RC 1,17 14.56** 14.92** ns 3.71&RC � AP 2,34 5.66* 0.68 16.03*** 0.62 3.56& 0.59 ns

RC � LR 2,34 8.13** 0.84 4.39* 0.93 ns ns

RC � AP � LR 4,68 ns 2.73* 0.81 2.92& 0.60 ns

Abbreviations and meanings of symbols are as for Table 2.


the 500–700 ms epoch, the RC � AP � LR interaction for

the contrast between the two classes of old word is due to

reasons similar to those responsible for the same interaction

term that was revealed in the target vs. new contrast: the fact

that the greater relative positivity for targets is most

pronounced at left hemisphere central and posterior scalp

locations.

3.2.5. Between-experiment analyses

These comprised direct contrasts of the ERP old/new

effects from the two experiments, and were completed on

mean amplitudes computed from subtraction waveforms

obtained by subtracting the ERPs evoked by correct

responses to new words from those evoked by target and

non-target hits. The initial analyses were completed sepa-

rately for targets and for non-targets, and were restricted to

targets for the 500–700 and 700–900 ms epochs as the

within-experiment analyses described above revealed no

reliable old/new effects for non-targets over these epochs in

Experiment 2. The analyses each included factors of

experiment, AP and LR, using the same epochs and electrode

locations described above. Where these analyses revealed

interactions between experiment and scalp locations, they

were followed up by analyses of rescaled data in order to

determine whether the interactions came about because of

differences between the scalp distributions of the ERP old/

new effects, or simply because of amplitude differences [33].

The data were rescaled using the vector length method [33],

and the rescaling was computed over all 25 sites from which

data were collected, although the analyses included the same

factors and electrode locations that were employed for the

data that were analysed prior to rescaling.

For the 300–500 ms epoch, the only reliable effect

involving experiment was an interaction with the LR

dimension for targets [F(1.6,54.2) = 3.83, P < 0.05],

reflecting the fact that the old/new effects in Experiment 1

are larger than those from Experiment 2 at left hemisphere

locations but smaller at the midline. This interaction was not

reliable for the analysis of rescaled data. For the 500–700

and 700–900 ms epochs, the target old/new effects in

Experiment 1 were larger than those from Experiment 2

[500–700: F(1,34) = 5.76, P < 0.05; 700–900: F(1,34) =

8.66, P < 0.01]. For the 900–1400 ms epoch, the contrast

for non-targets revealed a main effect of experiment

[F(1,34) = 4.31, P < 0.05], reflecting the fact that the

positive-going old/new effects at anterior locations were

larger in Experiment 1, while the negative-going effects

at posterior locations were larger in Experiment 2. The

contrast for targets over this epoch revealed an interaction

between experiment and the anterior/posterior dimension

[F(1.4,47.1) = 4.17, P < 0.05], reflecting the fact that while

the negative-going ERP old/new effects at posterior

electrodes differ minimally across experiments, at anterior

electrode locations the positive-going old/new effects are

markedly larger in Experiment 1 than in Experiment 2. This

interaction was not, however, reliable for the analysis of

rescaled data.

4. Discussion

Participants were able to discriminate between targets

and non-targets in both experiments, but the proportion of

correct target judgments was higher in Experiment 2 than in

Experiment 1. This pattern of behavioural data indicates that

participants were able to use recollection of the colour of

study presentation as a basis for distinguishing between

targets and non-targets, and that the amount or quality of

information available for this discrimination was greater in

the second experiment.


In and of themselves, however, the behavioural findings in

these experiments do not support unequivocally the assump-

tion that participants recollected information about both

targets and non-targets in order to make judgments at test.

Succeeding or failing in the recollection only of colour

information for targets can also serve as the basis for

distinguishing targets from non-targets and new words

[19,20]. As noted in the Introduction, this strategy becomes

increasingly effective as the likelihood of recollecting

information associated with targets increases, and the analysis

of ERPs evoked by test stimuli in these two experiments

provides an opportunity to assess the contribution of recol-

lection during the completion of exclusion tasks that differed

primarily in terms of the accuracy of target judgments.

