Abstract

Distractibility can lead to accidents and academic failures as well as memory problems. Recent evidence suggests that intentional recognition memory can be biased by unintentional recognition of distracting stimuli in the same environment. It is unknown whether unintentional and intentional recognition depend on the same underlying neurocognitive mechanisms. We assessed whether human participants' recognition of previously seen (old) or not seen (new) target stimuli was affected by whether a to-be-ignored distractor was old or new. ERPs were recorded to investigate the neural correlates of this bias. The results showed that the old/new status of salient distractors had a biasing effect on target recognition accuracy. Both intentional and unintentional recognition elicited early ERP effects that are thought to reflect relatively automatic memory processes. However, only intentional recognition elicited the later ERP marker of conscious recollection, consistent with previous suggestions that recollection is under voluntary control. In contrast, unintentional recognition was associated with an enhanced late posterior negativity, which may reflect monitoring or evaluation of memory signals. The findings suggest that unintentional and intentional recognition involve dissociable memory processes.

INTRODUCTION

The ability to ignore irrelevant distracting stimuli that interfere with our current goals is a critical skill for achieving many everyday tasks, such as driving or studying. Although progress has been made in understanding how distraction can impair perception and general decision-making (e.g., Gazzaley & Nobre, 2012), less is known regarding how distraction affects episodic recognition. In most laboratory-based memory experiments, researchers do their best to minimize distraction by presenting only the stimuli that participants are supposed to evaluate. However, in real life, we often need to recognize one stimulus in the context of multiple irrelevant stimuli in the same environment. Thus, distraction effects on recognition may be prevalent outside the laboratory.

Previous research has shown that participants' recognition of previously seen (old) or not seen (new) target stimuli can be biased by whether a simultaneously presented to-be-ignored distractor is old or new (Anderson, Jacoby, Thomas, & Balota, 2011; Ste-Marie & Jacoby, 1993). That is, people are more likely to report that a target item is old if the distractor item is also old and are more likely to report that a target item is new if the distractor item is also new, despite being explicitly instructed to always ignore the distractors. Similar results have been found in the literature on context effects on recognition memory, where previously encountered contexts can bias recognition responses to items superimposed on these contexts (e.g., Hockley, Bancroft, & Bryant, 2012; Murnane, Phelps, & Malmberg, 1999). Distractor-induced recognition biases are enhanced in young people when an additional task is conducted simultaneously that taxes their cognitive control abilities (Anderson et al., 2011; Ste-Marie & Jacoby, 1993), whereas older people show large recognition biases even without a secondary task (Anderson et al., 2011). In both young and old people, recognition biases are more likely when the distractors are pictures and targets words, rather than vice versa (Anderson et al., 2011). These findings suggest that a failure of control mechanisms to confine processing to targets can trigger unintentional recognition of particularly salient distractors and that the memory signal from distractors is then misattributed to targets.

A large body of research has suggested that intentional recognition is supported by multiple distinct retrieval processes, including rapid and relatively automatic assessments of item familiarity as well as slower, more controlled recollection of contextual details from a specific previous encounter with an item (Yonelinas & Jacoby, 2012; Mandler, 1980). This account predicts that unintentional recognition of distractors is more likely driven by familiarity than recollection, because familiarity is more automatic than recollection (Anderson et al., 2011). Thus far, however, there is no direct evidence on this issue, perhaps because the memory processes associated with unintentional distractor recognition can only be indirectly observed with behavioral measures in terms of their biasing effect on target memory judgments.

Therefore, we used EEG to directly measure recognition-related brain responses elicited by both targets and distractors to reveal the underlying neurocognitive processes associated with intentional and unintentional recognition. Young healthy adults were given a recognition test, where, in each trial, an old or new picture was simultaneously presented with an old or new word. In the first experiment, participants made recognition judgments on words while ignoring pictures, and they made judgments on pictures while ignoring words in the second experiment. We then measured how well-known ERP markers of familiarity and recollection (Wilding & Ranganath, 2012; Rugg & Curran, 2007; Rugg et al., 1998) were modulated by distractor and target old/new status.

