Abstract

Subjects studied pictures of common objects outlined in either red or green and were asked to memorize the objects and their associated colors. Event-related potentials (ERPs) were recorded during subsequent inclusion (i.e. item) and exclusion (i.e. source) memory tasks. The main goal of the experiment was to determine if brain signatures for familiarity and recollection, two behavioral processes thought to account for episodic memory performance, would be observed in the pattern of ERP results. For correctly recognized items, early, posterior old/new effects were recorded (~300–600 ms) that did not differ in magnitude or scalp distribution between item and source memory tasks. A subsequent long-duration occipitally focused negativity (~800 ms peak) was evident in the source but not the item memory task. The ERPs associated with ‘source errors’ in the source memory task also showed robust early old/new effects. However, ‘source error’ ERPs lacked frontal scalp activity compared to those associated with correct source attribution. The data suggest that a recollective response may require frontal involvement whereas a decision based on familiarity may not.

Introduction

One can have the strong feeling that a person's face looks so familiar that that individual must have been encountered at some time in the past. However, the context of that past experience, i.e. how long ago and in what situation, may be extremely difficult to retrieve. On the basis of these kinds of experiences, some memory theorists posit that two states of awareness underlie recognition memory judgments — one, familiarity, a relatively automatic process that might not be under conscious control; the other, recollection, a process requiring effort and conscious deliberation (Mandler, 1980; Jacoby and Dallas, 1981). Further, in addition to being cognitively distinct, some theorists argue that familiarity and recollection are both dependent upon the medial temporal lobe memory system but, in addition, recollection requires the participation of the frontal lobes (Squire, 1994; Squire and Knowlton, 2000).

Support for these arguments has come from a variety of evidence, including behavioral data from the ‘remember (recollection)/know (familiarity)’ paradigm (Tulving, 1985), neuropsychological studies of amnestic individuals with damage to the medial temporal lobe memory system (Cermak et al., 1992), memory studies of patients with localized frontal lobe lesions (Janowsky et al., 1989), event-related brain potential (ERP) investigations (Rugg et al., 1998), and neuroimaging investigations of blood flow during a variety of memory tasks (Schacter and Buckner, 1998; Henson et al., 1999).

The ERP technique is particularly well-suited for the study of these phenomena, as it is non-invasive, does not require averaging over long time periods as is necessary with some neuroimaging techniques, and, most important, is able to track the brain's processing of information in real time (on a millisecond time scale). In addition, this technique provides information concerning the scalp distribution of peaks and troughs in the ERP waveform. Such topographic differences are quite important, as a difference in scalp distribution between two states of awareness, such as familiarity and recollection, implies that different brain generator configurations gave rise to the electrical activity recorded at the scalp. This kind of result would be strong evidence in favor of the view that familiarity and recollection are cognitively distinct, and depend on at least partially non-overlapping brain systems (Squire and Knowlton, 2000). However, the relationship between the two processes could take the form of independence, exclusivity or redundancy (Jones, 1987; Joordens and Merikle, 1993; Knowlton and Squire, 1995), all of which might lead to topographic differences (see Discussion below).

ERP investigators have recorded brain electrical activity in an attempt to disentangle the roles that familiarity and recollection play in a variety of memory-related phenomena (Smith, 1993; Trott et al., 1997, 1999; Rugg et al., 1998; Senkfor and Van Petten, 1998; Wilding and Rugg, 1996, 1997) [reviewed by a number of authors (Johnson, 1995; Rugg, 1995; Friedman and Johnson, 2000)]. In a commonly used paradigm (labeled sequential response), subjects are asked to study verbal items for a subsequent memory test. These items can be presented, for example, in two different lists (Trott et al., 1997), or by male and female voices (Wilding and Rugg, 1996). At test, subjects are asked to discriminate, via choice, speeded and accurate reaction times, between old and new items. For any item that is judged old, the subjects are then asked to provide a source memory judgment, e.g. was the item presented during the first or second list? In male or female voice? In this procedure, the ERPs are averaged to correctly recognized old items (i.e. hits) according to whether the source was or was not correctly judged. As correct retrieval of the initial learning context (i.e. source) is considered a hallmark of recollection, the assumption is made that the electrical activity associated with a hit trial on which the source is also correctly judged reflects brain activity recorded during a recollective response. On the other hand, the ERP associated with a hit trial on which the source is not judged correctly presumably reflects brain activity obtained during a response based on familiarity alone. However, other explanations are viable, such as misattribution of the initial learning context, or the retrieval of ‘non-diagnostic’ contextual information (Mulligan and Hirshman, 1997).

Two kinds of ERP old/new effects have been observed consistently in the sequential response paradigm described earlier, as well as in other direct memory tasks such as the three-button source memory (source 1, source 2, new) (Senkfor and Van Petten, 1998), and two-button exclusion (target old; nontarget old or new) (Wilding and Rugg, 1997) recognition memory paradigms. The first old/new effect, with an onset at ∼300 ms, and a duration of ~400–500 ms, has a posterior scalp distribution and tends to be somewhat left lateralized with words as stimuli. The second, either coincident with (Wilding and Rugg, 1997) or subsequent to (Trott et al., 1997) the posterior old/new effect is of long duration (to the end of the recording epoch) and has a right-sided, prefrontal scalp distribution. Trott et al. (Trott et al., 1999) have interpreted the early, posterior old/new effect as being consistent with the retrieval of item content (perhaps devoid of contextual information). The late, prefrontal old/new effect has been interpreted as reflecting the search for and/or retrieval of contextual information (Wilding and Rugg, 1996). Trott et al. (Trott et al., 1999) have also suggested that item retrieval and contextual retrieval might be subserved, respectively, by medial temporal and right prefrontal cortical mechanisms [see also (Wilding and Rugg, 1996)]. The medial temporal lobe interpretation is based on the fact that in dense amnesia due to medial temporal lobe dysfunction, the early, posterior old/new effect is absent or reduced (Johnson, 1995). The prefrontal interpretation is based, in part, on a large amount of functional neuroimaging data that suggest a role for the right prefrontal cortex in the retrieval of episodic memories (Buckner and Tulving, 1995; Fletcher et al., 1998; Wagner et al., 1998).

Because differences in scalp distribution of an ERP component imply differences in the intracranial generators of those components, one would predict that the scalp distribution of the ERP old/new effect associated with correct source judgments should differ from that associated with incorrect source judgments. This is based on the assumption that retrieving the source correctly involves both familiarity and recollection and hence both the medial temporal and frontal lobes, whereas a failure to retrieve the source involves familiarity only and hence does not receive a contribution from the frontal lobes. However, in the two investigations where this was tested directly (Wilding and Rugg, 1996; Trott et al., 1999) and correctly by normalizing the data (McCarthy and Wood, 1985; Ruchkin et al., 1999), the old/new effects associated with these two behavioral outcomes did not differ in scalp topography. Thus, strong evidence for differential effects of familiarity and recollection on the ERP waveforms has not been obtained to date [however, Rugg and co-workers provided evidence in a paradigm that did not include a source memory component (Rugg et al., 1998)].

To deal with some of the problems associated with threebutton and sequential response source memory paradigms, one can employ the inclusion/exclusion paradigm developed by Jacoby (Jacoby, 1991). In this task, subjects are asked to study a list of words, e.g. some read and some heard, and then make old/new (inclusion) or heard/not heard (exclusion) choice responses at test. During the exclusion test, for not-heard stimuli, two classes of items need to be excluded, those that were read (referred to as nontarget old items), and new items. Hence, a selective response is made to only one class of studied items (target old), while the other class of studied items (nontarget old) or new items receive the same response. Performance on the exclusion task can only be above chance levels if the two classes of old items can be distinguished; old/new discrimination will not be sufficient. In addition, the assumption is made that responding on the basis of familiarity alone will not yield accurate performance, because target old and nontarget old items are equally familiar. In the exclusion test, which pits recollection against familiarity, subjects must make covert source judgments in order to perform accurately. Therefore, accurate performance on the exclusion task is assumed to be based on the retrieval of contextual information. We will refer hereafter to these two types of recognition memory procedures, inclusion and exclusion, as, respectively, item and source memory tasks.