Reliable left-parietal ERP old/new effects were obtained

for targets in both experiments, but for non-targets in

Experiment 1 only, suggesting that participants depended

upon recollection of information associated with non-targets

to a greater degree in Experiment 1 than in Experiment 2.

Since the accuracy of task judgments was superior in the

second experiment, this finding is consistent with Herron

and Rugg’s proposal that a strategy of depending primarily

on recollection of target information for task judgments is

employed when the likelihood of recollecting information

about targets is high [14,19,20].

It is also of critical importance here that this evidence for

selective recollection was obtained in tasks where the

principal feature distinguishing targets from non-targets

was colour. Thus, participants were able to restrict recol-

lection of information associated with non-targets, while in

both experiments, recollecting colour information about

targets. These results suggest that a considerable degree of

control can be exerted over what is and what is not

recollected, even under circumstances when there is a close

correspondence between these two classes of information.

A second important aspect of the findings in Experiment

1 is the fact that the left-parietal ERP old/new effect for non-

targets, while reliable, was smaller than the effect for targets.

This finding raises the possibility that, while relying on

recollection of non-target colour information to a greater

extent in Experiment 1 than Experiment 2, participants in

Experiment 1 also relied upon recollection of target colour

information to a greater degree than upon recollection of

non-target colour information. This could have come about

because of variability within participants with respect to the

retrieval strategy that was adopted during the course of the

retrieval phases, or because of variability across participants,

whereby some but not all participants in Experiment 1

depended solely upon the success or failure of recollection

of colour information about targets in order to make task

judgments [52,54].

While the present data do not permit a clear separation of

these alternatives, the accuracy of either does not challenge

two central claims. First, that the degree of dependence

upon recollection of colour information associated with non-

targets is correlated inversely with the accuracy of target

judgments. Second, that this strategic control of retrieval can

occur despite an apparently close correspondence between

the sources of information associated with targets and with

non-targets.

Whether the same conclusions hold for auditory source

information remains to be determined. As noted in the

Introduction, reliable non-target ERP old/new effects were

reported by Wilding and Sharpe [54] in two experiments

where target designation was based on the voice (male/

female) in which words were spoken at study and target

accuracy was�0.70. Whether these non-target effects would

be attenuated as accuracy increased to the levels achieved by

participants in Experiment 1 is a question for future research,

as is the question of whether there are other forms of

information for which selective recollection does not or

cannot occur during completion of the exclusion task [21].

It is also important to note that the analyses of the study-

phase RT data provide no support for the possibility that

some of the differences between the ERPs at test in Expe-

riments 1 and 2 came about because of differential encoding

of targets and of non-targets. In both experiments, the

colour of words designated as targets was changed halfway

through the experiment—after four study–test blocks in

Experiment 1 and 6 study–test blocks in Experiment 2.

Participants were not informed of target designation until

the start of each test phase, they were not informed of the

total number of study–test blocks in each experiment, and

they were told that target designation might change for each

retrieval phase. Despite these aspects of the design,

however, the possibility remains that participants made

assumptions about likely target designation and encoded

information at study differentially on the basis of these

assumptions.

In order to assess this possibility, the RTs for words

presented during the study phases in the blocks where the

same target designation had been required in the three

preceding blocks were analysed after being separated

according to whether they were designated subsequently

as targets or non-targets. The analyses revealed no effects

involving block position or target designation, a finding

consistent with the view that the differences between the

ERPs at test can be ascribed to factors operating at retrieval

rather than at encoding2. Support for the claim that the test

data is a function of strategies adopted at test also stems

from the fact that qualitatively similar patterns of target and

non-target ERP old/new effects have been obtained in

studies where there was only one study phase [20,21].

A second challenge for the conclusions drawn above

stems from the consideration that there may be variations

across the two experiments in the levels of confidence

associated with task judgments. Johnson et al. [29] reported

that ERPs elicited during a recognition memory task may be

sensitive to variations in response confidence, and in their

study, these variations manifested themselves as an increased

positivity with increasing confidence, the effects being

largest at left-central scalp locations (see also [15]). Due to


the fact that these modulations linked with decision confi-

dence were identified at scalp locations and in time periods

(¨500–900 ms post-stimulus) that overlap with those in

which the left-parietal ERP old/new effect is typically

observed, it is possible that variations in decision confidence

may be responsible for the patterns of ERP old/new effects

that were obtained in the experiments reported here.