We expected both intentional and unintentional recognition to elicit an early frontal ERP positivity—the FN400—which is thought to reflect a relatively automatic familiarity process (Curran, 2000; Rugg et al., 1998; although see Paller, Voss, & Boehm, 2007). In contrast, intentional target recognition was predicted to be uniquely associated with a later parietal ERP positivity that indexes conscious recollection and that can be voluntarily controlled (Hu, Bergström, Bodenhausen, & Rosenfeld, 2015; Bergström, Anderson, Buda, Simons, & Richardson-Klavehn, 2013; Bergström, De Fockert, & Richardson-Klavehn, 2009a, 2009b; Hanslmayr, Leipold, Pastötter, & Bäuml, 2009; Mecklinger, Parra, & Waldhauser, 2009; Bergström, Velmans, De Fockert, & Richardson-Klavehn, 2007). Unintentional recognition was instead expected to engage additional postretrieval monitoring processes (Rugg & Wilding, 2000) that may be recruited to evaluate the automatic memory signals elicited by old distractors to counteract their biasing influence. Such postretrieval monitoring was expected to be manifested as late ERP slow drifts (e.g., Johansson & Mecklinger, 2003). Finally, we expected behavioral and ERP correlates of unintentional recognition to be primarily expressed when distractors were visually salient pictures rather than less salient words (Anderson et al., 2011).

METHODS

Participants

Twenty-four right-handed, native English speakers participated in each experiment (Experiment 1: M age = 21 years, range = 18–24 years, eight men and 16 women; Experiment 2: M age = 19 years, range = 18–22 years, five men and 19 women). Participants were recruited through the offer of course credit or were awarded money for their participation. All participants gave written informed consent, and the experiment was approved by the University of Kent Psychology Research Ethics Committee.

Materials

Stimuli were 336 words taken from the ANEW database (Bradley & Lang, 1999) with valence ratings ranging from 3.79 to 7.58 on a 9-point scale and 336 color photographs of a range of objects, events, and scenes. Of the photographs, 277 were from the IAPS database (Lang, Bradley, & Cuthbert, 2008) with valence ratings between 1.51 and 6.62 on a 9-point scale, and 43 were from the GAPED database (Dan-Glauser & Scherer, 2011) with valence ratings between 1.35 and 45.7 on a 100-point scale. We initially aimed to investigate the effect of emotional valence of distracting pictures on word target recognition and ERPs, so half of the pictures were emotionally negative and the other half were neutral. However, as there were no effects of valence on behavior or ERPs in either experiment, all results are presented collapsed across this factor. The words ranged from four to eight letters in length, each of no more than two syllables. Of each type, 16 were assigned to a practice round, and the remaining 320 were used in the experiment. Assignment of words and pictures to experimental conditions was fully counterbalanced across participants.

Design and Procedure

Our experimental design was closely based on the “Memory Stroop” paradigm devised by Anderson et al. (2011), which was designed to investigate the effect of unintentional distractor recognition on intentional target recognition performance. The name of this paradigm stems from its similarities with the traditional color-word Stroop (1935) task, which investigates the effect of unintentional word reading on people's ability to name the ink in which the word is printed (see Anderson et al., 2011, for more details). Participants were first given task instructions and completed a short practice phase. The actual experiment consisted of 10 study-test cycles (split into multiple cycles to ensure adequate recognition test performance, in line with Anderson et al., 2011). In each study phase, 16 pictures and 16 words were presented individually and randomly interspersed at the center of the screen for 3000 msec, preceded by a 500-msec fixation cross. Participants rated the pleasantness of the words and pictures on a scale between 1 and 4 (with 1 being very unpleasant and 4 being very pleasant) by pressing buttons on a keyboard and were told that their memory for all items would later be tested.