Wilding and Rugg were the first to record ERPs during a source memory exclusion paradigm (Wilding and Rugg, 1997). Subjects in their investigation studied auditorily presented words delivered by either a male or female speaker. At test, subjects were required to respond to one class of old items (either originally spoken in a male or female voice; designated the target) with one response button and to the other category of old items (designated the nontarget) and new items with the other response button. Given sufficient numbers of trials, this procedure enables ERP averages to be formed to correctly recognized targets and nontargets (both old items), correctly rejected new items and new false alarms, as well as to nontarget false alarms and target misses. Wilding and Rugg recorded the early posterior old/new effect, and the late prefrontal old/new effect to correct trials, but not in association with misses and false alarms (Wilding and Rugg, 1997). Thus, Wilding and Rugg concluded that both effects ‘reflect processes associated with the recollection of an item in its study context’ (p. 125).

Two potential difficulties can be raised with respect to the Wilding and Rugg data (Wilding and Rugg, 1997). First, they did not incorporate a simple, old/new recognition or item memory task, so that a direct comparison between the old/new effects from the two tasks could not be made. Hence, they could not have determined whether the two old/new effects were specific to the source paradigm (in which retrieval of context was necessary) or would also have occurred in an item memory paradigm in which source retrieval was unnecessary for above- chance performance. Second, the ERPs to false alarms were averaged across both new and nontarget items that attracted a ‘target’ judgment. As nontarget items are truly old, whereas new items have never been seen before, this averaging scheme may have precluded the possibility of observing a ‘familiarity’ effect for old nontarget items that have been assigned to the incorrect category.

To counteract these shortcomings, the current experimental design included both item and source memory tasks. Subjects studied pictures outlined in either red or green and were asked to remember the pictures and their associated colors for a subsequent memory test. They then received old/new recognition (item) and target/other (source) memory test blocks. At test, all pictures were presented outlined in black. During the item memory test, subjects made speeded and accurate old (either red or green)/new judgments, whereas during source memory they made target/other (nontarget, new) judgments. There were sufficient numbers of error trials to separately evaluate ERP averages for target misses and nontarget false alarms. It was expected that during both item memory and source memory blocks, posterior old/new effects of similar magnitude and scalp distribution would be recorded, whereas brain electrical activity related to source search and/or retrieval would be observed only during the source memory task, or would be dramatically reduced during the item memory task. If familiarity is evident in the ERP waveforms, then it was expected that robust old/new effects would be observed not only for correct target and nontarget old trials, but also for old target and nontarget trials on which the source had not been retrieved (i.e. respectively, target misses and nontarget false alarms). In addition, if familiarity and recollection reflect somewhat independent states of consciousness, based on at least partially non-overlapping brain networks, another expectation was that ERP scalp distribution would differ during the source memory portion of the paradigm between old items associated with correct and incorrect covert source categorizations.

Materials and Methods

Subjects

Sixteen subjects (11 female) between the ages of 20 and 26 were recruited by notices posted within the Columbia Presbyterian Medical Center community. All subjects reported themselves to be in good health and to have no major medical, neurological or psychiatric problems. All subjects signed informed consent, were native English speakers and received payment for their participation.

Stimuli and Procedures

The experimental stimuli were 312 unambiguous line drawings of common objects that were divided into six lists of 52 items each, with lists carefully constructed so that they were equated on category membership, concept agreement, name agreement, familiarity and visual complexity [normative data bases (Snodgrass and Vanderwart, 1980; Berman et al., 1989; Cycowicz et al., 1997)]. Statistical analysis of the variables characterizing the picture sets revealed no significant differences among lists (P > 0.10). An additional 52 pictures from the same normative sources, not used in the experimental phases, were used for practice study and test blocks, and as fillers. The experiment was divided into six phases. Each phase consisted of one study and two test blocks (item recognition, and source recognition). In each phase, one of the six lists of pictures was used, with the order of list presentation randomized across phases separately for each subject. Of the 52 pictures in a list, 32 were randomly assigned to the study block, while the remaining 20 were assigned as foils to the test block.

In the study block, the subject viewed 36 pictures (half outlined in green and half in red), including four fillers, two of which were presented at the beginning and two at the end of the block to avoid primacy and recency effects (subjects were not tested on these fillers). For each subject, the picture's color was randomly assigned. To ensure that the subject attended to the picture's color, s/he was asked whether the picture was red or green by pressing one of two buttons. The subject was asked to memorize both the picture and its associated color for a subsequent memory test.

In the item recognition memory test block, the subject viewed a total of 26 pictures outlined in black; 14 new and 12 old (six previously presented in red and six in green). Subjects were asked to press one response button if the item was old and the other if it was new. The numbers of old items differed between the two tasks in order to approximately equate the probability of an ‘old’ (in the item memory task) or ‘target’ (in the source memory task) response. During item memory, the proportion of old items was 0.46 (12 of 26) while in source memory, the proportion of target items was 0.38 (10 of 26). Hence, in the source memory test block, the subject viewed a total of 26 pictures outlined in black: 6 new and 20 old (10 previously presented in red and 10 in green). In half of these test blocks, red was defined as the target and, in the other half, green was defined as the target. The subject was asked to press one button if s/he thought the picture was seen during study in the target color, and a second button if s/he thought the picture had been presented during study in the nontarget color, or was new (i.e. had not been seen during study). Figure 1 depicts schematically examples of study and test trial sequences.

During study, item memory and source memory blocks, each picture was presented at the center of a computer screen for 500 ms, followed by a 1500 ms period in which a fixation point was shown on the screen, for a total interstimulus interval (ISI) of 2000 ms. In order to ensure that all subjects knew which button to press, cues were presented on the computer screen during the entire block. Hence, during the study block, small rectangles were presented below and to the right and left of the to-be-remembered pictures. In one rectangle the word ‘RED’ appeared in red and in the other the word ‘GREEN’ appeared in green. The left/right positions of these cues reflected the hand assigned to each color. Similarly, during the item memory blocks, ‘OLD’ and ‘NEW’ cues in black lettering appeared in rectangles below and to the right and left of the pictures, again consistent with the assigned hand of response. During source memory test blocks, the cues contained the word ‘OTHER’ in black letters and either ‘RED’ in red letters (when red was the target), or ‘GREEN’ in green letters (when green was the target color). During study, item memory and source memory blocks, subjects made choice, speeded and accurate respectively, ‘red’/‘green’, ‘old’/‘new’ and ‘target’ (old)/other (either nontarget ‘old’ color or new) decisions to each picture. The hands assigned during study to ‘red’ and ‘green’ buttons, during item memory to ‘old’ and ‘new’, and during source memory to ‘target’ and ‘other’ were counterbalanced across subjects. The horizontal visual angles ranged from 0.85 to 4.81°, and the vertical visual angles from 0.56 to 3.40° respectively, for the largest and smallest pictures.

To counterbalance order effects, in half the sessions, item memory preceded source memory testing and, in the other half, source memory preceded item memory testing. Subjects were not informed prior to the study-block which test block would be administered first, or which color would be the target. The sequence of stimuli was separately randomized for each subject.

EEG Recording

EEG (5 s time constant; 50 Hz upper cutoff; 200 Hz digitization rate) was recorded (Sensorium Amplifiers) continuously using an Electrocap (Electrocap International) from 62 scalp sites, including left and right mastoids, referred to nosetip, from extended 10–20 system placements (Nuwer et al., 1998). Vertical EOG was recorded bipolarly from electrodes placed on the supraorbital and infraorbital ridges of the right eye, and horizontal EOG was recorded bipolarly from electrodes placed on the outer canthi of the two eyes. Trials containing eye movement artifact were corrected off-line using the procedure developed by Gratton et al. (Gratton et al., 1983). Trials were epoched off-line with 100 ms pre- and 1900 ms post-stimulus periods.