For the following reason, however, this account seems

unlikely. Response confidence may on average have been

greater in Experiment 2 than in Experiment 1 because of

superior response accuracy. In Experiment 2, however, left-

parietal ERP old/new effects were non-significant for non-

targets, while the effects for the same category were reliable

in Experiment 1. Given that the data indicating ERP

sensitivity to response confidence predict relatively greater

positivity with increasing confidence [15,29], and the

reasonable inference that the confidence in judgments

increases with response accuracy, this pattern of data

suggests that variations in confidence do not explain

satisfactorily the pattern of data across the two experiments:

for non-targets, the left-parietal ERP old/new effects are

larger in the experiment where response confidence is if

anything likely to be lower. Rugg et al. [40], moreover,

show data illustrating that ERP old/new effects at left-

parietal scalp locations differ little according to response

confidence (see also [36,38]).

What mechanisms might operate during retrieval tasks in

order to permit control over episodic retrieval? Common to

several conceptual frameworks is the assumption that

processes engaged before and/or after an interaction bet-

ween a retrieval cue and a memory trace can influence what

information is available and/or employed for memory

judgments [3,35,48]. Anderson and Bjork [2] identified

two classes of process that operate prior to this interaction

and which might support selective retrieval (also see [1,30]).

The first, target bias, describes processes that operate

directly on memory representations themselves. This class

of process enables selective retrieval by making some

representations more likely to interact with an appropriate

retrieval cue than others. A second class of process

described by Anderson and Bjork is cue bias, which

describes processes that operate directly on retrieval cues,

increasing the likelihood that they will interact with some

memory representations rather than with others. As for

target bias processes, cue bias operations are engaged prior

to a cue/trace interaction and influence the nature of that

interaction. A third class of process discussed by Anderson

and Bjork—attention bias—has a different locus, operating

downstream of any interactions between a retrieval cue and

a memory trace. This class of process enables selective

retrieval by ensuring that processing resources are allocated

only to task-relevant material that becomes available as a

result of the cue/trace interaction.

There is to date no electrophysiological data that dis-

tinguish unequivocally between these classes of bias

mechanism, although it has been argued that target bias

mechanisms are likely to be maintained throughout a

retrieval task and thus will not be revealed in time-locked

and baseline-corrected event-related potential measures of

neural activity [22]. Herron and Rugg [19] favoured a cue-

bias account of the ERP data they acquired in a task where

the encoding phase comprised presentation of words and

pictures. Test stimuli were words, and the old words were

either re-presentations of words, or of words corresponding

to the objects shown in the pictures. In separate retrieval

phases, targets were designated as old words encountered

either as words or pictures at encoding.

There were reliable parietal old/new effects for targets in

both target designations, but reliable parietal effects for non-

targets only when pictures were designated as targets.

Herron and Rugg [19] noted that these data could not be

accommodated straightforwardly within an attention bias

account, but could be within a cue-bias account if it was

assumed that: (1) in the word target condition, participants

were able to process cues sufficiently selectively to restrict

recollection to information associated with studied words,

perhaps because of the perceptual match for the target

stimuli, while (2) in the picture target condition, the absence

of the perceptual match for target stimuli meant that

processes engaged to recover information about studied

pictures involved processing operations that also resulted in

recollection of information associated with studied words.

This account does not preclude the possibility that a

combination of these classes of bias mechanism is typically

responsible for selective episodic retrieval, nor whether the

extent to which these processes act in concert varies

according to factors such as the content that is to be retrieved

and the structure of retrieval tasks. Identification of boundary

conditions for when selective retrieval can occur—a goal of

the study described here—is a precursor to attempts to

delineate further the mechanisms that enable selective

retrieval of episodic information to be accomplished.

