Recognition memory for emotionally negative and neutral words: an ERP study

Similar documents
Material-speci c neural correlates of memory retrieval

In what way does the parietal ERP old new effect index recollection?

Task-switching and memory retrieval processing: Electrophysiological evidence.

ERP correlates of retrieval orientation: Direct versus indirect memory tasks

The relationship between electrophysiological correlates of recollection and amount of information retrieved

Dissociable neural correlates for familiarity and recollection during the encoding and retrieval of pictures

False Memory: P300 Amplitude, Topography, and Latency. Antoinette R. Miller, Christopher Baratta, Christine Wynveen, and J.

Age effects on brain activity associated with episodic memory retrieval Mark, Ruth; Rugg, M.D.

An event-related potential study of the revelation effect

ERP correlates of Remember/Know decisions: Association with the late posterior negativity

ARTICLE IN PRESS. Introduction

Neural Correlates of Human Cognitive Function:

Implicit effects of emotional contexts: An ERP study

The contribution of recollection and familiarity to yes±no and forced-choice recognition tests in healthy subjects and amnesics

Neurophysiological correlates of memory illusion in both encoding and retrieval phases

ARTICLE IN PRESS. Neuropsychologia xxx (2009) xxx xxx. Contents lists available at ScienceDirect. Neuropsychologia

Identifying the ERP correlate of a recognition memory search attempt

Left Anterior Prefrontal Activation Increases with Demands to Recall Specific Perceptual Information

Importance of Deficits

ERP evidence for successful voluntary avoidance of conscious recollection

Neural events that underlie remembering something that never happened

The worth of pictures: Using high density event-related potentials to understand the memorial power of pictures and the dynamics of recognition memory

Electrophysiological dissociation of the neural correlates of recollection and familiarity

Event-related brain potentials distinguish processing stages involved in face perception and recognition

Levels of processing and selective attention e ects on encoding in memory

Attention modulates the processing of emotional expression triggered by foveal faces

Event-related potentials as an index of similarity between words and pictures

Semantic and perceptual effects on recognition memory: Evidence from ERP

The neural basis of the butcher-on-the-bus phenomenon: when a face seems familiar but is not remembered

Remembering the Past to Imagine the Future: A Cognitive Neuroscience Perspective

Mental representation of number in different numerical forms

Resistance to forgetting associated with hippocampus-mediated. reactivation during new learning

Non-conscious recognition of affect in the absence of striate cortex

Electrophysiological evidence of two different types of error in the Wisconsin Card Sorting Test

Independence of Visual Awareness from the Scope of Attention: an Electrophysiological Study

Neural correlates of conceptual implicit memory and their contamination of putative neural correlates of explicit memory

Figure 1. Source localization results for the No Go N2 component. (a) Dipole modeling

A study of the effect of auditory prime type on emotional facial expression recognition

Remember/Know Judgments Probe Degrees of Recollection

Age-related changes in source memory retrieval: an ERP replication and extension

Evaluating Models of Object-Decision Priming: Evidence From Event-Related Potential Repetition Effects

Overt vs. Covert Responding. Prior to conduct of the fmri experiment, a separate

Conceptually driven encoding episodes create perceptual misattributions

BRIEF REPORTS Modes of cognitive control in recognition and source memory: Depth of retrieval

Henry Molaison. Biography. From Wikipedia, the free encyclopedia

Emotional arousal enhances lexical decision times

An electrophysiological signature of unconscious recognition memory

Synchronous cortical gamma-band activity in task-relevant cognition

The Impact of Emotion on Perception Bias or Enhanced Processing?

Prefrontal cortex and recognition memory Functional-MRI evidence for context-dependent retrieval processes

On the relationship between recognition familiarity and perceptual uency: Evidence for distinct mnemonic processes

Neural correlates of retrieval processing in the prefrontal cortex during recognition and exclusion tasks

Examining Event-Related Potential (ERP) Correlates of Decision Bias in Recognition Memory Judgments

Discovering Processing Stages by combining EEG with Hidden Markov Models

Comparing event-related and epoch analysis in blocked design fmri

Neuroscience of Consciousness II

The influence of criterion shifts on electrophysiological correlates of recognition memory. N. Azimian-Faridani* & E.L. Wilding. School of Psychology

Activation of brain mechanisms of attention switching as a function of auditory frequency change

Frontal steady-state potential changes predict long-term recognition memory performance

Reward prediction error signals associated with a modified time estimation task

Viewing a Map Versus Reading a Description of a Map: Modality-Specific Encoding of Spatial Information

A systems neuroscience approach to memory

Woollams et al. 1. In Press, Journal of Cognitive Neuroscience. ERPs associated with masked priming of test cues reveal multiple potential

Immediate and delayed stimulus repetitions evoke different ERPs in a serial-probe recognition task.

Event-Related fmri and the Hemodynamic Response

The role of selective attention in visual awareness of stimulus features: Electrophysiological studies

The influence of directed attention at encoding on source memory retrieval in the young and old: An ERP study

NeuroImage 49 (2010) Contents lists available at ScienceDirect. NeuroImage. journal homepage:

Multiple Forms of Learning Yield Temporally Distinct Electrophysiological Repetition Effects

Accounts for the N170 face-effect: a reply to Rossion, Curran, & Gauthier

Source memory retrieval is affected by aging and prefrontal lesions: Behavioral and ERP evidence

Distinguishing between Category-based and Similarity-based Induction

Title of Thesis. Study on Audiovisual Integration in Young and Elderly Adults by Event-Related Potential

MEASURING CONSCIOUS MEMORY 0

October 2, Memory II. 8 The Human Amnesic Syndrome. 9 Recent/Remote Distinction. 11 Frontal/Executive Contributions to Memory

Brainpotentialsassociatedwithoutcome expectation and outcome evaluation

Psychology Midterm Exam October 20, 2010 Answer Sheet Version A. 1. a b c d e 13. a b c d e. 2. a b c d e 14. a b c d e

SOMETHING ALWAYS STICKS? HOW EMOTIONAL LANGUAGE MODULATES NEURAL PROCESSES INVOLVED IN FACE ENCODING AND RECOGNITION MEMORY AUTHOR'S COPY ABSTRACT

Context-dependent repetition effects on recognition memory

encoding and predict subsequent memory. Michael Griffin, Melissa DeWolf, Alexander Keinath, Xiaonan Liu and Lynne Reder* Carnegie Mellon University

Early posterior ERP components do not reflect the control of attentional shifts toward expected peripheral events

ERP correlates of item recognition memory: Effects of age and performance

Brain Potentials Reflect Behavioral Differences in True and False Recognition. Tim Curran Department of Psychology, Case Western Reserve University

WHAT HOLDS YOUR ATTENTION? THE NEURAL EFFECTS OF MEMORY ON ATTENTION. Emily Leonard Parks

An electrophysiological analysis of modality-specific aspects of word repetition

Recognition of facial emotion in nine individuals with bilateral amygdala damage

Implicit memory influences the allocation of attention in visual cortex

Effects of attention and confidence on the hypothesized ERP correlates of recollection and familiarity

Final Summary Project Title: Cognitive Workload During Prosthetic Use: A quantitative EEG outcome measure

Dual Mechanisms for the Cross-Sensory Spread of Attention: How Much Do Learned Associations Matter?