Data Analyses

ERPs were averaged to correctly recognized old and new items during the item memory test blocks, and to correctly recognized target and nontarget old items, as well as new items during the source memory test blocks. When 10 or more trials were available for a given subject and condition, ERP averages were also computed to error trials (e.g. false alarms and misses during the item memory and source memory test phases).

Maps of surface potential activity were computed using a third-order spherical spline interpolation algorithm (Perrin et al., 1989), which uses the average reference (Scherg, 1990) rather than, for example, the nose or mastoids as reference sites. With this method, the scalp distribution of an ERP component is not influenced as heavily by the reference site used for the original recordings. To compare scalp distributions of ERP surface potential activity between conditions, the data were normalized using the root mean square method described by McCarthy and Wood (McCarthy and Wood, 1985). This manipulation removes overall amplitude differences between conditions to allow a comparison of the shape of the distribution across the scalp. A significant difference in scalp distribution is revealed as an interaction of a particular variable with electrode location.

The BMDP-4V (repeated measures ANOVA) computer program was used for all analyses. The Greenhouse–Geisser epsilon (ε) correction (Jennings and Wood, 1976) was used where appropriate. Uncorrected degrees of freedom are reported below along with the epsilon value; the P values reflect the epsilon correction. Where appropriate, significant main effects and interactions were followed-up with post hoc analyses using the Tukey honestly significant difference (HSD) test. In addition, two types of planned comparisons were performed, described in the Results section below.

Results

Behavioral Data

During the study phase, subjects were highly accurate in detecting red (mean percent correct = 98.4) and green (mean percent correct = 98.2) items. Mean reaction time (RT) did not differ between these two response categories (P > 0.10; mean red = 609.5 ms; mean green = 608.2 ms), suggesting that, during the study phase, items outlined in red and green were processed similarly.

Preliminary analyses indicated that there was no difference in memory test performance between pictures outlined in red and green. Therefore, all subsequent analyses were performed on data collapsed across these two classes of stimuli. Table 1 presents the behavioral data during the item memory and source memory tasks. As can be observed, subjects performed well above chance during both types of memory tests, although discrimination (Pr) (Snodgrass and Corwin, 1988) was significantly lower during the source memory compared to the item memory blocks, as confirmed by t-test [tPr(15) = 6.31, P < 0.0001]. Similarly, the false alarm rate was higher during source memory compared to item memory blocks [tFA(15) = 4.70, P < 0.0001], but there was no difference in the bias measure (Br) (Snodgrass and Corwin, 1988) [tBr(15) = 1.55, P > 0.10]. Values of Br <0.5 indicate that subjects responded conservatively — i.e. they chose to respond ‘new’ when uncertain.

Table 2 presents the response time data. Statistical comparisons were only conducted for correct responses. During item memory blocks, old items were responded to faster than new items [t(15) = 2.18, P < 0.05]. A one-way ANOVA of the RTs during the source memory task followed by post hoc testing revealed that RTs to old target and new items did not differ reliably, but RTs to both were faster than those to nontarget old items.

ERP Data

As for the behavioral data during the study phase, there were no systematic differences between the ERPs elicited by pictures outlined in red and green, suggesting that they were processed similarly by the brain.

Memory Test Waveforms

For the item memory task, the analysis comprised hit (correct old) trials (M = 60.7 trials) and correct rejection (correct new) trials (M = 78.0) but neither type of error trial (false alarms or misses) because they occurred too infrequently to provide stable data. For the source memory task, the analysis comprised target old trials (M = 45.3), nontarget old trials (M = 45.0), new trials (M = 34.1), target incorrect trials (M = 14.1) and nontarget incorrect trials (M = 13.7). There were too few new error trials in the source memory task (i.e. false alarms) to permit statistical analysis.

Figure 2 depicts the grand mean ERPs associated with correct old and new responses during the item (left) and source (right) memory tasks. For ease of viewing, the data are shown at 24 of the 62 scalp locations on a schematic figure of the head. These electrode sites provide good coverage of the scalp, and capture the major effects of interest. For the source memory task, the ERPs elicited by target and nontarget old items are depicted. As can be seen, the waveforms are characterized by a series of early negative and positive visual evoked potential components (largest at the occipital scalp sites) between about 100 and 300 ms that do not appear to differ between the item and source memory waveforms. The ERPs elicited by both new and old items are characterized by a large-amplitude positive component (at ~500 ms peak latency). This positivity is also larger for old compared to new items in both tasks, and most likely can be identified with the posterior old/new effect observed by previous investigators during similar kinds of memory tasks (Wilding and Rugg, 1996; Trott et al., 1999). The ERPs elicited by target and nontarget old items in the source memory task do not appear to differ dramatically.

To facilitate the observance of between-task differences, Figure 3 depicts the grand mean ERPs associated with correct rejections of new items (left) and correct recognition of previously studied old pictures (right) during the item and source memory tasks. The positivity appears to be larger during the item memory task compared to the source memory task for both new and old items. The ERPs associated with old items during the source memory task show a large amplitude, posteriorly focused negativity that is markedly smaller during the item memory task. Finally, based on Figures 2 and 3, there is little evidence of a late-onset, right-lateralized prefrontal old/ new effect in the source memory data, an effect that has been observed previously in source memory tasks (Trott et al., 1999).

To enable easier visualization, Figure 4 depicts some of the same data as shown in Figures 2 and 3 with larger amplitude and broader time scales, at three midline posterior scalp sites, where these effects were largest. The data are segregated by task (left- most column, item memory; middle-column, source memory correct), with old and new waveforms superimposed. Also depicted (in the right-most column) are the ERPs associated with incorrect source memory judgments. For all categories of response, including those associated with incorrect source judgments, old and new waveforms begin to diverge at ~350 ms, with the old/new difference reaching peak amplitude at ~500 ms, and taking the form of a positive-going displacement of the ERP to the repeated (i.e. old) item, hereafter referred to as the early, posterior old/new effect. The later, negative old/new effect, most prominent in the ERPs of the source memory task (hereafter referred to as the late, posterior old/new effect), peaks at ~900 ms post-stimulus.

One potential problem with the analyses of the ERP data from the source memory task is that responses to critical items (e.g. old target versus new; old target correct versus old target incorrect) were made with different hands and with different response probabilities. This means that neither hand of response nor probability of response is equated between either pair of conditions, precluding unequivocal interpretation of the results of these comparisons. However, there are four sets of comparisons in which ERPs were associated with responses from the same hand, and these are depicted in Figure 5. As can be seen, for comparisons of the ERPs elicited by old items (the first two columns), there is a remarkable degree of similarity regardless of whether the ERPs were associated with correct or incorrect responses. Thus, response probability does not seem to be a modulating factor. In addition, it is obvious that hand of response is not modulating the ERP waveforms, as the new/old comparisons (last three columns) are highly similar regardless of whether the same or different hands were used to respond to both classes of items.

Averaged Voltage Analyses

The ERP data were quantified using averaged voltages corresponding to areas of difference observed in the grand mean and individual subject mean ERPs. These spanned the early posterior and late posterior old/new effects. For each of these old/new effects, two windows were defined. For the early old/new effect, these values were 260–360 ms and 415–615 ms [roughly corresponding to the N400 and P3b portions of the waveform observed in many previous ERP studies of recognition memory (Wilding and Rugg, 1996; Senkfor and Van Petten, 1998; Paller et al., 1999; Trott et al., 1999)]. Hereafter, these will be referred to as early1 and early2. For the late old/new effect, the measurement windows were 700–950 ms and 955–1200 ms. Hereafter, these will be referred to as late1 and late2. To assess the effects discerned from visual analysis above, a series of ANOVAs was performed. To capture any anterior/posterior and/or left/right asymmetries, these analyses were performed on the data recorded from 24 scalp sites along saggital (left, midline, right) and anterior/posterior planes. The 24 scalp sites included on the left, FP1, F3, FC3, C3, CP3, P3, PO1 and O1; on the midline, FPz, Fz, Cz, CPz, Pz, Poz and Oz; on the right, FP2, F4, FC4, C4, CP4, P4, PO2 and O2.