Over and above the implications of the pattern of left-

parietal ERP old/new effects in these two experiments for

the resolution with which recollection can be constrained,

the contrasts between the target and non-target ERP old/new

effects revealed other modulations of the electrical record

that have been linked to memory processes. The analyses of

the ERP old/new effects provided no evidence for qualita-

tive differences between the ERP old/new effects in the two

experiments, but there were amplitude differences of note.

In the 300–500 ms post-stimulus time window, a fronto-

central modulation comprising a greater positivity for

correctly identified old than for new items has been linked

to familiarity [4,6,34,41,59], which according to dual-

process accounts of recognition memory is one basis for

making old/new recognition memory judgments [23,32,57].

Although the scalp maps in Fig. 2 show that there are some

differences according to experiment and target/non-target

status in this epoch, these were not borne out by the

analyses, which revealed only quantitative differences

between the ERP old/new effects for targets.


A disparity between the ERP old/new effects in the two

experiments that does receive statistical support is evident

primarily at frontal electrode locations, where the right-

lateralised ERP old/new effects were markedly more

prominent in Experiment 1 than in Experiment 2, and larger

for targets than for non-targets. According to one account of

the right-frontal ERP old/new effect, it indexes processes

responsible for monitoring retrieval processing in service of

task goals [42,43]. The current data are broadly compatible

with this account, in so far as the relatively lower level of

memory accuracy in Experiment 1 might well have en-

couraged a greater degree of content and/or response

monitoring than in Experiment 2, where the superior level

of memory accuracy may have precluded requirements for

extensive monitoring.

A further notable aspect of the ERP old/new effects is the

fact that common to both experiments is the relatively focal

negativity for old compared to new words that is evident

primarily during the 900–1400 ms epoch, and which has a

posterior midline maximum. While Fig. 1 shows that there

are small differences between these posterior negativities for

targets and for non-targets across experiments, the principal

divergence is between the ERPs evoked by new words and

those evoked by both classes of old word. This late posterior

negativity is evident in a number of recent ERP studies of

source memory (for examples, see [27,31]), and the func-

tional significance of the modulation is a matter of

continuing debate. In a series of studies, Cycowicz, Fried-

man, and colleagues [7–9] have reported pronounced

posterior negativities in tasks where the information to be

retrieved was the colour (red/green) in which pictures were

presented in a prior study phase. In their experiments, the

effect did not differ reliably according to the accuracy of

source judgments, leading to the proposal that it reflects

processes linked closely to the search for and/or retrieval of

colour information [7], or perhaps processes that are linked

with retrieval and/or evaluation of source information but

not necessarily restricted to colour [17].

The data in this experiment contribute in two ways to

questions about the functional significance of this posteri-

orly distributed old/new effect. First, the fact that the effect

is of comparable magnitude across the two experiments

while the right-frontal ERP old/new effect is markedly

larger in Experiment 1 is consistent with the view that the

two effects are functionally dissociable (see also [49]). The

second point stems from the fact that the ERP waveforms in

the studies of Cycowicz and colleagues have been displayed

with respect to a nose-tip reference only, with the exception

of one recent publication [17]. In the majority of ERP

memory studies, a linked mastoid reference has been

employed. The data from Experiment 2 are relevant in this

regard, as EEG was also acquired from the nose-tip,

permitting the data to be re-referenced to that site. The

outcome of this procedure is shown in Fig. 3, where it can

be seen that the consequence of this change of reference is a

considerable increase in the amplitude of the late posterior

negativity, and a small reversal of the positive-going ERP

old/new effects at anterior electrodes.

The data referenced to linked mastoids that is shown in

Figs. 1 and 2 bear strong similarities to that which has been

reported for different kinds of source information such as

voice [51] or study task [49]. One interpretation of this

correspondence is that the posterior negativity is not linked

strongly to colour-specific processing, given the apparent

match between the data reported here and in other studies in

which a linked mastoid reference has been employed and in

which retrieval of different kinds of source information has

been encouraged. The data in Fig. 3 support this account in

so far as when a nose-tip reference is employed, the pattern

of data resembles closely that reported previously by

Cycowicz and colleagues [7–9]. In combination, therefore,

the data provide little support for a colour-specific account

of the late posterior negativity.