Brain potentials associated with recollective processing of spoken words

An ERP Examination of the Different Effects of Sleep Deprivation on Exogenously Cued and Endogenously Cued Attention

A Basic-Systems Approach to Autobiographical Memory David C. Rubin

Does scene context always facilitate retrieval of visual object representations?

The Mechanism of Valence-Space Metaphors: ERP Evidence for Affective Word Processing

Electrophysiological Indices of Target and Distractor Processing in Visual Search

Self-construal priming modulates visual activity underlying global/local perception

Imaginal, Semantic, and Surface-Level Processing of Concrete and Abstract Words: An Electrophysiological Investigation

Online Publication Date: 15 th July 2012 Publisher: Asian Economic and Social Society

The time-course of intermodal binding between seeing and hearing affective information

Transcription:

Neuropsychologia 38 (2000) 1452±1465 www.elsevier.com/locate/neuropsychologia Recognition memory for emotionally negative and neutral words: an ERP study Elizabeth J. Maratos*, Kevin Allan, Michael D. Rugg Institute of Cognitive Neuroscience and Department of Psychology, University College London, 17 Queen Square, London WC1N 3AR, UK Received 12 July 1999; received in revised form 27 March 2000; accepted 3 April 2000 Abstract Scalp recorded event-related potentials were used to investigate the neural activity elicited by emotionally negative and emotionally neutral words during the performance of a recognition memory task. Behaviourally, the principal di erence between the two word classes was that the false alarm rate for negative items was approximately double that for the neutral words. Correct recognition of neutral words was associated with three topographically distinct ERP memory `old/new' e ects: an early, bilateral, frontal e ect which is hypothesised to re ect familiarity-driven recognition memory; a subsequent left parietally distributed e ect thought to re ect recollection of the prior study episode; and a late onsetting, right-frontally distributed e ect held to be a re ection of post-retrieval monitoring. The old/new e ects elicited by negative words were qualitatively indistinguishable from those elicited by neutral items and, in the case of the early frontal e ect, of equivalent magnitude also. However, the left parietal e ect for negative words was smaller in magnitude and shorter in duration than that elicited by neutral words, whereas the right frontal e ect was not evident in the ERPs to negative items. These di erences between neutral and negative words in the magnitude of the left parietal and right frontal e ects were largely attributable to the increased positivity of the ERPs elicited by new negative items relative to the new neutral items. Together, the behavioural and ERP ndings add weight to the view that emotionally valenced words in uence recognition memory primarily by virtue of their high levels of `semantic cohesion', which leads to a tendency for `false recollection' of unstudied items. 7 2000 Elsevier Science Ltd. All rights reserved. Keywords: Emotional memory; Familiarity; Recollection; Episodic memory; Semantic cohesion 1. Introduction * Corresponding author. E-mail address: e.maratos@ucl.ac.uk (E.J. Maratos). Numerous studies have demonstrated that performance on episodic memory tasks is in uenced by the emotional nature of the test items. Notably, recall of emotionally negative material (and, in some studies, positive material also) is enhanced relative to the recall of neutral items [5,23,26]. It has been suggested that di erences in memory performance for emotional and non-emotional material arise through multiple mechanisms, two of the most important of which are `arousal', and `semantic cohesiveness' [22]. Items that are arousing (i.e. items that elicit signi cant autonomic responses as indexed, for example, by skin conductance) gain a mnemonic advantage over non-arousing items through the enhancement of their encoding and consolidation in memory. This advantage seems likely to be mediated at the neural level by the modulatory in uence of the amygdala on hippocampal and cortical components of the network supporting episodic memory [10,22]. By contrast, non-arousing emotional items (e.g. words with negative emotional valence) in uence memory largely because, unlike emotionally neutral items, they tend to belong to categories that are semantically `cohesive', that is, categories in which the constituent items share strong inter-item associations. The foregoing distinction provides a useful framework in which to view ndings of studies comparing 0028-3932/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S0028-3932(00)00061-0