Because of the long latency of the Old/New and Task effects observed in Figures 2–5, all of the analyses described below were also performed on two additional measurement intervals, 1205–1500 and 1505–1800 ms. The great majority of significant effects were identical to those observed for the late1 and late2 indices (and will not be discussed further), suggesting that, from ~900 ms onwards, the latency regions encompassed by the late negativity reflected functionally and topographically homogeneous brain activity. The first analysis employed the data for the ERPs associated with correct old and new responses from the item memory task and the ERPs associated with correct target old and new responses from the source memory task. The results of the Task (Item memory, Source memory) by Old/New by Saggital Plane (left, midline, right) by Anterior/Posterior (Frontal Pole, Frontal, Fronto-central, Central, Centro-parietal, Parietal, Parieto-occipital, Occipital) ANOVA on these data are presented in Table 3. Three-way interactions of Task and/or Old/New with Electrode Location (as manifested by interactions with Saggital Plane and Anterior/Posterior Region) are only considered in the scalp topography section below, after normalization of the data to enable unequivocal interpretation (Table 4). The main effects of Saggital Plane and Anterior/Posterior Region were not interpreted as, by themselves, they do not reflect memory- related differences.

The early2 measurement was significantly more positive in the item than in the source memory task. By contrast, the late1 and late2 measurements were significantly more negative in the source memory task. The early2 index was significantly more positive to old than to new items in both tasks, reflecting robust old/new effects. For late1 and late2, Task interacted with Old/New. For both measurements, post hoc testing revealed that old items produced larger amplitudes in the source compared to the item memory task, whereas new items did not differ reliably between tasks. The Task by Anterior/Posterior interaction was reliable for both early1 and early2 intervals. Post hoc tests indicated that the greater positive amplitudes elicited during the source memory task had a posterior scalp distribution centered around centro-parietal scalp; see Fig. 3). Post hoc testing of the Old/New by Anterior/Posterior interaction for the early1 index revealed that the greater positive amplitudes for old items were reliable only for the four frontal sites (fronto-polar to central) and the two most posterior sites. For the early2 window, post hoc testing of this same interaction indicated that all of the differences were significant, but the largest differences were distributed posteriorly. The Task by Anterior/Posterior interactions were significant for both the late1 and late2 indices. Post hoc testing showed that, for both, the greater negative amplitudes elicited during the source task were only reliable at the more posterior electrode sites.

To summarize briefly, the early, posterior Old/New effect did not differ between item and source memory tasks, whereas, for the later negative Old/New effect, greater amplitudes were consistently elicited during the source memory procedure. Moreover, differences between tasks were larger at posterior scalp sites with no evidence of the robust left/right asymmetries or right-sided prefrontal old/new effects that have been observed previously (Wilding and Rugg, 1996; Trott et al., 1999).

In the second analysis, using planned comparisons, the ERPs from the source memory task elicited by old items associated with the same response hand (depicted in the first two columns of Figure 5) were compared between correct and incorrect source judgments. The ANOVAs were constructed identically to those described above. The Target Correct versus Nontarget Incorrect by Saggital Plane by Anterior/Posterior Region revealed no reliable main or interaction effects with Condition, as was also the case for the Nontarget Correct by Target Incorrect by Saggital Plane by Anterior/Posterior Region ANOVA (Fs < 2.69, Ps > 0.10), indicating that these classes of old items did not differ.

In the third analysis, a series of planned comparisons on the data from the source memory task contrasted separately the ERPs associated with old items (target correct, target incorrect, nontarget correct, nontarget incorrect) with the ERPs elicited by correctly categorized new items, in order to determine if each yielded robust old/new effects. Again, the ANOVAs were identical to those detailed immediately above (Old/New by Saggital Plane by Anterior/Posterior Region). With the exception of the early1 window, which did not consistently produce reliable old/new effects, the four ANOVAs revealed significant old/new main effects for each of the remaining three measurement windows [Fs(1,15) > 5.26, Ps < 0.05]. Primarily for the two late indices, the ANOVAs also revealed significant Old/New by Saggital Plane [Fs(2,30) > 4.9, Ps < 0.05] or Old/New by Anterior/ Posterior Region [Fs(7,105) > 5.0, Ps < 0.05] interactions. Post hoc testing of the Old/New by Saggital Plane interaction indicated that the differences between old and new ERPs were greatest at the midline; for the Old/New by Anterior/Posterior Region interaction, post hoc tests revealed that the old/new differences were greatest between parietal and occipital scalp.

In brief, whether an old item was associated with a correct or incorrect source judgment, the early and late posterior indices could not be differentiated on the basis of amplitude. Further, robust old/new effects were found for both correct and incorrect target and nontarget items in the source memory task. Again, no reliable left/right asymmetries or right-sided, prefrontal old/new effects were observed.

Scalp Topography

Figure 6 depicts the old minus new difference waveforms for correct old/new decisions in the item memory task and correct and incorrect judgments in the source memory task. As can be observed, for the early, posterior old/new effect, the waveforms associated with incorrect source judgments show a more posterior scalp distribution than those associated with correct judgments of source, or with correct old/new judgments in the item recognition task. The late, negative old/new effect is markedly larger in the source memory task. Both of these effects can be clearly observed in Figure 7, which depicts the surface potential (SP) maps for the four measurement windows, early1 through late2, and are based on the data depicted in Figure 6. The three rows depict, respectively, the scalp distributions associated with correctly recognized old pictures during the item memory task, correctly recognized old pictures (with correct source judgments) during the source memory task, and old pictures whose source was incorrectly judged. The distributions depicted in the latter two rows were computed on the ERPs collapsed across target and nontarget items. This was done to increase the signal-to-noise ratio and was based on the fact that there were no differences between target and nontargets (whether correct or incorrect) for any of the amplitude analyses. Moreover, separate Target and Nontarget ANOVAs on the normalized data revealed identical differences in scalp distribution between correct and incorrect trials as reported below for the collapsed data. The topography of the early1 phase is characterized by an anterior scalp distribution during both item memory and source memory trials in which the initial color was correctly judged. By contrast, the scalp distribution associated with source memory trials for which color was incorrectly judged shows a more posterior distribution. Similarly, the topographies of the early2 index associated with item and correct source memory trials are more frontally oriented than those associated with incorrect source memory judgments. The distributions for the late1 and late2 measurements associated with both types of source memory trials show highly focused negative activity over parieto-occipital scalp sites and, unlike the two early indices, do not appear to differ for correct and incorrect source memory trials. On the other hand, the late1 index for item memory trials is not nearly as large or as clearly focal as those recorded during the source memory task.

To assess potential differences in scalp distribution, the old minus new averaged voltages whose scalp distributions are depicted in Figure 7 were normalized using the root mean square method described by McCarthy and Wood (McCarthy and Wood, 1985). Table 4 shows the results of the topographic ANOVAs and which Condition by Electrode Location interactions were reliable. In the first topographic analysis, the normalized data for all four indices from the source memory task (comparing rows 2 and 3 of Fig. 7) were subjected to a Response Type (correct, incorrect) by Electrode Location (24 scalp sites) ANOVA. These 24 sites were chosen to be consistent with the amplitude analyses reported earlier. For both early1 and early2 indices the scalp distributions associated with correct source judgments were more frontally oriented (as assessed by post hoc testing) than those associated with incorrect source judgments. The two response types showed no differences in scalp distribution for the two late measurement intervals (Fs < 1).

To determine whether the scalp distributions of the early and late posterior old/new effects differed between the item and source memory tasks, a second analysis was performed, in which the normalized data were compared between the first and second rows depicted in Figure 7 (bottom of Table 4). The two early indices did not differ topographically between tasks, as can clearly be observed in Figures 6 and 7. However, the Task by Electrode Location interaction was significant for the late1 index, but not for the late2 interval. As can be seen by inspection of Figures 5 and 6, although late1 negative activity was clearly posteriorly distributed, the largest differences between tasks (as assessed by post hoc testing) occurred for the positive aspect of the distributions over frontal scalp locations. The item task showed relatively greater positivity at these sites than the source memory task.