In each target recognition test phase, 32 pairs of pictures and words were presented with the word superimposed over the picture. Each phase contained four combinations: old word and old picture (eight trials), old word and new picture (eight trials), new word and old picture (eight trials), and new word and new picture (eight trials), displayed in random order. Participants were asked to press one keyboard button if they recognized the target stimulus (words in Experiment 1 and pictures in Experiment 2) as “old” (i.e., previously presented in the experiment) and another button to classify the target as “new” (i.e., not previously presented in the experiment), with response hand counterbalanced across participants. Participants were instructed to always ignore distractor stimuli (pictures in Experiment 1 and words in Experiment 2). Each trial began with a 500-msec fixation cross, after which the word-and-picture pair was presented for 3000 msec, and participants were asked to respond while the pair was still on the screen.

During each test phase, participants also did a simultaneous working memory (WM) task, because previous research had shown that unintentional distractor recognition is more likely during dual-task conditions. Anderson et al. (2011) and Ste-Marie and Jacoby (1993) found that young participants showed larger distractor-induced recognition biases when they were given a secondary task that involved listening to a recorded list of digits and verbally responding whenever they detected particular sequences of digits. However, because a continuous listening task with verbal responses would interfere with the EEG recording, we used an alternative secondary task that involved covert rehearsal of digit sequences, which have been shown to increase interference from distractor processing in other tasks (e.g., De Fockert, Rees, Frith, & Lavie, 2001). To this end, a number string of five digits (0–4, always beginning with 0 but with 1–4 in random order) was shown for 3000 msec every four to six trials (randomly determined), and participants were instructed to maintain the sequence of numbers in WM while completing the episodic recognition task. After four to six trials, a single digit probe was displayed for 3000 msec, and participants pressed the number corresponding to the next digit in the number sequence that they were currently rehearsing. To encourage participants to pay attention to the WM task, they were given visual feedback regarding the accuracy of each response (either “incorrect,” “correct,” or “no response”). Next, participants were shown a new number sequence to maintain in WM during the following four to six recognition trials until the next probe. No number sequence was repeated within the same test cycle.

After the target recognition test in each cycle, participants were given a very short distractor recognition test consisting of two previously seen distractors intermixed with two novel items from the same stimuli class (pictures in Experiment 1 and words in Experiment 2) and were asked to press one button to classify distractors as “old” (previously seen) and another to classify them as “new” (not seen at any point in the experiment). The purpose of this test was to ensure that participants attempted to memorize distractors as well as targets during the study phases because both item types would be tested. Stimulus presentation durations and response buttons during the distractor test were the same as in the target recognition test.

EEG Recording and Analysis

EEG was recorded at 500 Hz with a 0.05- to 70-Hz bandwidth using FCz as the reference electrode for 64 scalp electrodes placed in an actiCAP (Brain Products GmbH, München, Germany), with locations according to the extended 10–20 system. EOG was recorded from below the left eye (vertical EOG) and from the right outer canthi (horizontal EOG). Recorded data were analyzed using EEGLAB (UC San Diego; Delorme & Makeig, 2004). The EEG was rereferenced to the average of the mastoids and segmented into 1700-msec epochs (including a 200-msec prestimulus baseline) that were time-locked to the onset of the word–picture pair in the target recognition test. Epochs were concatenated and submitted to extended infomax independent component analysis using Runica from the EEGLAB toolbox, with default extended-mode training parameters (Delorme & Makeig, 2004). Independent components reflecting eye movements and other sources of noise were identified by visual inspection of component scalp topographies, time courses, and activation spectra and were discarded from the data by back-projecting all but these components to the data space. Corrected data were subsequently low-pass filtered digitally at 30 Hz (two-way least-squares finite impulse response filter). Any trials that still contained visible artifacts after filtering were removed, as were trials where participants failed to respond within the allocated time. Only a very small percentage of trials (5% in Experiment 1 and 3% in Experiment 2) were deleted in total. Finally, ERPs were formed for the four conditions: old word old picture (mean trial numbers in Experiment 1 = 76.4 and Experiment 2 = 77.7), old word new picture (mean trial numbers in Experiment 1 = 75.8 and Experiment 2 = 77.4), new word old picture (mean trial numbers in Experiment 1 = 76.3 and Experiment 2 = 77.8), and new word new picture (mean trial numbers in Experiment 1 = 76.4 and Experiment 2 = 77.4).