E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 1453 memory for emotionally valenced and emotionally neutral words. With the exception of certain classes of item such as `taboo' words, emotional and nonemotional words di er little in their arousing properties [19] raising the possibility that di erences in memory for these two word classes are due mainly to di erences in semantic cohesiveness. Consistent with this possibility, Phelps and LaBar (cited in LaBar and Phelps, [19]) found that when inter-item associations were controlled, the recall advantage normally found for emotionally valenced words over neutral items was eliminated. Also consistent are ndings showing that the above-mentioned recall advantage does not extend to tests of recognition memory (e.g. see [5,20]). In one of these studies [5], discrimination (d') between studied (old) and unstudied (new) words was lower for emotional than for neutral items (hit and false alarm rates were not reported). In the second study [20], emotional words were associated with a higher hit rate but, because this e ect was o set by an equivalent elevation of the false alarm rate, recognition accuracy did not di er for the two word types. The disparate pattern of ndings for recall and recognition t well with the idea that emotional valence exerts much of its e ect on memory for words through the mechanism of semantic cohesion. The fact that emotional words generally belong to a relatively `closed' semantic category means that recall for these items bene ts for the same reasons that underlie the recall advantage for words from any semantically categorised set relative to uncategorised words [2]. In the case of recognition, the ndings of Leiphart et al. [20] suggest that emotionally valenced words may act like the `related lures' that elicit high false alarm rates in studies of `false recollection'. In these studies (e.g. [24,25,35] see [15] for review) subjects typically study lists of semantically related items and subsequently attempt to discriminate between these items and two kinds of new word Ð semantic associates of studied words (related lures), and words semantically unrelated to items that had been shown at study (unrelated lures). Relative to unrelated lures, related lures generate high false alarm rates, with subjects reporting many of these responses to be based on a recollection of the lure as a member of the study list (`Remember' responses, e.g. [25]). Neuroimaging and electrophysiological evidence suggest that the illusory recollection of related lures relies on much of the same neural circuitry that underlies veridical recollection [7,14,34,35]. Since sets of emotionally valenced words tend to be semantically and associatively related, it is easy to see how these items might also engage processes supporting false recollection. According to the foregoing analysis, therefore, emotionally valenced words do not exert their e ect on memory by virtue of their valence per se unless they are arousing. Rather, they do so because emotionally valenced words have, on average, stronger inter-item associations than do sets of unselected neutral words. Thus, memory for emotional words is mediated not by systems or processes to which these items have privileged access, but by the same cognitive operations that support memory for non-emotional material. In the present experiment, we assess the foregoing proposal by investigating the neural correlates of recognition memory for emotionally valenced and neutral words using event-related brain potentials (ERPs). In general terms, the proposal that recognition memory for the two classes of word engages equivalent cognitive operations can be assessed by determining whether the memory-related ERP e ects they elicit di er with respect to scalp distribution. If the e ects for emotional and neutral words do di er in this respect, it would indicate that memory for the two kinds of material is neurally (and, therefore, most likely functionally) dissociable [29]. Such a nding would be inconsistent with the proposal that memory for emotional and neutral words relies on functionally equivalent memory mechanisms. The results of previous research investigating the ERP correlates of recognition memory allow predictions arising from the `semantic cohesiveness' hypothesis to be addressed more speci cally. It has long been known that, relative to new words, ERPs elicited by correctly classi ed old words in tests of recognition memory are more positive-going Ð the so-called ERP `old/new' e ect (see [33] for review). Recent work (for review see [28]; see also [31]) has led to the identi cation of three old/new e ects which appear to index functionally distinct aspects of recognition. For present purposes, the most important e ect is one which onsets around 400±500 ms post-stimulus, and is maximal over the left parietal scalp (the `left parietal' old/ new e ect). The e ect appears to be a neural correlate of the episodic retrieval (recollection) of study items [39] and, in studies of false recollection [7,14] is elicited by both genuinely old words and related lures. Another old/new e ect involves an earlier (ca. 300±500 ms), bilateral shift with a frontal distribution. Rugg et al. [31] proposed that this e ect was a neural correlate of familiarity-based recognition. This is a form of recognition memory held by some authors to be independent of recollection, and to underlie recognition judgements associated with `Know' rather than `Remember' responses [9,38]. The third old/new e ect onsets quite late (ca. 500±700 ms), persists for a second or more, and is distributed over the right frontal scalp [6,39]. The `right frontal' old/new e ect is held to re ect processes that operate on the products of memory retrieval [28]. If, as suggested above, unstudied emotionally valenced words in tests of recognition memory act like

1454 E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 associative lures in studies of false recollection, these items should be di cult to discriminate from studied emotional items because of their tendency to elicit `recollection' of their study presentation. As a consequence, unstudied emotional items should be associated with an elevated false alarm rate relative to unstudied neutral words. Furthermore, even on trials on which unstudied emotional items are correctly classi ed as new, the tendency of these items to elicit illusory recollection might be expected to both impede the decision to respond `new' and, critically, to be manifest in a pattern of neural activity Ð the left parietal ERP e ect Ð signifying the engagement of neural systems supporting episodic retrieval. Thus, relative to the left parietal old/new e ect elicited by neutral words, the e ect elicited by emotionally valenced words should be smaller in magnitude. Furthermore, the di erence in the magnitude of the left parietal e ects elicited by the two word types should be carried mainly by the ERPs elicited by new items, those elicited by emotional words exhibiting the greater positivity. We tested these predictions by recording ERPs while subjects discriminated between studied and unstudied emotionally negative and neutral words. For the reasons already noted, we expected negative words to give rise to an excess of false alarms, a smaller left parietal old/new e ect, but no evidence for the recruitment of memory mechanisms additional to those engaged by neutral items. Further, to the extent that emotional valence in uences recognition memory exclusively through its e ects on recollection, the ERP old/new e ect held by Rugg et al. [31] to index familiarity-driven recognition should be una ected by this variable. 2. Method 2.1. Subjects Twenty young male and female right-handed subjects were employed in the study. Each subject gave informed consent prior to participation in the study and all were remunerated at the rate of 5 per hour. Four subjects' data were discarded prior to data analysis due to excessive electro-oculographic artefact leading to a failure to provide 16 or more artefact free trials for one or more of the ERP categories. The 16 subjects whose data were analysed consisted of 11 females and ve males. 2.2. Experimental material The critical stimuli consisted of 224 words selected from The Balanced A ective Word Project (a corpus of words normed for emotional valence; [36]). Half of these words were emotionally negative (e.g. Fright), whereas the remainder were emotionally neutral (e.g. Mention). Five neutral bu er words were also used. The words varied in length between three and nine letters. All words were semantically distinct from each other in the sense that derivatives of words from a common root were not used. Two study lists (A and B) were created, each containing a set of 56 negative and 56 neutral words. The words in the study lists were randomly ordered and a neutral bu er word was added to the beginning and end of each list. Two test lists (1 and 2) were created using all 224 negative and neutral critical words in two randomised orders, with a bu er word added to the beginning of both lists. 2.3. Procedure The experiment took the form of a single study-test cycle. The combination of study and test list used was counterbalanced across subjects, thus ensuring there was no correlation between word type and old/new status. The stimuli were presented in central vision on a computer monitor (white upper case words on a dark screen). Each word subtended a maximum vertical visual angle of approximately 0.48 and a maximum horizontal angle of approximately 1.28. Study words were presented with their rst letter replacing the xation character. Test words were presented with the third letter replacing the xation character. Subjects were tested in a sound attenuated recording booth. Prior to the experiment, subjects were tted with an electrode cap. Subjects were informed that the experiment consisted of two parts but they were not informed that the second part would involve a memory test. An interval of around 5 min separated the study and test phases, and during this time subjects were instructed to relax. 2.4. Study phase Each trial consisted of a presentation of the word for 300 ms, followed immediately by the restoration of the xation character. This character remained in view until the next trial was initiated (see below). Subjects were required to give an a ective rating to each study word using a scale that ranged from 3 (negative) through 0 (neutral) to +3 (positive). They were instructed to make positive ratings to words they associated with feelings of happiness, content or satisfaction, negative ratings to words they associated with unpleasant feelings of sadness, anger or anxiety, and neutral ratings for words which were neither negative