In summary, unlike the amplitude data, the topographic data suggest that there is an ERP sign of ‘recollection’, because there was a reliable trend for the early, posterior old/new effects associated with correct source judgments to be more frontally oriented than those associated with incorrect source judgments. As observed in Figure 7, the scalp distribution of the late1 index of negative activity associated with trials on which a covert source judgment had to be made differed from that associated with trials on which no such source judgment was necessary (item memory), suggesting that the processes reflected by this activity differed in the two tasks.

Discussion

The current investigation provides several novel findings concerning the retrieval of episodic information not observed in previous ERP investigations. First, this is the only ERP study (to our knowledge) to compare directly inclusion (item memory) with exclusion (source memory) tasks. Second, although previous investigators have reported findings that have supported the distinction between recollection and familiarity, those conclusions were not accompanied by the strong evidence of scalp distribution differences (as is the case here) between posterior old/new effects associated with the retrieval of source information and those for which correct color information was not retrieved. Third, late negative activity was present only during the source memory paradigm (or was much larger), as evidenced by its large amplitude and different scalp distribution.

The Early, Posterior Old/New effect.

This old/new activity appears highly similar to the posterior old/new effects recorded by others in recognition memory paradigms with (Wilding and Rugg, 1996, 1997; Senkfor and Van Petten, 1998; Trott et al., 1999) and without (Rugg and Nagy, 1989; Friedman et al., 1993) a source memory component. In previous source memory investigations, however, this activity has been larger when associated with a hit trial for which the source was correctly judged compared to a hit trial on which the source was incorrectly judged (Wilding and Rugg, 1996; Senkfor and Van Petten, 1998; Trott et al., 1999). On this basis, Wilding and Rugg concluded that it reflected recollection, as the hallmark of a recollective response is the retrieval of the context associated with the initial learning episode (Wilding and Rugg, 1996). Even though this early old/new activity has been found to be greater for correct than for incorrect source judgments, no differences in scalp distribution of this activity have been observed between the ERPs associated with correct and incorrect source judgments (Wilding and Rugg, 1996; Trott et al., 1999). As some two-process theorists argue for distinct cognitive and neurological aspects for familiarity and recollection, the failure to find topographic differences between trials presumably based on recollection (hit + source correct) and those presumably based on familiarity (hit + source incorrect) argue against an effect of familiarity on the ERP waveforms. In the current data, considering magnitude alone, the early old/new effect also does not appear to qualify as a sign of recollection, as it was as large in association with correct as it was with incorrect source judgments. On the other hand, based on scalp topography, the current data provide more compelling evidence for these two processes — ERPs on trials for which source was correctly attributed showed more frontally oriented scalp topographies than those on trials for which source was incorrectly judged (or not retrieved) during the source memory task.

These data suggest that the more posterior scalp topography associated with error trials might reflect a decision based on familiarity. Although there were instances when only a posterior scalp distribution was observed (error trials during the source memory task; see Fig. 7), there were no instances of a purely frontal scalp topography. Moreover, during item memory blocks, in which both familiarity and recollection are assumed to work in concert (Jacoby et al., 1993), the scalp distribution of the old/new effect associated with correctly recognized pictures showed both frontal and posterior aspects. Therefore, the data argue for a redundant relationship between familiarity and recollection, such that recollection always includes familiarity, but familiarity can occur in the absence of recollection [for discussion see (Jones, 1987; Jacoby et al., 1993; Joordens and Merikle, 1993; Knowlton and Squire, 1995; Cowan and Stadler, 1996)].

For our interpretation to be viable, it must be assumed that the lack of frontal scalp activity indicates a lack of a frontal lobe contribution to the early old/new effect, whereas the presence of such activity on correct trials implies the presence of a frontal lobe generator. Although, based on scalp distribution alone these assumptions are tenuous, the current data are consistent with a large variety of neuropsychological (Schacter, 1987) and neuroimaging data (Nolde et al., 1998) that have demonstrated contributions of the prefrontal cortex to reinstatement of the initial learning context, a critical sign of recollection-based retrieval.

The early old/new effect onset occurred several hundred milliseconds prior to the old/new decision in item memory blocks and the target/nontarget decision in source memory blocks. Hence, the relationship between the onset and peak of the early old/new activity and the generation of mean reaction time raises the possibility that it could be one of the brain events causally related to the decision to respond (Ritter et al., 1972). If this were the case during source memory blocks, these data would suggest that sufficient information about source to render a target/nontarget decision would have been available during the interval when the early old/new effect was active at the scalp. On this basis, the early old/new effect could reflect, at least in part, some aspect of recollection. The fact that this early, posterior old/new activity is absent or reduced in patients with dense amnesia due to lesions of the medial temporal lobes (Johnson, 1995), patients whose ability to recollect is severely compromised, is consistent with this interpretation.

A relationship between the medial temporal and frontal lobes that is consistent with the current data has been advanced by Moscovitch (Moscovitch, 1994). In Moscovitch's model, the output of hippocampally mediated retrieval is relayed to a frontal lobe mechanism. This frontal lobe mechanism is then responsible for working with the information retrieved by the hippocampal system. The frontal lobe system serves to make this information conscious and to place it in its particular spatiotemporal context (i.e. source), data that are not necessarily retrieved by the hippocampal system. In Moscovitch's model, the output of the hippocampal system is associated with an automatic, non-strategic judgment of prior occurrence (Moscovitch, 1994). The frontal aspect of this old/new effect (presumably mediated by prefrontal cortex) would be associated with a strategic, effortful search and/or retrieval of at least some source information. That is, the prefrontal cortex is responsible for organizing strategically the output of the hippocampal system. On this view, the current early, posteriorly distributed old/new effect observed in association with error trials would reflect an obligatory recognition response devoid of contextual information (a ‘pure’ hippocampal retrieval). By contrast, the presence of posterior as well as frontal activity in the scalp distribution associated with correct trials would reflect both an obligatory recognition response and one in which at least some contextual information is retrieved under direction of the prefrontal cortex.

The use of pictures in the current study may have led to the presence of large-amplitude old/new effects in association with error trials. Pictures are known to be remarkably resistant to forgetting, perhaps due to the fact that pictures have distinct sensory codes (Nelson, 1979) and may be encoded both perceptually and semantically (Paivio, 1986). On these bases, pictorial stimuli may be more likely than words to engender an ‘automatic’ recognition response, resulting in robust old/new effects for both target misses and nontarget false alarms.

The Late, Posterior Old/New Effect

Unlike previous ERP studies with a source memory component (Wilding and Rugg, 1996, 1997; Senkfor and Van Petten, 1998; Trott et al., 1999; Wilding, 1999), there was little evidence of a right-lateralized prefrontal old/new effect in these data. Rather, a late-onsetting, symmetrical, parietal-occipital negativity was much more prominent in the source memory compared to the item memory task. The presence of this activity in the current data may be due to differences in experimental paradigms and/or stimulus characteristics. However, the observation of this late, negative old/new effect in the current investigation is most probably not due to the retrieval of pictures per se (as opposed to words) for a few reasons. First, this activity was not observed with old pictures during the item memory task. Second, it was also not observed with new items in either task. Third, previous investigators who also presented pictures during study and test (Friedman, 1990; Schloerscheidt and Rugg, 1997) did not observe this kind of negative brain activity.

In the current study, pictures outlined in color were presented, while in previous studies words were presented either in the auditory or visual modalities. Therefore, the use of a distinct perceptual attribute to define source (i.e. color) may have led to the presence of the large-amplitude occipital negativity (consistent with visual cortical generators) that would have tended to reduce markedly any frontal positive activity that would have been elicited in this task. This negativity was considerably larger during source memory compared to item memory blocks (and showed a different scalp distribution), suggesting that it reflects the activity of brain generators that have something to do with source search and/or retrieval. As the nature of the contextual information was perceptual (rather than semantic, for example), it is reasonable to assume that this kind of information would have been stored in the cortical regions that originally processed the information (i.e. occipital cortex).