We first tested our specific predictions by statistically analyzing ERP mean amplitudes from two time windows and electrode sites where the FN400 and left parietal old/new effects are typically maximal, 300–500 msec at the midfrontal site (Fz) and 500–800 msec at the left parietal site (P3), respectively. These a priori selected time windows and locations were based on a large body of previous research (reviewed in Rugg & Curran, 2007). The targeted analysis did not include the late ERP slow drifts that are thought to index retrieval monitoring processes. This is because retrieval monitoring-related slow drifts can have very different scalp distributions across studies (cf. Hayama, Johnson, & Rugg, 2008; Johansson & Mecklinger, 2003) so we were unable to make clear predictions about their spatial locations.

Because selecting only a few time windows and electrode sites for analysis may overlook effects at other sites and time points, we also conducted a whole-head, fully data-driven multivariate “nonrotated” task partial least squares (task-PLS) analysis (McIntosh & Lobaugh, 2004). PLS allows the examination of distributed patterns of spatial and temporal dependencies in the ERP data with minimal assumptions regarding the timing and distribution of potential effects. Task-PLS analyzes the “cross-block” covariance between the spatiotemporal ERP distribution and orthogonal contrast vectors representing differences between experimental conditions. In nonrotated PLS (Bergström, Anderson, et al., 2013; Bergström et al., 2009a, 2009b; McIntosh & Lobaugh, 2004), the sums of squares of the cross-block covariance between the contrast vector and the spatiotemporal data matrix are directly tested for significance using random permutation test, thus allowing a direct assessment of the hypothesized experimental effects. Correction for multiple comparisons is not required, because the PLS only tests the same number of contrasts as degrees of freedom in the design. The PLS analysis outputs electrode saliences that identify the electrodes that most strongly covary at a particular point in time with the experimental effect expressed in the contrast vector. The standard errors of the electrode saliences are estimated through bootstrap resampling. The ratio of the electrode salience to the bootstrapped standard error gives a standardized measure of reliability that is approximately equivalent to a z score, whereby values above 1.96 and below −1.96 are reliably different from zero with a 95% confidence interval (McIntosh & Lobaugh, 2004).

In the current analysis, nonrotated task-PLS was used to test the full factorial design with contrasts coding for the main effects of word and picture memory status (old vs. new) as well as their interaction term. Data from all scalp channels across the time window from 0 to 1000 msec poststimulus were included. The covariance of the experimental contrasts with the spatiotemporal data was tested for significance using 1000 permutations, and the reliability of the electrode saliences was tested using 200 bootstraps (see McIntosh and Lobaugh, 2004, for a full description of PLS). MATLAB code to perform PLS is available at www.rotman-baycrest.on.ca/pls/.

RESULTS

Behavior

WM task accuracy was similar and high in both experiments (Experiment 1: M = 0.84, SD = 0.10; Experiment 2: M = 0.83, SD = 0.13; t < 1, p > .79), suggesting that participants complied with instructions and successfully managed to combine performing both tasks. One participant scored lower than 2 SDs below the mean in each experiment (accuracy of 0.57 in Experiment 1 and 0.41 in Experiment 2), but excluding those participants who did not change the pattern of results on the target recognition task; therefore, all participants were included in the final analysis.

For the recognition task, we first analyzed raw hit rates and correct rejection rates to make our results directly comparable with previous research (Anderson et al., 2011; Ste-Marie & Jacoby, 1993). Mean accuracy and RT are presented in Table 1.

Table 1. 

Mean Accuracy and RT of Target Recognition Decisions in Both Experiments

Experiment 1, Word Targets and Pictures DistractorsExperiment 2, Picture Targets and Words Distractors
Mean (SD) AccuracyMean (SD) RTMean (SD) AccuracyMean (SD) RT
Old Word Old Picture 0.93 (0.06) 1206 (200) 0.95 (0.03) 1145 (161) 
Old Word New Picture 0.90 (0.07) 1164 (187) 0.97 (0.03) 1234 (187) 
New Word Old Picture 0.90 (0.11) 1280 (201) 0.95 (0.04) 1134 (197) 
New Word New Picture 0.92 (0.06) 1248 (219) 0.97 (0.03) 1186 (211) 
Experiment 1, Word Targets and Pictures DistractorsExperiment 2, Picture Targets and Words Distractors
Mean (SD) AccuracyMean (SD) RTMean (SD) AccuracyMean (SD) RT
Old Word Old Picture 0.93 (0.06) 1206 (200) 0.95 (0.03) 1145 (161) 
Old Word New Picture 0.90 (0.07) 1164 (187) 0.97 (0.03) 1234 (187) 
New Word Old Picture 0.90 (0.11) 1280 (201) 0.95 (0.04) 1134 (197) 
New Word New Picture 0.92 (0.06) 1248 (219) 0.97 (0.03) 1186 (211) 