E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 1455 nor positive. The experimenter initiated each trial after recording the response to the preceding item. 2.5. Test phase Subjects were presented with one of the two test lists. Each trial consisted of the presentation a xation character for 2100 ms, followed by a 124 ms period during which the screen was blank. This period was followed by the display of a test item for 300 ms, following which the monitor was blanked for 2700 ms. Subjects were required to judge whether or not each word had appeared previously in the study phase. Responses were made by button press using one or other index nger. The mapping of keys to response type was counterbalanced across subjects. Subjects were instructed to relax, to keep body movements to a minimum, and to blink only when the xation character was in view. The test phase was computer controlled. Responses faster than 300 ms or slower than 2500 ms were treated as errors. 2.6. Event-related potential recording EEG was recorded from 25 tin electrodes embedded in an elasticated cap and from an electrode placed on the right mastoid process. All channels were referenced to a left mastoid electrode, and re-referenced o -line to represent recordings with respect to linked mastoids. EOG was recorded bipolarly from electrodes placed above the supraorbital ridge of the right eye and on the outer canthus of the left eye. EEG electrodes were located according to the International 10-20 system [13]. These were located over the midline (Fz, Cz, Pz), and at the following additional locations: lateral frontal (FP1/FP2, F3/F4, LF/RF (75% of the distance from Fz to F7/8), and F7/F8), central/anterior temporal (C3/C4, LT/RT (75% of the distance from Cz to T3/4), and T3/T4), parietal/posterior temporal (P3/P4, LP/RP (75% of the distance from Pz to T5/6), T5/T6), and occipital (O1, O2). Data were sampled at a rate of 8 ms per point and digitised with 12 bit resolution. The duration of the recording epoch was 2048 ms with a 104 ms pre-stimulus baseline period. All channels were ampli ed with a bandpass of 0.032±35 Hz. Trials on which base to peak EOG activity exceeded 98 mv were rejected, as were trials on which A/D saturation occurred, or on which baseline drift across the recording epoch (i.e. amplitude of the rst point minus the amplitude of the last point of the epoch) exceeded 255 mv in any EEG channel. Table 1 Behavioural data. Columns 1 and 2: mean proportion (standard deviation in brackets) of hits and correct rejections to each word type, column 3 and 4: discrimination index (Pr) and bias indices (Br) Word type Hit CR Pr Br Neutral 0.83 (0.10) 0.86 (0.12) 0.73 0.41 Negative 0.89 (0.07) 0.66 (0.14) 0.57 0.72 ERPs were formed for four response conditions: correctly recognised old neutral words (neutral hits); correctly rejected new neutral words (neutral correct rejections); correctly recognised old negative words (negative hits); and correctly rejected new negative words (negative correct rejections). The waveforms were smoothed with a ve point binomally weighted lter. There were insu cient trials to form ERPs to items associated with error trials (i.e. misses and false alarms). 3. Results 3.1. Behavioural data 3.1.1. A ective ratings The mean rating assigned to the negative words was 1.73, (across subject SD=0.35; range 2.38 to 1.07) whereas the mean rating for the neutral words was 0.52, (across subject SD=0.40; range 0.04 to 1.30). These means di ered reliably (t15=15.28, p < 0.001). For every subject the mean rating given for the negative words was signi cantly lower than that given for the neutral items (minimum t110=11.66, p < 0.001). Hit and correct rejection rates for the neutral and negative items, along with discrimination (ph-pfa, or `Pr') and bias (`Br') indices [37], are shown in Table 1. Pr for neutral items was signi cantly greater than that for negative items (t15=5.01, p < 0.001) 1. Bias also di ered signi cantly between word types (t15=7.45, p < 0.001), such that subjects were more willing to respond `yes' to negative items. Additional t-tests revealed that hit rates and false alarm rates were both signi cantly greater for negative items (t15=2.93, p = 0.01, and t15=6.28, p < 0.001, respectively). Table 2 Mean reaction times (standard deviations in brackets) for hits, correct rejections and false alarms to each word type Word type Hit CR False alarm 1 The same result was obtained when d' rather than Pr was used as a discrimination index [d'=2.36 and 1.81 for neutral and negative words, respectively; t15=4.98, p < 0.001]. Neutral 999 (242) 1097 (267) 1313 (272) Negative 975 (249) 1184 (268) 1188 (292)

1456 E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 Reaction time (RT) data are shown in Table 2. ANOVA of the RTs for hits and correct rejections revealed main e ects of response type (F(1,15)=10.32, p < 0.01), valence (F(1,15) < 17.90, p = 0.001), and an interaction between these factors (F(1,15)=17.46, p < 0.001). Tukey HSD tests revealed that RTs were faster for correct rejections to the neutral rather than the negative items, and that RTs for hits were faster than those for correction rejections in the case of the negative items only. A further test revealed that RTs to false alarms were faster for negative words (t15=2.81, p < 0.01). Fig. 1. Top: Grand average ERP waveforms elicited by correctly classi ed old and new neutral words. Bottom: Grand average ERP waveforms elicited by correctly classi ed old and new negative words. Electrode locations are described in the text. Gain for the EOG channel is 5 lower than for the EEG.

E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 1457 3.1.2. ERPs The mean numbers of trials (range in brackets) contributing to the ERPs for neutral hits, neutral correct rejections, negative hits and negative correct rejections were 38 (22±49), 39 (23±50), 38 (20±50) and 29 (18± 38), respectively. The proportion of trials lost because of artefact was 18, 19, 22 and 20%, respectively. These proportions did not di er signi cantly. Grand average waveforms from selected lateral and midline sites are shown overlaid according to response category and valence in Figs. 1 and 2 respectively. ERPs begin to diverge as a function of response category and word type from around 300 ms post-stimulus onset. ERPs to old words become more positive-going than those to Fig. 2. Top: Grand average ERP waveforms elicited by correctly classi ed neutral and negative new words. Bottom: Grand average ERP waveforms elicited by correctly classi ed neutral and negative old words.