The occipitally focused scalp distribution of the current negativity is compatible with cerebral blood flow data reported by Schacter and colleagues (Schacter et al., 1996), who demonstrated that, in addition to activation of the medial temporal region during retrieval, correct recognition of events originally presented aurally was associated with activation of the superior temporal region (i.e. auditory cortex). Taken together, the results are consistent with the principle that information processed in a given cortical area during study is stored as a representation in similar cortical regions, and is subsequently retrieved from those regions, perhaps under the direction of the prefrontal cortex (Squire and Kandel, 1999).

The peak latency of the late old/new effect (between ~800 and 900 ms; Figs 2 and 4) is consistent with its potential role in source search and/or retrieval, as peak latency occurred at about the same time as mean reaction time (see Table 2). As the negative activity onset occurred as the early old/new effect was returning to baseline, one speculative hypothesis is that sufficient time would have elapsed for this late negative activity to have reflected brain activity related to an attempt to reinstate the initial image along with its associated color during source memory blocks (Farah et al., 1988), i.e. a continued search and/or evaluation of the correctness of the behavioral response. This interpretation would be consistent with the fact that the occipitally prominent negative activity was present even in those ERPs that were associated with trials on which an incorrect, covert ‘source’ decision had been made (i.e. target misses and nontarget false alarms).

Conclusions

In general, the current data suggest a series of memory-related brain activities engendered by these item and source recognition memory tasks that are similar in some respects to previously described ERP data (Wilding and Rugg, 1996, 1997; Senkfor and Van Petten, 1998; Trott et al., 1997, 1999). The early, posteriorly distributed old/new effect may reflect a relatively automatic recognition response to material that was previously apprehended consciously (regardless of the correctness of the source judgment). When accompanied by a frontal scalp focus, at least some of the initial learning context (i.e. source) is more likely to have been recovered as well. The later-onsetting, negative old/new effect appears to reflect activity coincident with or following the initial decision. This could be a continued, material-specific search (i.e. within the visual cortex that initially processed the perceptual attribute) for contextual information under the direction of the prefrontal cortex.

Notes

The authors thank Mr Charles L. Brown III for computer programming, Mr Jeff Cheng and Mr Martin Duff for technical assistance. We are grateful to all volunteers who generously gave their time to participate in the study. This research was supported in part by grant HD14959, and by the New York State Department of Mental Hygiene. David Friedman is supported, in part, by Senior Scientist Award no. MH01225 from NIMH.

Address correspondence to: Dr Yael M. Cycowicz, Cognitive Electrophysiology Laboratory — Unit 6, New York State Psychiatric Institute, 1051 Riverside Drive, New York, NY 10032, USA. Email: yc60@columbia.edu.

Table 1

Accuracy data summary during the item and source memory test blocks

Item memory (Old versus New) 
Accuracy: Old  New Miss FAa  Prb Brb 
Mean 0.85  0.93 0.15 0.07  0.77 0.33 
SD 0.07  0.05 0.07 0.05  0.08 0.19 
Item memory (Old versus New) 
Accuracy: Old  New Miss FAa  Prb Brb 
Mean 0.85  0.93 0.15 0.07  0.77 0.33 
SD 0.07  0.05 0.07 0.05  0.08 0.19 
Source memory (Target versus Nontarget) 
Accuracy: Target Nontarget New Miss New FA NT FA Prc Brc 
Miss: in Item, an old item called new; in source memory, a target called a nontarget. 
aFA, false alarm. In the item memory task, a FA is a new item called old; in the source memory task, a FA is a nontarget old item called a target (NT FA) or a new item called a target (New FA). 
bPr = discrimination measure; Br = bias index; both computed according to the methods provided by Snodgrass and Corwin (Snodgrass and Corwin, 1988). 
cComputed (Snodgrass and Corwin, 1988) using the overall false alarm rate to nontarget old and new items. 
Mean 0.76 0.77 0.97 0.24 0.03 0.23 0.61 0.40 
SD 0.09 0.09 0.03 0.09 0.03 0.09 0.11 0.11 
Source memory (Target versus Nontarget) 
Accuracy: Target Nontarget New Miss New FA NT FA Prc Brc 
Miss: in Item, an old item called new; in source memory, a target called a nontarget. 
aFA, false alarm. In the item memory task, a FA is a new item called old; in the source memory task, a FA is a nontarget old item called a target (NT FA) or a new item called a target (New FA). 
bPr = discrimination measure; Br = bias index; both computed according to the methods provided by Snodgrass and Corwin (Snodgrass and Corwin, 1988). 
cComputed (Snodgrass and Corwin, 1988) using the overall false alarm rate to nontarget old and new items. 
Mean 0.76 0.77 0.97 0.24 0.03 0.23 0.61 0.40 
SD 0.09 0.09 0.03 0.09 0.03 0.09 0.11 0.11 
Table 2

Reaction time in ms (± SD) during the two retrieval phases

Item memory 
 Old (hit)  New (CR) Miss FA 
Mean 750  782 857 928 
SD (85)  (93) (146) (170) 
Item memory 
 Old (hit)  New (CR) Miss FA 
Mean 750  782 857 928 
SD (85)  (93) (146) (170) 
Source memory 
 Old target NT Old New Target miss NT FA 
CR, correct rejection; FA, false alarm; NT, nontarget. 
Miss: in the item memory task, an old item called new; in the source memory task, a target called a nontarget. 
Mean 854 905 823 972 978 
SD (76) (77) (97) (157) (161) 
Source memory 
 Old target NT Old New Target miss NT FA 
CR, correct rejection; FA, false alarm; NT, nontarget. 
Miss: in the item memory task, an old item called new; in the source memory task, a target called a nontarget. 
Mean 854 905 823 972 978 
SD (76) (77) (97) (157) (161) 
Table 3

ANOVA results from the first averaged voltage analysis

  Measurement window: 
  Early1 Early2 Late1 Late2 
Effect: dfa F ε F ε F ε F ε 
Bold values are significant at P < 0.05 or less. 
aIdentical for all four measurement windows. 
Task (T)  1,15  0.1  11.0  18.7  8.7  
Old/New (O)  1,15  1.7  22.9  25.5  15.3  
Saggital (S)  2,30 18.6 0.96  0.1 0.71  1.7 0.94  0.7 0.93 
Anterior/Posterior (A)  7,105 4.5 0.21 7.2 0.20 16.9 0.26 21.2 0.31 
T × O  1,15  0.0   0.7  35.4  17.5  
T × S  2,30  0.8 0.99  3.2 0.76  1.9 0.75  0.7 0.79 
T × A  7,105  7.2 0.21 17.6 0.22 24.1 0.32 17.3 0.29 
O × S  2,30  2.9 0.92 5.6 0.71  3.8 0.67 7.6 0.62 
O × A  7,105 19.8 0.21 4.3 0.20 49.2 0.30 9.5 0.30 
S × A 14,210  2.1 0.30 3.4 0.32 6.9 0.30 7.7 0.31 
T × O × S  2,30  1.4 0.86  2.8 0.76 4.4 0.95 6.0 0.76 
T × O × A  7,105  0.2 0.27  0.03 0.21  2.3 0.23  2.2 0.27 
T × S × A 14,210  1.6 0.32 6.9 0.40 5.5 0.34  1.6 0.37 
O × S × A 14,210  2.0 0.29 3.2 0.21  1.8 0.16 3.2 0.18 
T × O × S × A 14,210  0.8 0.32  0.6 0.70 2.8 0.39  2.0 0.32 
  Measurement window: 
  Early1 Early2 Late1 Late2 
Effect: dfa F ε F ε F ε F ε 
Bold values are significant at P < 0.05 or less. 
aIdentical for all four measurement windows. 
Task (T)  1,15  0.1  11.0  18.7  8.7  
Old/New (O)  1,15  1.7  22.9  25.5  15.3  
Saggital (S)  2,30 18.6 0.96  0.1 0.71  1.7 0.94  0.7 0.93 
Anterior/Posterior (A)  7,105 4.5 0.21 7.2 0.20 16.9 0.26 21.2 0.31 
T × O  1,15  0.0   0.7  35.4  17.5  
T × S  2,30  0.8 0.99  3.2 0.76  1.9 0.75  0.7 0.79 
T × A  7,105  7.2 0.21 17.6 0.22 24.1 0.32 17.3 0.29 
O × S  2,30  2.9 0.92 5.6 0.71  3.8 0.67 7.6 0.62 
O × A  7,105 19.8 0.21 4.3 0.20 49.2 0.30 9.5 0.30 
S × A 14,210  2.1 0.30 3.4 0.32 6.9 0.30 7.7 0.31 
T × O × S  2,30  1.4 0.86  2.8 0.76 4.4 0.95 6.0 0.76 
T × O × A  7,105  0.2 0.27  0.03 0.21  2.3 0.23  2.2 0.27 
T × S × A 14,210  1.6 0.32 6.9 0.40 5.5 0.34  1.6 0.37 
O × S × A 14,210  2.0 0.29 3.2 0.21  1.8 0.16 3.2 0.18 
T × O × S × A 14,210  0.8 0.32  0.6 0.70 2.8 0.39  2.0 0.32 
Table 4