From this perspective, Johannsen and Mecklinger’s

recent proposal that one aspect of the posterior negativity

reflects processes related to maintaining and/or forming

representations of attribute conjunctions is parsimonious,

since it makes no claims about the content of those

conjunctions [17,26]. It is important to observe, however,

that these suggestions are based on informal comparisons

across studies, and in determining the functional signifi-

cance of this late posterior modulation, it will be

important in future studies to contrast ERPs associated

with retrieval of different contents while equating all other

factors [17].

To summarise, the principal findings in these two

experiments concern the pattern of left-parietal ERP old/

new effects—the electrophysiological signature of recollec-

tion. In the first experiment, overall task accuracy was

inferior to that in the second experiment, and targets as well

as non-targets were associated with reliable left-parietal

ERP old/new effects. In the second experiment, only targets

were associated with the effect. In combination, these

findings are consistent with Herron and Rugg’s proposal

that relying upon the success or failure of recollection of

information associated with targets only is a strategy that is

adopted when memory for targets is high [19,20]. In

addition, the findings in these two studies demonstrate that

control over what is and what is not recollected can be

exercised with a high degree of precision, as attested to by

the fact that the data in Experiment 2 suggest that

participants were able to recollect some colour information

while not recollecting other highly similar information.

Acknowledgments

This research was supported by the UK Biotechnology

and Biological Sciences Research Council (BBSRC). We

would like to thank Mariam Dzulkifli and Nazanin

Azimian-Faridani for comments on a draft of the

manuscript.


References

[1] M.C. Anderson, Rethinking interference theory: executive control

and the mechanisms of forgetting, J. Mem. Lang. 49 (2003)

415–445.

[2] M.C. Anderson, R.A. Bjork, Mechanisms of inhibition in long term

memory, in: D. Dagenbach, T.H. Carr (Eds.), Inhibitory Processes in

Attention, Memory, and Language, Academic Press, San Diego, CA,

1994, pp. 265–325.

[3] P.W. Burgess, T. Shallice, Confabulation and the control of recollec-

tion, Memory 4 (1996) 359–411.

[4] T. Curran, Brain potentials of recollection and familiarity, Mem. Cogn.

28 (2000) 923–938.

[5] T. Curran, Effects of attention and confidence on the hypothesized

ERP correlates of recollection and familiarity, Neuropsychologia 42

(2004) 1088–1106.

[6] T. Curran, A.M. Cleary, Using ERPs to dissociate familiarity from

recollection in picture recognition, Cognit. Brain Res. 15 (2003)

191–205.

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

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

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

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

colour of objects: an ERP investigation of source memory, Cereb.

Cortex 11 (2001) 322–334.

[9] Y.M. Cycowicz, D. Friedman, M. Duff, Pictures and their colors: what

do children remember? J. Cogn. Neurosci. 15 (2003) 759–770.

[10] D.I. Donaldson, M.D. Rugg, Recognition memory for new associa-

tions: electrophysiological evidence for the role of recollection,

Neuropsychologia 36 (1998) 377–395.

[11] D.I. Donaldson, E.L. Wilding, K. Allan, Fractionating retrieval from

episodic memory using event-related potentials, in: A.E. Parker, E.L.

Wilding, T.J. Bussey (Eds.), The Cognitive Neuroscience of

Memory: Episodic Encoding and Retrieval, Psychology Press, Hove,

2003, pp. 39–58.

[12] E. Duzel, A.P. Yonelinas, G.R. Mangun, H.J. Heinze, E. Tulving,

Event-related brain potential correlates of two states of conscious

awareness in memory, Proc. Natl. Acad. Sci. U. S. A. 94 (1997)

5973–5978.

[13] E. Duzel, F. Vargha-Khadem, H.J. Heinze, M. Mishkin, Brain

activity evidence for recognition without recollection after early

hippocampal damage, Proc. Natl. Acad. Sci. U. S. A. 98 (2001)

8101–8106.