RTs measured in millisecond.

In Experiment 1, a 2 × 2 ANOVA on the accuracy data revealed a significant interaction between Word memory status and Picture memory status (F(1, 23) = 7.29, p = .01, ηp2 = .24), but no main effects (Fs < 1, p > .75). When the distracting picture was new, it decreased the likelihood that old target words would be correctly recognized compared with when the distracting picture was old (t(23) = 2.84, p = .009, Cohen's d = 0.46, calculated here and subsequently as the difference between means divided by the pooled standard deviation to ensure unbiased effect size estimates; Dunlap, Cortina, Vaslow, & Burke, 1996), whereas new distracting pictures facilitated correct rejections of new words compared with old distracting pictures, although this difference was only at trend-level significance (t(23) = 1.83, p = .08, d = 0.23).

For RTs in Experiment 1, the 2 × 2 ANOVA revealed only significant main effects and no interaction (F < 1, p = .70). RTs were slower for new (M = 1264, SEM = 42) than old (M = 1185, SEM = 39) target words (F(1, 23) = 25.89, p < .001, ηp2 = .53) and slower for old (M = 1243, SEM = 40) than new (M = 1206, SEM = 41) distracting pictures (F(1, 23) = 12.58, p = .002, ηp2 = .35).

In Experiment 2, the accuracy pattern was different. A 2 × 2 ANOVA on the accuracy data revealed only a significant main effect of Picture memory status (F(1, 23) = 11.60, p = .002, ηp2 = .34) with significantly higher accuracy for new pictures (M = 0.97, SEM = 0.01) than old pictures (M = 0.95, SEM = 0.01). The main effect of Word memory status and the interaction were far from significant (Fs < 1, p > .51).

RT differences in Experiment 2 were reversed compared with Experiment 1. The 2 × 2 ANOVA revealed only significant main effects and no significant interaction (F(1, 23) = 2.24, p = .15). Now, RTs were slower for new (M = 1210, SEM = 39) than old (M = 1139, SEM = 36) target pictures (F(1, 23) = 12.52, p = .002, ηp2 = .35) and slower for old (M = 1190, SEM = 34) than new (M = 1160, SEM = 39) distracting words (F(1, 23) = 4.87, p = .04, ηp2 = .18).

To formally assess whether accuracy and RT patterns were qualitatively different across the two experiments, we analyzed both measures with three-way mixed ANOVAs with the factors Experiment, Word memory status, and Picture memory status. For accuracy, there was indeed a significant three-way interaction (F(1, 46) = 4.84, p = .033, ηp2 = .10), confirming that the Word × Picture memory status interaction was unique to Experiment 1. For RTs, the three-way interaction was not significant (F(1, 46) = 1.85, p = .18, ηp2 = .04), consistent with the lack of Word × Picture memory status interactions in both experiments. However, Experiment interacted with Word memory status (F(1, 46) = 27.89, p < .001, ηp2 = .38) and Picture memory status (F(1, 46) = 22.87, p < .001, ηp2 = .33) factors individually. These two-way interactions arose because RTs were slower for new than old targets (i.e., new words > old words in Experiment 1 and new pictures > old pictures in Experiment 2) and slower for old than new distractors (i.e., new pictures < old pictures in Experiment 1 and new words < old words in Experiment 2) in both experiments, leading to a crossover pattern when words and pictures swapped target/distractor assignment.