1458 E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 new words, and ERPs to negative items become more positive-going than those to the neutral items. From approximately 500 ms post-stimulus, old/new e ects demonstrate a left posterior maximum. From around 800 ms post-stimulus, the old/new e ects for negative words are no longer evident, whereas those for neutral words begin to exhibit a right frontal maximum which extends for approximately a further 500 ms. ERPs were quanti ed by measuring (with respect to the mean of the pre-stimulus baseline) the mean amplitudes of four consecutive latency regions (300±500 ms, 500±800 ms, 800±1100 ms, 1100±1400 ms). These intervals correspond closely with the regions chosen for analysis in previous studies (e.g. [6,39]). Two sets of analyses were performed on these data. The rst set investigated di erences in the amplitude of ERP e ects associated with the factors of valence and category. The second set tested for di erences in the scalp distribution of the e ects, and whether the distribution of these e ects changed over time. In all ANOVAs degrees of freedom were corrected for nonsphericity by application of the Greenhouse±Geisser procedure [16] and F ratios are reported with corrected degrees of freedom. Data employed in the topographic analyses were rescaled to eliminate the confounding e ects of between-condition and between-epoch di erences in amplitude [21]. 3.2. Mean amplitude analyses To assess amplitude di erences between the conditions, ANOVAs were conducted for each latency region on the data from lateral frontal (F7/8, LF/RF, F3/4), temporal/central (T3/4, LT/RT, C3/4), and parietal (T5/6, LP/RP, P3/4) electrode sites, employing the factors of valence (negative vs neutral), response category (hit vs correct rejection), hemisphere, location (frontal, temporal/central, parietal) and site (inferior, middle, superior). The results of these ANOVAs with respect to the valence and response category factors are shown in Table 3. The ANOVAs were followed up by additional subsidiary ANOVAs as reported below. As can be seen in Table 3, analysis of the 300±500 ms latency range revealed e ects of valence and response category, re ecting the greater positivity of ERPs elicited by negative words and hits respectively. Crucially, while the factor of valence interacted with the factors of location, hemisphere, and site (see Table 3), it did not interact with response category (maximum F = 1.57). A further ANOVA was conducted on the data from the 300±500 ms latency region. This focused on the three sites Ð F3, Fz, F4 Ð at which the putative ERP correlate of familiarity described by Rugg et al. [30,31] was of maximum amplitude in that study. The waveforms from these sites are shown in Fig. 3. The ANOVA revealed signi cant e ects of valence (F(1,15)=26.53, p < 0.001) and response category (F(1,15=7.67, p < 0.025), but no interaction between these factors (max F(1.6 24.4)=2.27, p > 0.1). Separate ANOVAs showed that the e ect of valence was reliable for both old and new items (F(1,15)=18.21, p < 0.001 and F(1,15)=18.68, p = 0.001, respectively), and that the e ect of response category was reliable for both neutral and negative items (F(1,15)=5.95, p < 0.05, and F(1,15)=5.14, p < 0.05, respectively). Table 3 shows that the ANOVA of data from the 500±800 ms region revealed main e ects of valence and response category, along with three interactions (one of marginal signi cance) involving these factors. As can be seen in Fig. 4, old/new e ects in this latency range showed a left posterior maximum for both Table 3 Magnitude analyes: Table summarising the results of ANOVAs during the four latency intervals shown. Only signi cant e ects involving the condition factors (word type, response category) are reported a 300±500 ms 500±800 ms 800±1100 ms 1100±1400 ms NE F(1,15)=31.6, p < 0.001 F(1,15)=8.4, p < 0.025 ± ON F(1,15)=6.9, p < 0.025 F(1,15)=11.1, p = 0.005 ± NE ON F(1,15)=4.2, p = 0.058 F(1,15)=14.3, p < 0.0025 NE CH F(1.6,24)=19.6, p < 0.001 ± ± ON HM ± F(1,15)=22.1 p < 0.001 F(1,15)=7.4, p = 0.01 ON ST ± F(1,15.4)=6.5, p < 0.025 F(1.1,16.1)=4.7, p < 0.05 NE ON ST ± F(1.1,16.2)=9.9, p = 0.005 F(1,15.7)=8.5, p = 0.01 NE CH HM ± F(1.9,28.5)=7.3, p < 0.005 F(1.5,22.6)=4.3, p < 0.05 NE CH ST F(2.6,38.4)=6.6, p < 0.005 ± ± F(2.5,37.2)=3.1, p = 0.05 NE HM ST F(1.8,26.9)=6.6, p < 0.01 F(1.5,22.9)=5.9, p = 0.01 F(1.4,20.4)=5.1, p < 0.05 ON CH HM ± F(1.7,25.3)=6.7, p < 0.01 ± NE ON CH HM ± F(1.4,21.5)=6.0, p = 0.01 F(1.7,25.5)=9.5, p = 0.001 F(1.3,19.9)=4.5, p < 0.05 a ON=old/new (response category), NE=neutral/negative (valence), CH=frontal/anterior/temporal location, HM=hemisphere (left, right), ST=electrode site (Inferior, Mid-lateral, Superior). Signi cant F values of interactions involving the factors of valence and response category are shown in bold.

E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 1459 classes of word. Furthermore, while old/new e ects were larger for the neutral than for the negative words at all scalp regions, the disparity between the e ects varied across regions. In order to assess the reliability of the old/new e ects elicited by each class of word, separate ANO- VAs were conducted on the data for the neutral and negative items. The ANOVA for the neutral items revealed a main e ect of response category (F(1,15)=20.64, p < 0.001), an interaction between response category and hemisphere (F(1,15)=15.10, p = 0.001), and a three-way interaction between response category, location and hemisphere (F(1.5, 22.2)=12.33, p = 0.001). The e ects re ect the robust nature of the old/new e ects elicited by the neutral items, and their left posterior maximum. ANOVA of the data for the negative items revealed interactions between response category and hemisphere (F(1,15)=8.36, p = 0.01), and between response category and location (F(1.3,19.7)=4.35, p < 0.05). These e ects re ect the tendency for the old/new e ects elicited by these items to be left lateralised, and in addition to show an anterior-posterior gradient (see Fig. 4). A nal set of analyses were conducted on the data from the 500±800 ms latency in order to focus speci cally on the left parietal old/new e ect, the putative ERP signature of episodic retrieval (see introduction). An ANOVA was conducted on the data from the left parietal electrode alone. This revealed e ects of valence (F(1,15=5.40, p < 0.05) and response category (F(1,15)=33.90, p = 0.001), as well as a marginally signi cant interaction between these factors (F(1,15)=4.32, p < 0.06). Separate ANOVAs revealed that the e ects of response category were reliable for both the neutral and negative items (F(1,15)=50.05, p < 0.001, and F(1,15)=12.89, p < 0.005, respectively). Two further ANOVAs revealed that the waveforms elicited by the new negative words were more positive than those for the new neutral items (F(1,15)=7.09, p = 0.025), whereas no such e ect was evident for the ERPs to the two classes of old item (F(1,15) < 1). As shown in Table 3, ANOVA of data from the 800±1100 ms region revealed three interactions between the factors of valence and response category, including the four-way interaction between these factors, location and hemisphere. As shown in Fig. 4, these e ects re ect the left posterior maximum of the old/new e ects in this latency range and the apparent absence of such e ects in the ERPs to negative words. As for the previous latency region, old/new e ects in the 800±1100 ms region were assessed separately for neutral and negative words. ANOVA of the data for the neutral items revealed a main e ect of response category (F(1,15)=8.62, p = 0.010), and interactions between response category and hemisphere (F(1,15)=7.04, p < 0.025), and between response category, location and hemisphere (F(1.7,25.8=8.56, p < 0.005). The e ects re ect the left posterior distribution of the old/new e ects for these items. ANOVA of the data for the negative items revealed no e ects involving the factor of response category. Once again, a further set of analyses was directed toward the left parietal electrode alone. An initial ANOVA revealed a signi cant e ect of response category (F(1,15)=19.12, p = 0.001), and a response category valency interaction (F(1,15)=14.62, p < 0.01). Further ANOVAs revealed a reliable old/new e ect for the neutral items (F(1,15)=38.88, p < 0.001) but not for the negative items. A nal pair of ANOVAs indicated that the ERPs for new negative words were more positive than those for new neutral items (F(1,15)=8.63, p = 0.01), whereas the reverse pattern Fig. 3. Grand average waveforms for all word types (neutral; negative; old; new) from a subset of frontal electrode sites (F3, Fz, F4).