Comparison of scalp topographies based on normalized averaged voltages at 24 scalp sites

 Measurement window: 
 Early1 Early2 Late1 Late2 
 F df ε F df ε F df ε F df ε 
Bolded values are significant at P < 0.05 or less. 
Source memory only (1)             
Response Type × Electrode Location 5.4 23,345 0.13 3.7 23,345 0.15 0.7 23,345 0.13 0.2 23,345 0.12 
Item versus source memory (2)             
Task × Electrode Location 0.6 23,345 0.10 0.2 23,345 0.08 4.1 23,345 0.11 1.1 23,345 0.09 
 Measurement window: 
 Early1 Early2 Late1 Late2 
 F df ε F df ε F df ε F df ε 
Bolded values are significant at P < 0.05 or less. 
Source memory only (1)             
Response Type × Electrode Location 5.4 23,345 0.13 3.7 23,345 0.15 0.7 23,345 0.13 0.2 23,345 0.12 
Item versus source memory (2)             
Task × Electrode Location 0.6 23,345 0.10 0.2 23,345 0.08 4.1 23,345 0.11 1.1 23,345 0.09 
Figure 1.

Schematic diagram of the study and test phase trial sequences.

Figure 1.

Schematic diagram of the study and test phase trial sequences.

Figure 2.

Grand mean ERPs elicited by new and old items during the item task (left) and new and target and nontarget old items during the source memory task (right). The depicted data were recorded at 24 of the 62 scalp locations.

Figure 2.

Grand mean ERPs elicited by new and old items during the item task (left) and new and target and nontarget old items during the source memory task (right). The depicted data were recorded at 24 of the 62 scalp locations.

Figure 3.

Grand mean ERPs elicited by new (left) and old (right) items. In order to show task effects, the waveforms from the item and source memory tasks are superimposed. The depicted data were recorded at 24 of the 62 scalp locations.

Figure 3.

Grand mean ERPs elicited by new (left) and old (right) items. In order to show task effects, the waveforms from the item and source memory tasks are superimposed. The depicted data were recorded at 24 of the 62 scalp locations.

Figure 4.

Grand mean ERPs elicited during the item (left-most column) and source (middle and right columns) memory tasks. The ERPs are depicted at three posterior midline sites on larger amplitude scales and broader time bases than in Figures 2 and 3. Arrows mark stimulus onset with timelines every 400 ms.

Grand mean ERPs elicited during the item (left-most column) and source (middle and right columns) memory tasks. The ERPs are depicted at three posterior midline sites on larger amplitude scales and broader time bases than in Figures 2 and 3. Arrows mark stimulus onset with timelines every 400 ms.

Figure 5.

Grand mean ERPs recorded during the source memory task. The first four columns depict ERP comparisons in which responses were made with the same hand to both classes of ERPs within each column. The last column depicts the ERPs in which the two classes were associated with different hands. Arrows mark stimulus onset with time lines every 400 ms.

Figure 5.

Grand mean ERPs recorded during the source memory task. The first four columns depict ERP comparisons in which responses were made with the same hand to both classes of ERPs within each column. The last column depicts the ERPs in which the two classes were associated with different hands. Arrows mark stimulus onset with time lines every 400 ms.

Figure 6.

Grand mean old minus new (item memory task), target minus new and nontarget minus new (source memory task) difference waveforms associated with, respectively, correct old/new, target correct and nontarget correct source judgments. The depicted data were recorded at 24 of the 62 scalp locations.

Figure 6.

Grand mean old minus new (item memory task), target minus new and nontarget minus new (source memory task) difference waveforms associated with, respectively, correct old/new, target correct and nontarget correct source judgments. The depicted data were recorded at 24 of the 62 scalp locations.

Figure 7.

Surface potential (SP) maps for each of the four indices. Top row: correct item memory trials. Middle row: source memory trials associated with correct color judgments. Bottom row: source memory trials associated with incorrect color judgments. The source memory maps are based on the ERPs averaged across target and nontarget trials. The SP isopotential lines are separated by 0.30 μV.

Figure 7.

Surface potential (SP) maps for each of the four indices. Top row: correct item memory trials. Middle row: source memory trials associated with correct color judgments. Bottom row: source memory trials associated with incorrect color judgments. The source memory maps are based on the ERPs averaged across target and nontarget trials. The SP isopotential lines are separated by 0.30 μV.

References

Bentin S, Moscovitch M, Heth I (
1992
) Memory with and without awareness: performance and electrophysiological evidence of savings.
J Exp Psychol Learn Mem Cogn
 
18
:
1270
–1283.
Berman S, Friedman D, Hamberger M, Snodgrass JG (
1989
) Name agreement, familiarity and visual complexity norms for 320 line drawings in children and adults.
Behav Res Methods Instrum
 
21
:
371
–382.
Buckner RL, Tulving E (1995) Neuroimaging studies of memory: theory and recent PET results. In: Handbook of neuropsychology, Vol. 10 (Boller F, Grafman J, eds), pp. 439–466. Amsterdam: Elsevier.
Cermak LS, Verfaellie M, Sweeney M (
1992
) Fluency versus conscious recollection in the word completion performance of amnesic patients.
Brain Cogn
 
20
:
367
–377.
Cowan N, Stadler MA (
1996
) Estimating unconscious processes: implications of a general class of models.
J Exp Psychol Gen
 
125
:
195
–200.
Cycowicz YM, Friedman D, Rothstein M, Snodgrass JG (
1997
) Picture naming norms in young children: Norms for name agreement, familiarity and visual complexity.
J Exp Child Psychol
 
65
:
171
–237.
Fletcher PC, Shallice T, Frith CD, Frackowiak RS, Dolan RJ (
1998
) The functional roles of prefrontal cortex in episodic memory. II. Retrieval.
Brain
 
121
:
1249
–1256.
Farah MJ, Peronnet F, Gonon MA, Giard MH (
1988
) Electrophysiological evidence for a shared representational medium for visual images and visual percepts.
J Exp Psychol Gen
 
117
:
248
–257.
Friedman D, Johnson R Jr (
2000
) Event-related potential (ERP) studies of memory encoding and retrieval: a selective review.
Microsc Res Techn
 
51
:
6
–28.
Henson RNA, Rugg MD, Shallice T, Josephs O, Dolan RJ (
1999
) Recollection and familiarity in recognition memory: an event-related functional magnetic resonance imaging study.
J Neurosci
 
19
:
3962
–3972.
Jacoby LL (
1991
) A process dissociation framework: separating automatic from intentional uses of memory
J Mem Lang
 