[14] A. Dzulkifli, E.L. Wilding, Electrophysiological indices of strategic

episodic retrieval processing, Neuropsychologia 43 (2005) 1152–1162.

[15] S. Finnigan, M.S. Humphreys, S. Dennis, G. Geffen, ERP ‘‘old/new’’

effects: memory and decisional factor(s), Neuropsychologia 40 (2002)

2288–2304.

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

memory encoding and retrieval: a selective review, Microsc. Res.

Tech. 51 (2000) 6–28.

[17] D. Friedman, Y.M. Cycowicz, M. Bersick, The late negative episodic

memory effect: the effect of recapitulating study details at test, Cognit.

Brain Res. 23 (2005) 185–198.

[18] G.W. Greenhouse, S. Geisser, On methods in the analysis of repeated

measures designs, Psychometrika 49 (1959) 95–112.

[19] J.E. Herron, M.D. Rugg, Retrieval orientation and the control of

recollection, J. Cognit. Neurosci. 15 (2003) 843–854.

[20] J.E. Herron, M.D. Rugg, Strategic influences on recollection in the

exclusion task: electrophysiological evidence, Psychon. Bull. Rev. 10

(2003) 703–710.

[21] J.E. Herron, E.L. Wilding, An electrophysiological investigation of

factors facilitating strategic recollection, J. Cognit. Neurosci. 17

(2005) 1–11.

[22] M. Hornberger, A.M. Morcom, M.D. Rugg, Neural correlates of

retrieval orientation: effects of study– test similarity, J. Cognit.

Neurosci. 16 (2004) 1196–1210.

[23] L.L. Jacoby, A process dissociation framework: separating auto-

matic from intentional uses of memory, J. Mem. Lang. 30 (1991)

513–541.

[24] L.L. Jacoby, Invariance in automatic influences of memory: toward a

user’s guide for the process-dissociation procedure, J. Exper. Psychol.,

Learn., Mem., Cogn. 24 (1998) 3–26.

[25] L.L. Jacoby, C. Kelley, Unconscious influences of memory: dissocia-

tions and automaticity, in: A.D. Milner, M.D. Rugg (Eds.), The

Neuropsychology of Consciousness, Academic Press, London, 1992,

pp. 201–233.

[26] M. Johannson, A. Mecklinger, The late posterior negativity in ERP

studies of episodic memory: action monitoring and retrieval of

attribute conjunctions, Biol. Psychol. 64 (2003) 91–117.

[27] M. Johansson, G. Stenberg, M. Lindberg, I. Rosen, Memory for

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


[28] M.K. Johnson, S. Hashtroudi, D.S. Lindsay, Source monitoring,

Psychol. Bull. 114 (1993) 3–28.

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

analysis of recognition-related event-related brain potentials, Int. J.

Psychophysiol. 29 (1998) 83–104.

[30] B.J. Levy, M.C. Anderson, Inhibitory processes and the control of

memory retrieval, Trends Cognit. Sci. 6 (2002) 299–305.

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

performed and planned actions, Int. J. Psychophysiol. 45 (2002)

197–210.

[32] G. Mandler, Recognising: the judgment of previous occurrence,

Psychol. Rev. 87 (1980) 252–271.

[33] G. McCarthy, C.C. Wood, Scalp distributions of event-related

potentials: an ambiguity associated with analysis of variance models,

Electroencephalogr. Clin. Neurophysiol. 62 (1985) 203–208.

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

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

[35] D.A. Norman, D.G. Bobrow, Descriptions: an intermediate stage in

memory retrieval, Cogn. Psychol. 11 (1979) 107–123.

[36] S.R. Rubin, C. Van-Petten, E.L. Glisky, W.M. Newberg, Memory

conjunction errors in younger and older adults: event-related

potential and neuropsychological data, Cogn. Neuropsychol. 16

(1999) 459–488.

[37] M.D. Rugg, ERP studies of memory, in: M.D. Rugg, M.G.H.

Coles (Eds.), Electrophysiology of Mind: Event-Related Brain

Potentials and Cognition, Oxford University Press, London,

1995, pp. 132–170.