1460 E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 was found for the ERPs to old words (F(1,15)=9.02, p < 0.01); (see Fig. 4). ANOVA of the 1100±1400 ms data revealed yet another four way interaction between valence, response category, anterior/posterior locality and hemisphere (Table 3). Fig. 4 shows that old/new e ects in this latency region were con ned largely to ERPs elicited over the right anterior scalp by neutral words. In keeping with the impression given by the gure, a further ANOVA con ned to data from the frontal sites alone gave rise to an interaction between valence, response category, and hemisphere (F(1,15)=8.46, p < 0.025), as well as to an interaction between valence, response category and site (F1.3,19.8=4.62, p < 0.05). ANOVAs performed on these data for the two word classes separately revealed, for the neutral items, interactions between response category and hemisphere, and response category and electrode site (F1,15=8.56, p < 0.01, and F1.3,18.9=4.18, p < 0.025, respectively). These e ects re ect the tendency for the old/new e ects elicited by these words to be greater over the right than the left hemisphere, and at superior rather than inferior sites. For the negative items, the Fig. 4. Di erences in mean amplitude for old negative minus new negative, and old neutral minus new neutral, for the 500±800, 800±1100, and 1100±1400 ms, latency regions. Amplitude measures are averaged over the electrode site indicated and the sites immediately adjacent to it.

E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 1461 Fig. 5. Topographic maps illustrating the distribution of the di erences between ERPs to correctly classi ed old and new neutral words (upper row), and between ERPs to correctly classi ed old and new negative words (lower row), in successive latency regions.

1462 E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 ANOVA revealed no e ects involving the factor of response category. 3.3. Topographic analyses Four sets of topographical analyses were performed. The rst set compared the scalp distributions of the old/new e ects (i.e. the di erences in voltage between the ERPs elicited by correctly classi ed old and new words) for negative and neutral words between 300 and 500 ms. These distributions are illustrated in Fig. 5. ANOVA (factors of electrode site and valence) of these data revealed no evidence of a valence site interaction (F = 1), indicating that the two scalp distributions are statistically equivalent. A second analysis contrasted the topographies of the foregoing old/new e ects with the e ects of valence in the same latency range, that is, over the time interval in which the analyses of the mean amplitudes between 300 and 500 ms suggested that old/new and valence e ects were additive. The old/new e ects consisted of di erence scores (old±new) collapsed over the factor of valence. Valence e ects consisted of the di erence scores (negative Ð neutral) collapsed over the factor of response category. The topographies of the two e ects are illustrated in Fig. 6, where it can seen that each displays a mid-frontal maximum. Unsurprisingly, given their similarity, an ANOVA contrasting these topographies gave rise to a site e ect (F(3.6,53.7)=7.38, p < 0.001), but not to a site condition interaction (F < 1). A third analysis compared the topographies of the old/new e ects elicited by neutral and negative items in the 500±800 ms latency region (i.e. the only region post-500 ms in which e ects were reliable for the negative words). The scalp distributions of these e ects are illustrated in Fig. 5, where it can be seen that both show a strong left posterior focus. ANOVA of these data revealed a reliable site e ect (F(3.9, 58.7)=4.40, p < 0.005, but no evidence of a site valence interaction (F = 1.3). The nal set of analyses compared the scalp distributions of the old/new e ects elicited by the neutral words as a function of latency region, in order to con- rm that the e ects do indeed re ect the contributions of generator con gurations that di er over time. The comparison was conducted across the three latency regions Ð 300±500, 500±800, and 1100±1400 ms Ð in which old/new e ects appeared to show the most disparate distributions (see Fig. 5). ANOVA revealed a signi cant site by latency region interaction (F(4.9, 73.3)=4.06, p < 0.005), indicating that the old/new e ects for the neutral words do indeed di er with time. As shown in Fig. 5, these distributions initially evolve from a fronto-central to a left parietal maximum, and culminate in a right frontal focus. 3.4. Summary of ERP results Reliable old/new e ects were elicited by negative and neutral words in both the 300±500 and 500±800 ms latency regions. In the rst of these regions, the e ects had a fronto-central scalp distribution, whereas in the later region their topography showed a left parietal maximum. Neither the magnitude nor the scalp topography of the old/new e ects di ered according to Fig. 6. Topographic maps illustrating the distribution of the di erences between correctly classi ed old and new items (left), and between neutral and negative items (right), in the 300±500 ms epoch.