In a second analysis, we also calculated independent measures of discrimination and response bias (see Snodgrass & Corwin, 1988), to assess whether unintentional recognition of distractors primarily affected participant response biases rather than their ability to discriminate between old versus new targets. The Pr discrimination measure is calculated by subtracting each individual's false alarm rate from their hit rate on a recognition task and thus provides a measure of discrimination between old and new items that is corrected for response biases. For Experiment 1, therefore, new word false alarm rates were subtracted from old word hit rates separately for when distractor pictures were old versus new. For Experiment 2, new picture false alarm rates were subtracted from old picture hit rates separately for when distractor words were old versus new. The Br response bias measure is calculated by dividing each individual's false alarm rate by 1 − Pr. Values of Br that are above 0.5 indicate a tendency to guess “old” rather than “new” when uncertain (a positive response bias), whereas values below 0.5 indicate the opposite tendency. For Experiment 1, therefore, new word false alarm rates were divided by the Pr measure calculated in the previous step, separately for when distractor pictures were old versus new. For Experiment 2, new picture false alarm rates were divided by the Pr measure calculated in the previous step, separately for when distractor words were old versus new. These measures are presented in Table 2.

Table 2. 

Mean Discrimination Performance (Pr) and Response Bias (Br) for Target Recognition Decisions in Both Experiments

Experiment 1, Word Targets and Pictures DistractorsExperiment 2, Picture Targets and Words Distractors
Mean (SD) PrMean (SD) BrMean (SD) PrMean (SD) Br
Old distractors 0.83 (0.13) 0.57 (0.23) 0.93 (0.06) 0.35 (0.18) 
New distractors 0.83 (0.11) 0.44 (0.20) 0.93 (0.06) 0.32 (0.19) 
Experiment 1, Word Targets and Pictures DistractorsExperiment 2, Picture Targets and Words Distractors
Mean (SD) PrMean (SD) BrMean (SD) PrMean (SD) Br
Old distractors 0.83 (0.13) 0.57 (0.23) 0.93 (0.06) 0.35 (0.18) 
New distractors 0.83 (0.11) 0.44 (0.20) 0.93 (0.06) 0.32 (0.19) 

Two-way ANOVAs with the factors Experiment (1 vs. 2) × Distractor memory status (old vs. new) on Pr and Br revealed a significant main effect of Experiment on discrimination (F(1, 46) = 14.69, p < .001, ηp2 = .24), with significantly higher discrimination in Experiment 2 (M = 0.93, SEM = 0.02) than Experiment 1 (M = 0.83, SEM = 0.02). There was no effect of Distractor memory status and no interaction (both Fs < 0.15). For response bias, there was also a significant main effect of Experiment (F(1, 46) = 12.28, p = .001, ηp2 = .21) because participants showed a significantly more positive response bias in Experiment 1 (M = 0.50, SEM = 0.03) than in Experiment 2 (M = 0.34, SEM = 0.03). Participants were also significantly more likely to respond “old” when the distractors were old (M = 0.46, SEM = 0.03) than new (M = 0.38, SEM = 0.03; F(1, 46) = 6.59, p = .014, ηp2 = .13). However, the Experiment × Distractor memory status interaction was not significant (F(1, 46) = 1.85, p = .18, ηp2 = .04).

In both experiments, accuracy on the WM task was negatively (but nonsignificantly) correlated with the size of the congruency accuracy effect (congruent minus incongruent conditions) on the target recognition task (Experiment 1: rs = −.25, p = .24; Experiment 2: rs = −.24, p = .26). This finding shows that recognition biases were not simply related to how much participants complied with instructions to divide attention over both tasks, because this account would predict that recognition biases should increase as accuracy on the WM task increased (i.e., a positive correlation). To the contrary, participants who showed larger distraction-induced recognition biases also performed more poorly on the WM task.

In summary, target recognition accuracy and RTs showed qualitatively different patterns in both experiments, albeit in different ways. In Experiment 1, where words were targets and pictures were distractors, accuracy was highest on congruent (where targets and distractors had the same memory status, i.e., were both either old or new) compared with incongruent (where targets and distractors had opposite memory status) trials. RTs, however, were longer for old distractors than new distractors and longer for new targets than old targets. This pattern suggests that distractor effects on performance in Experiment 1 cannot be accounted for by a simple speed-accuracy trade-off.