[38] M.D. Rugg, M.C. Doyle, Event-related potentials and recognition

memory for low- and high-frequency words, J. Cognit. Neurosci. 4

(1992) 69–79.

[39] M.D. Rugg, R.C. Roberts, D.D. Potter, C.D. Pickles, M.E. Nagy,

Event-related potentials related to recognition memory. Effects of

unilateral temporal lobectomy and temporal lobe epilepsy, Brain 114

(1991) 2313–2332.

[40] M.D. Rugg, C.J.C. Cox, M.C. Doyle, T. Wells, Event-related

potentials and the recollection of low and high frequency words,


[41] M.D. Rugg, R.E. Mark, P. Walla, A. Schloerscheidt, C.S. Birch, K.

Allan, Dissociation of the neural correlates of implicit and explicit

memory, Nature 392 (1998) 595–598.

[42] M.D. Rugg, K. Allan, C.S. Birch, Electrophysiological evidence for

the modulation of retrieval orientation by depth of study processing,

J. Cognit. Neurosci. 12 (2000) 664–678.

[43] M.D. Rugg, J. Herron, A. Morcom, Electrophysiological studies of

retrieval processing, in: L.R. Squire, D.L. Schacter (Eds.), Neuro-

psychology of Memory, Guilford Publications, New York, 2002,

pp. 154–165.

[44] H.V. Semlitsch, P. Anderer, P. Schuster, O. Presslich, A solution for

reliable and valid reduction of ocular artifacts, applied to the P300

ERP, Psychophysiology 23 (1986) 695–703.

[45] M.E. Smith, Neurophysiological manifestations of recollective expe-


rience during recognition memory judgements, J. Cognit. Neurosci. 5

(1993) 1–13.

[46] M.E. Smith, E. Halgren, Dissociation of recognition memory

components following temporal lobe lesions, J. Exper. Psychol.,

Learn., Mem., Cogn. 15 (1989) 50–60.

[47] I. Tendolkar, A. Schoenfeld, G. Golz, G. Fernandez, K. Kuhl, R.

Ferszt, H. Heinze, Neural correlates of recognition memory with and

without recollection in patients with Alzheimer’s disease and healthy

controls, Neurosci. Lett. 263 (1999) 45–48.

[48] E. Tulving, Ecphoric processes in episodic memory, Philos. Trans. R.

Soc., London B302 (1983) 361–371.

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

event-related potential study of source memory, Neuropsychologia 37

(1999) 441–454.

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

recollection? Int. J. Psychophysiol. 35 (2000) 81–87.

[51] 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.

[52] E.L. Wilding, M.D. Rugg, Event-related potentials and the recognition

memory exclusion task, Neuropsychologia 35 (1997) 119–128.

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

recent insights from event-related potentials, in: A. Zani, A. Mado-

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

Academic Press, San Diego, 2003, pp. 169–196.

[54] E.L. Wilding, H.L. Sharpe, The influence of response– time demands

on electrophysiological correlates of successful episodic retrieval,

Cognit. Brain Res. 18 (2004) 185–195.

[55] E.L. Wilding, M.C. Doyle, M.D. Rugg, Recognition memory with and

without retrieval of context: an event-related potential study, Neuro-

psychologia 33 (1995) 743–767.

[56] A.P. Yonelinas, The nature of recollection and familiarity: a review of

the 30 years of research, J. Mem. Lang. 46 (2002) 441–517.

[57] A.P. Yonelinas, I. Dobbins, M.D. Szymanski, H.S. Dhaliwal, L. King,

Signal-detection, threshold, and dual-process models of recognition

memory: ROCs and conscious recollection, Conscious. Cogn. 5

(1996) 418–441.

[58] A.P. Yonelinas, N.E. Kroll, I. Dobbins, M. Lazzara, R.T. Knight,

Recollection and familiarity deficits in amnesia: convergence of

remember-know, process dissociation and receiver operating charac-

teristic data, Neuropsychology 12 (1998) 323–339.

[59] 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.

Indexing strategic retrieval of colour information with event-related potentials

Documents

Transcript of Indexing strategic retrieval of colour information with event-related potentials