E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 1463 valence in the 300±500 ms region. In the 500±800 ms region, the scalp topography of the old/new e ects elicited by the two classes of word was again equivalent, but the magnitude of the e ects was greater for the neutral items. This di erence was largely due to the enhanced positivity of the ERPs to new negative items, rather than an attenuation of the waveforms elicited by old negative words. From 800 ms onwards, old/new e ects were reliable for the neutral items only, and evolved with time from a left posterior to a right frontal focus. 4. Discussion 4.1. Performance data Recognition memory was poorer for negative than for neutral items. Crucially, this di erence in recognition accuracy was entirely due to a marked di erence in the rate of false alarms elicited by the two classes of word (34 vs 14% for negative and neutral items, respectively). These ndings are broadly consistent with previous research investigating recognition memory for emotionally valenced words [5,20] and are well accounted for by the `semantic cohesiveness' hypothesis outlined in the introduction. Speci cally, it is proposed that the excessive false alarm rate for negative words re ects the same mechanisms that give rise to false recollection e ects in studies investigating recognition memory for semantic associates of emotionally neutral study items [25]. The RT data add support to the foregoing hypothesis. Subjects were some 100 ms quicker to respond correctly to new neutral words than to new negative words. By contrast, false alarm responses to negative words were 125 ms faster than those to neutral words. This pattern of results presumably re ects the propensity of new negative items to elicit information which is either di cult (in the case of correct rejections) or impossible (in the case of false alarms) to discriminate from that associated with a veridical memory of the study episode. 4.2. ERP data Between 300 and 500 ms post-stimulus ERPs were modulated by both the valence and the study status of the test items. Crucially, however, these two factors did not interact with respect to either the magnitude of their e ects, or their scalp topographies. To the extent that Rugg et al. [31] are correct in their proposal that frontally distributed old/new e ects in this latency range re ect familiarity-driven recognition rather than episodic retrieval (recollection), these ndings suggest that valence has little e ect on familiarity. This conclusion adds weight to the proposal (see below) that valence in uences recognition memory primarily through its e ects on recollection. Valence exerted a marked e ect on ERPs between 300 and 500 ms. Speci cally, neutral items elicited a fronto-centrally distributed negative-going de ection, peaking around 400 ms, that was approximately 3 mv greater than the de ection elicited by negative words. One possibility is that this e ect re ects the modulation of one or more of the generators of the `N400', an ERP component well known for its sensitivity to semantic relatedness and association [3,12,18,32]. According to this hypothesis, the high level of interitem association between members of the negatively valenced word set meant that these items tended to prime one another semantically, leading to an attenuation of N400 in a fashion similar to that seen in studies of semantic priming (e.g. [12,32]). To the extent that this account is correct, it underscores the di erences that exist with respect to semantic `cohesiveness' between otherwise unselected sets of emotionally negative and neutral words. As is evident from Fig. 6, the scalp topographies of the old/new and valence e ects in the 300±500 ms latency range were remarkably similar. While this nding does not rule out the possibility that there exist brain regions, undetectable by the ERP method, which respond di erentially to the two variables, it suggests that valence and study status modulate a common generator population. According to this suggestion, therefore, these generators are sensitive both to `semantic' (valence) and `episodic' (old vs new) factors. A discussion of possible reasons for this observation can be found in Ref. [27]. 4.3. Left parietal old/new e ect Neutral words elicited a prominent left parietally distributed old/new e ect which extended from approximately 500 to 1200 ms post-stimulus. In the early part of the same latency range, negative words also elicited a reliable left parietal e ect. While this latter e ect was topographically indistinguishable from that elicited by the neutral items, it was both smaller in magnitude and considerably more short-lived. Crucially, the smaller magnitude of the left parietal e ect elicited by negative items was carried (wholly, in the 500±800 ms latency range, and partially, in the later time region) by the greater positivity of the ERPs elicited by new negative words relative to those elicited by new neutral words. For the reasons outlined in the introduction, we interpret this pattern of results as a re ection of the capacity of new negative items to elicit spurious or `illusory' episodic memories. Presumably, the fact that this e ect was observed in the ERPs elicited by items which were correctly (albeit relatively

1464 E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 slowly) classi ed as new means that the information `retrieved' on these trials was su ciently distinct from that elicited by genuinely old words to permit a false alarm to be avoided. Had there been su cient trials to allow ERPs with adequate signal-to-noise ratio to be formed for false alarm trials, we assume that false alarms to negative items would have elicited a left parietal e ect as large as or larger than that elicited by correctly rejections [7,15]. Indeed, this prediction necessarily follows from the interpretation given above for the e ects of emotionality on the left parietal e ect, and a failure to obtain the nding would call that interpretation into question. 4.4. Right frontal old/new e ect A nal old/new e ect elicited by the neutral words took the form of a late-onsetting, sustained right frontal positive shift. No such e ect was observed in the ERPs to the negative words. Instead, the ERPs elicited by both old and new negative items in this latency range over the right frontal scalp were only slightly smaller in amplitude than those elicited by the old neutral items (see Fig. 5). `Right frontal' e ects similar to the one observed here have been described in several previous studies (for review see [28]). The e ect has been interpreted functionally as a re ection of `post-retrieval' operations carried out on the products of memory retrieval, which are brought into play when retrieved information must be `monitored' with respect to its relevance to current behavioural goals [17]. It has been further proposed [39] that the right frontal old/new e ect re ects di erential neural activity in the same regions of the right prefrontal cortex that are frequently activated in functional neuroimaging studies of memory [4,8], and which are hypothesised by some authors to be the neural substrate of memory monitoring operations (e.g. [11,30]). Although not predicted, the ndings regarding the right frontal e ect in the present study can be accounted for fairly comfortably within the framework outlined above. Unlike in `standard' tests of yes/no recognition memory (when right frontal old/new e ects tend to be small or non-existent [1]) the mere retrieval of episodic information was not a reliable guide as to whether a test item was old or new. Rather, as in tests of source memory, it was necessary for subjects to evaluate the content of the `retrieved' information in order to ascertain the likelihood that it represented a veridical episode from the study phase (a similar argument was advanced by Schacter et al. [35] to account for their observation of right prefrontal activation during the processing of associative lures in a PET study of false recollection). Thus, post-retrieval monitoring operations were engaged whenever a test item elicited episodic information from memory. The only class of items in the present experiment for which such operations were not necessary, therefore, were the new, neutral items. Hence, neutral items, but not negative ones, elicited a right frontal e ect. 5. Concluding comments To summarise, we found that ERP old/new e ects di ered according to the emotional valence of the words that elicited them. We further found, however, that these di erences involved modulations of a common set of memory-related e ects, which, along with the behavioural ndings, could be understood on the assumption that negatively valenced words share higher levels of inter-item associations than do words of neutral valence (the semantic cohesion hypothesis of Phelps et al. [22]). There was no evidence that recognition memory for the two di erent word classes engaged qualitatively distinct neural systems. With the caveat that accompanies any conclusion based on a null result, the ndings suggest that emotionally negative words engage memory retrieval operations that are neither functionally nor neurally distinct from those that support the retrieval of emotionally neutral items. Acknowledgements EJM is supported by the UK Medical Research Council. MDR and KA are supported by the Wellcome Trust. This study was conducted while the authors were in the School of Psychology, University of St Andrews. References [1] Allan K, Wilding EL, Rugg MD. Electrophysiological evidence for dissociable processes contributing to recollection. Acta Psychologia 1998;98:231±52. [2] Baddeley AD. Human memory: theory and practice. Hove: Psychology Press, 1997. [3] Bentin S, McCarthy G, Wood CC. Event-related potentials, lexical decision and semantic priming. Electroencephalography and Clinical Neurophysiology 1985;60(4):343±55. [4] Buckner RL, Koutstaal W. Functional neuroimaging studies of encoding, priming, and explicit memory retrieval. PNAS 1998;95:891±8. [5] Danion J, Kau mann-muller F, Grange D, Zimmermann M, Greth P. A ective valence of words, explicit and implicit memory in clinical depression. Journal of a ective disorders 1995;34:227±34. [6] Donaldson DI, Rugg MD. Recognition memory for new associations: electrophysiological evidence for the role of recollection. Neuropsychologia 1998;36:377±95. [7] Duzel E, Yonelinas AP, Mangun GR, Heinze H-J, Tulving E.