30
:
513
–541.
Jacoby LL, Dallas M (
1981
) On the relationship between autobiographical memory and perceptual learning.
J Exp Psychol Gen
 
110
:
306
–340.
Jacoby LL, Toth JP, Yonelinas AP (
1993
) Separating conscious and unconscious influences of memory: measuring recollection.
J Exp Psychol Gen
 
122
:
139
–154.
Janowsky JS, Shimamura AP, Squire LR (
1989
) Source memory impairment in patients with frontal lobe lesions.
Neuropsychologia
 
27
:
1043
–1056.
Jennings JR, Wood CC (
1976
) The ε-adjustment procedure for repeated- measure analyses of variance.
Psychophysiology
 
13
:
277
–278.
Johnson, R Jr (1995) Event-related potential insights into the neurobiology of memory systems. In: Handbook of neuropsychology, Volume 10 (Boller F, Grafman J, eds), pp. 135–164. Amsterdam: Elsevier.
Jones GV (
1987
) Independence and exclusivity among psychological processes: Implications for the structure of recall.
Psychol Rev
 
94
:
229
–235.
Joordens S, Merikle PM (
1993
) Independence or redundancy? Two models of conscious and unconscious influences.
J Exp Psychol Gen
 
122
:
462
–467.
Knowlton BJ, Squire LR (
1995
) Remembering and knowing: two different expressions of declarative memory.
J Exp Psychol Learn Mem Cogn
 
21
:
699
–710.
Mandler G (
1980
) Recognizing: the judgement of previous occurrence.
Psychol Rev
 
87
:
252
–271.
McCarthy G, Wood CC (
1985
) Scalp distributions of event-related potentials: an ambiguity associated with analysis of variance models.
Electroenceph clin Neurophysiol
 
62
:
203
–208.
Mulligan, NW, Hirshman E. (
1997
). Measuring the bases of recognition memory: an investigation of the process-dissociation framework.
J Exp Psychol Learn Mem Cogn
 
23
:
280
–304.
Nelson DL (1979) Remembering pictures and words: appearance, significance, and name. In: Levels of processing in human memory (Cermark LS, Craik FIM, eds), pp. 45–76. Hillsdale, NJ: Erlbaum.
Nolde SF, Johnson MK, Raye CL (
1998
) The role of prefrontal cortex during tests of episodic memory.
Trend Cogn Sci
 
2
:
399
–406.
Nuwer MR, Comi G, Emerson R, Fugslang-Frederiksen A, Guerit JM, Hinrichs H, Ikeda A, Lucas FJC, Rappelsburger P (
1998
) IFCN standards for digital recording of clinical EEG.
Electroenceph clin Neurophysiol
 
106
:
259
–261.
Paivio A (1986) Mental representations: a dual coding approach. New York: Oxford University Press.
Paller KA, Bozic VS, Ranganath C, Grabowecky M, Yamada S (
1999
). Brain waves following remembered faces index conscious recollection.
Cogn Brain Res
 
7
:
519
–531.
Perrin P, Pernier J, Bertrand O, Echallier JF (
1989
) Spherical splines for scalp potential and current density mapping.
Electroenceph clin Neurophysiol
 
72
:
184
–187.
Ritter W, Simson R, Vaughan HG. (
1972
) Association cortex potentials and reaction time in auditory discrimination.
Electroenceph clin Neurophysiol
 
33
:
547
–555.
Ruchkin D, Johnson R, Friedman D (
1999
) Scaling is necessary when making comparisons between shapes of event-related potential topographies: a reply to Haig et al.
Psychophysiology
 
36
:
832
–834.
Rugg MD (1995) ERP studies of memory. In: Electrophysiology of mind: event-related brain potentials and cognition, Volume XV (Rugg MD, Coles MGH, eds), pp. 132–170 New York: Oxford University Press.
Rugg MD, Nagy ME (
1989
) Event-related potentials and recognition memory for words.
Electroenceph Clin Neurophysiol
 
72
:
395
–406.
Rugg MD, Fletcher PC, Frith CD, Frackowiak RSJ, Dolan RJ (
1996
) Differential activation of the prefrontal cortex in successful and unsuccessful memory retrieval.
Brain
 
119
:
2073
–2083.
Rugg MD, Mark RE, Walla P, Schloerscheidt AM, Birch CS, Allan K (
1998
). Dissociation of the neural correlates of implicit and explicit memory.
Nature
 
392
:
595
–598.
Schacter DL (
1987
). Implicit memory: history and current status.
J Exp Psychol Learn Mem Cogn
 
13
:
501
–518.
Schacter DL, Buckner RL (
1998
) On the relations among priming, conscious recollection, and intentional retrieval: evidence from neuroimaging research.
Neurobiol Learn Mem
 
70
:
284
–303.
Schacter DL, Reiman E, Curran T, Yun LS, Bandy D, McDermott KB, Roediger HLR (
1996
) Neuroanatomical correlates of veridical and illusory recognition memory: evidence from positron emission tomography.
Neuron
 
17
:
267
–274.
Scherg M (1990) Fundamentals of dipole source analysis. In: Auditory evoked magnetic fields and potentials, Volume 6 (Hoke M et al., eds), pp. 40–69. Basel: Karger.
Schloerscheidt AM, Rugg MD (
1997
) Recognition memory for words and pictures: an event-related potential study.
NeuroReport
 ,
8
:
3281
–3285.
Senkfor AJ, Van Petten C (
1998
) Who said what: an event-related potential investigation of source and item memory.
J Exp Psychol Learn Mem Cogn
 
24
:
1005
–1025.
Smith ME (
1993
) Neurophysiological manifestations of recollective experience during recognition memory judgments.
J Cogn Neurosci
 
5
:
1
–13.
Snodgrass JG, Corwin J (
1988
) Pragmatics of measuring recognition memory: applications to dementia and amnesia,
J Exp Psychol Gen
 
116
:
34
–50.
Snodgrass JG, Vanderwart M (
1980
) A standardized set of 260 pictures: norms for name agreement, image agreement, familiarity, and visual complexity.
J Exp Psychol Hum Learn Mem
 
6
:
174
–215.
Squire LR (1994) Declarative and nondeclarative memory: multiple brain systems supporting learning and memory. In: Memory Systems 1994 (Schacter DL, Tulving E, eds), pp. 203–231. Cambridge: MIT Press.
Squire LR, Kandel ER (1999) Memory: from mind to molecules, Vol. 69. New York: W.H. Freeman.
Squire LR, Knowlton B (2000) The medial temporal lobe, the hippocampus, and the memory systems of the brain. In: The New Cognitive Neurosciences (Gazzaniga MS, ed.). Cambridge, MA: MIT Press.
Trott C, Friedman D, Ritter W, Fabiani M (
1997
) Item and source memory: differential age effects revealed by event-related potentials.
NeuroReport
 
8
:
3373
–3378.
Trott C, Friedman D, Ritter W, Fabiani M (
1999
) Episodic priming and memory for temporal source: event-related potentials reveal age-related differences in prefrontal functioning.
Psychol Aging
 
14
:
390
–413.
Tulving E (
1985
) Memory and consciousness.
Can Psychol
 
26
:
1
–12.
Wagner AD, Desmond JE, Glover GH, Gabrieli JD (
1998
) Prefrontal cortex and recognition memory. Functional-MRI evidence for context- dependent retrieval processes.
Brain
 
121
:
1985
–2002.
Wilding EL (
1999
) Separating retrieval strategies from retrieval success: an event-related potential study of source memory.
Neuropsychologia
 
37
:
441
–454.
Wilding EL, Rugg MD (
1996
) An event-related potential study of recognition memory with and without retrieval of source.
Brain
 
119
:
889
–905.
Wilding EL, Rugg MD (
1997
) Event-related potentials and the recognition memory exclusion task.
Neuropsychologia
 
35
:
119
–128.
Yonelinas AP, Regehr G, Jacoby LL (
1995
). Incorporating response bias in dual-process theories of memory.
J Mem Lang
 
34
:
821
–835.