In Experiment 2, where pictures were targets and words were distractors, there was no congruency effect and no main effect of distractors on accuracy but only an effect of target memory status whereby new targets were more accurately classified as new than old targets were correctly classified as old. RTs were, however, similar to Experiment 1, with longer RTs for old distractors than new distractors and longer RTs for new targets than old targets. This pattern suggests again a lack of a simple speed-accuracy trade-off in Experiment 2.

The analyses of discrimination and response bias measures showed that, across both experiments, unintentional recognition of distractors significantly influenced only response bias but not discrimination, consistent with the view that distractors were biasing responses toward the memory status of the distractor rather than influencing participants' ability to discriminate between old and new targets.

ERPs

Grand-averaged ERPs from the midfrontal (Fz) and left parietal (P3) electrode sites from both experiments are displayed in Figure 1.

Figure 1. 

Grand-averaged ERPs and scalp topographies of old/new effects for targets and distractors in both experiments. (A) ERPs from midfrontal (Fz, top row) and left parietal (P3, bottom row) sites in Experiments 1 (left) and 2 (right). (B) Scalp topographies of the mean amplitude old minus new difference for words irrespective of picture memory status (top row) and the old minus new difference for pictures irrespective of word memory status (bottom row) in Experiments 1 (left) and 2 (right). Whereas old targets elicited both typical early (300–500 msec) and late (500–800 msec) positive ERP amplitudes in both experiments, old picture distractors (Experiment 1) only elicited an early ERP positivity and was associated with a later enhanced negativity across posterior sites. Old word distractors did not differ based on old/new status.

Figure 1. 

Grand-averaged ERPs and scalp topographies of old/new effects for targets and distractors in both experiments. (A) ERPs from midfrontal (Fz, top row) and left parietal (P3, bottom row) sites in Experiments 1 (left) and 2 (right). (B) Scalp topographies of the mean amplitude old minus new difference for words irrespective of picture memory status (top row) and the old minus new difference for pictures irrespective of word memory status (bottom row) in Experiments 1 (left) and 2 (right). Whereas old targets elicited both typical early (300–500 msec) and late (500–800 msec) positive ERP amplitudes in both experiments, old picture distractors (Experiment 1) only elicited an early ERP positivity and was associated with a later enhanced negativity across posterior sites. Old word distractors did not differ based on old/new status.

Targeted Analysis

FN400 Old/New Effects

In Experiment 1 where words were targets and pictures were distractors, both old target words and old distractor pictures elicited significantly more positive FN400 amplitudes than new target words and new distractor pictures, respectively (2 × 2 ANOVA; main effect of Word memory status: F(1, 23) = 6.54, p = .02, ηp2 = .22; main effect of Picture memory status: F(1, 23) = 28.50, p < .0001, ηp2 = .55), but there was no interaction (F < 1, p = .53).

In Experiment 2 where pictures were targets and words were distractors, old target pictures elicited significantly more positive FN400 amplitudes than new target pictures (2 × 2 ANOVA; Picture memory status: F(1, 23) = 16.78, p < .001, ηp2 = .42), but there was no significant old/new effect for distractor words (F(1, 23) = 1.72, p = .20), nor an interaction (F < 1, p = .86).

A three-way ANOVA with Experiment as the third factor revealed significant main effects of Word memory status and Picture memory status, with significantly more positive FN400s for old than new words (F(1, 46) = 8.08, p = .007, ηp2 = .15) and for old than new pictures (F(1, 46) = 39.02, p < .0001, ηp2 = .46) across both experiments. The FN400 was also significantly more positive in Experiment 1 than in Experiment 2 (F(1, 46) = 6.26, p = .016, ηp2 = .12). However, there were no significant two- or three-way interactions (Word memory status × Experiment: F(1, 46) = 1.74, p = .19, ηp2 = .04; all other Fs < 1, ps > .43).