E.J. Maratos et al. / Neuropsychologia 38 (2000) 1452±1465 1465 Event-related brain potential correlates of two states of conscious awareness in memory. PNAS 1997;94:5073±978. [8] Fletcher PC, Frith CD, Rugg MD. The functional neuroanatomy of episodic memory. Trends in Neuroscience 1997;20:213± 8. [9] Gardiner JM, Java RI, Richardson-Klavehn A. How level of processing really in uences awareness in recognition memory. Canadian Journal of Experimental Psychology 1996;50:114±22. [10] Hamann SB, Ely TD, Grafton ST, Kilts CD. Amygdala activity related to enhanced memory for pleasant and aversive stimuli. Nature Neuroscience 1999;2:289±93. [11] Henson RNA, Rugg MD, Shallice T, Josephs O, Dolan RJ. Recollection and familiarity in recognition memory: an event-related functional magnetic resonance imaging study. Journal of Neuroscience 1999;19(10):3962±72. [12] Holcomb PJ. Automatic and attentional processing: an event-related brain potential analysis of semantic priming. Brain and Language 1988;35:66±85. [13] Jasper H. The ten-twenty system of the international federation. Electroencephalography and Clinical Neurophysiology 1958;10:371±5. [14] Johnson MK, Nolde SF, Mather M, Kounios J, Schacter DL, Curran T. The similarity of brain activity associated with true and false recognition memory depends on test format. Psychological Science 1997;8:250±7. [15] Johnson MK, Rye CL. False memories and confabulation. Trends in Cognitive Sciences 1998;2:137±45. [16] Keselman HJ, Rogan JC. Repeated measures F tests and psychophysiological research: controlling the number of false positives. Psychophysiology 1980;17:499±503. [17] Koriat A, Goldsmith M. Monitoring and control processes in the strategic regulation of memory accuracy. Psychological Review 1996;103:490±517. [18] Kutas M, Hillyard SA. Reading senseless sentences: brain potentials re ect semantic incongruity. Science 1980;207:203±5. [19] LaBar KS, Phelps EA. Arousal-mediated memory consolidation: role of the medial temporal lobe in humans. Psychological Science 1998;9:490±3. [20] Leiphart J, Rosenfeld JP, Gabrieli JD. Event-related correlates of implicit priming and explicit memory tasks. International Journal of Psychophysiology 1993;15:197±206. [21] McCarthy G, Wood CC. Scalp distributions of event-related potentials: an ambiguity associated with analysis of variance models. Electroencephalography and Clinical Neurophysiology 1985;62:203±8. [22] Phelps EA, LaBar KS, Anderson AK, O'Connor KJ, Fulbright RK, Spencer DD. Specifying the contributions of the human amygdala to emotional memory: a case study. Neurocase 1998;4(6):527±40. [23] Phelps EA, LaBar KS, Spencer DD. Memory for emotional words following unilateral temporal lobectomy. Brain and Cognition 1997;35:85±109. [24] Robinson KJ, Roediger III HL. Associative processes in false recall and false recognition. Psychological Science 1997;8:231±7. [25] Roediger III HL, McDermott KB. Creating false memories: Remembering words not presented in lists. Journal of Experimental Psychology-learning memory and cognition 1995;21:803±14. [26] Rubin RC, Friendly M. Predicting which words get recalled: measures of free recall, availability, goodness, emotionality, and pronunciability for 925 nouns. Memory and Cognition 1986;14:79±94. [27] Rugg MD, Allan K, Birch CS. Electrophysiological evidence for the modulation of retrieval orientation by depth of study processing. Neuropsychologia 2000 (in press). [28] Rugg MD, Allan K. Memory retrieval: an electrophysiological perspective. In: Gazzaniga MS, editor. The cognitive neurosciences. 2nd ed. Cambridge, MA: MIT Press 2000 (in press). [29] Rugg MD, Coles MGH. The ERP and cognitive psychology: conceptual issues. In: Rugg MD, Coles MGH, editors. Electrophysiology of mind: event-related brain potentials and cognition. Oxford: Oxford University Press, 1995. p. 27±39. [30] Rugg MD, Fletcher PC, Allan K, Frith CD, Frackowiak RSJ, Dolan RJ. Neural correlates of memory retrieval during recognition memory and cued recall. Neuroimage 1998;8(3):262±73. [31] Rugg MD, Mark RE, Walla P, Schloerscheidt EM, Birch CS, Allan K. Dissociation of the neural correlates of implicit and explicit memory. Nature 1998;392:595±8. [32] Rugg MD. The e ects of word of word repetition and semantic priming on event-related potentials. Pyschophsyiology 1985;22:642±7. [33] Rugg MD. Event-related potential studies of human memory. In: Gazzaniga MS, editor. The cognitive neurosciences. Cambridge, MA: MIT Press, 1995. p. 1341±56. [34] Schacter DL, Buckner RL, Koustaal W, Dale AM, Rosen BR. Late onset of anterior prefrontal activity during true and false recognition. Neuroimage 1998;6:259±69. [35] Schacter DL, Reiman E, Curran T, Sheng yun L, Bandy D, Mcdermott KB, Roediger HL. Neuroanatomical correlates of veridical and illusory recognition memory: evidence from positron emission tomography. Neuron 1996;17:267±74. [36] Siegle G. The Balanced A ective Word Project. San Diego State University. Web Site http;//www.sci.sdsu.edu/cal/wordlist/words.prn. [37] Snodgrass JG, Corwin J. Pragmatics of measuring recognition memory: applications to dementia and amnesia. Journal of Experimental Psychology 1988;117(1):34±50. [38] Tulving E. Memory and consciousness. Canadian Psychologist 1985;26:1±11. [39] Wilding EL, Rugg MD. An event-related potential study of recognition memory with and without retrieval of source. Brain 1996;119:889±906.