Parietal Old/New Effects

In contrast to the FN400, a typical increased parietal positivity for old compared with new items was only found for word targets in Experiment 1 (2 × 2 ANOVA; Word memory status: F(1, 23) = 31.11, p < .0001, ηp2 = .58). Distractor pictures in fact showed a difference in the opposite direction, with significantly more negative parietal amplitudes for old compared with new distractors (F(1, 23) = 5.94, p = .02, ηp2 = .21). There was no interaction between Word and Picture memory status (F < 1, p = .48).

When pictures were targets in Experiment 2, however, old target pictures did elicit significantly more positive parietal ERPs than new target pictures (F(1, 23) = 20.56, p < .001, ηp2 = .47), and there was also a nonsignificant trend for distractor words in the same direction with more positive ERPs for old than new distractors (F(1, 23) = 3.62, p = .07, ηp2 = .14). Again, there was no interaction between these factors (F < 1, p = .62).

A three-way ANOVA with Experiment as the third factor confirmed that both the word and picture parietal old/new effects were qualitatively different across the two experiments, as indicated by significant interactions between Word memory status and Experiment (F(1, 46) = 11.75, p = .001, ηp2 = .20) and Picture memory status and Experiment (F(1, 46) = 26.48, p < .0001, ηp2 = .37). The two-way interaction between Word and Picture memory status and the three-way interaction were not significant (both Fs < 1, ps > .40).

FN400 and Parietal Old/New Effects Differences Dependent on Target Status

To confirm that the FN400 and parietal old/new effects were qualitatively different across experiments, we calculated old minus new difference measures for targets and distractors for both effects (average difference between 300 and 500 msec at Fz for the FN400 and average difference between 500 and 800 msec at P3 for the left parietal effect), as displayed in Figure 2. These difference measures were analyzed in a 2 × 2 × 2 mixed ANOVA with factors Stimulus type (word/picture), ERP effect (FN400/parietal old/new effect), and Experiment (1/2). The ANOVA confirmed a significant three-way interaction (F(1, 46) = 17.91, p < .001, ηp2 = .28), which was followed up with separate Stimulus type × ERP effect ANOVAs within each experiment.

Figure 2. 

Mean FN400 and left parietal old minus new ERP differences for words and pictures in both experiments. Error bars represent 1 SEM.

Figure 2. 

Mean FN400 and left parietal old minus new ERP differences for words and pictures in both experiments. Error bars represent 1 SEM.

In Experiment 1, the FN400 and parietal old/new effects showed a qualitatively different pattern for the two stimulus types (interaction: F(1, 23) = 37.23, p < .00001, ηp2 = .62). For target words, the parietal old–new difference was significantly larger than the FN400 old/new effect (t(23) = 2.22, p = .04, d = 0.49), whereas for distractor pictures, the parietal old–new difference was significantly smaller (in fact negative) than the FN400 old–new difference (t(23) = 5.97, p < .00001, d = 1.62).

In Experiment 2, there was only a significant main effect of Stimulus type (F(1, 23) = 11.49, p = .003, ηp2 = .33) whereby the old–new difference was significantly larger on average for pictures (M = 3.71 μV, SEM = 0.78) than words (M = 0.67 μV, SEM = 0.32), and there was no interaction with ERP effect (F < 1, p = .96).

There were no significant correlations between the size of the ERP FN400 or parietal old/new effects and individual differences in WM accuracy or between the ERP effects and the size of the congruency accuracy effect (congruent minus incongruent conditions) on the recognition task.

In summary, the targeted ERP analysis revealed qualitatively different old/new effects depending on the target/distractor status of words and pictures. When participants were asked to recognize words and ignore pictures in Experiment 1, both target words and distractor pictures were associated with typical early FN400 old/new effects, but only target words were associated with a later parietal old/new effect, whereas distractor pictures showed a reversal of typical amplitudes with more negative parietal ERPs for old compared with new distractors.

In contrast, when participants were asked to recognize pictures and ignore words in Experiment 2, this reversal in amplitude for distractors between the FN400 and the parietal old/new effect was no longer present. Target pictures were associated with very large FN400 and parietal old/new effects, but word distractors did not elicit significant FN400 or parietal old/new effects (although the latter was a numerical trend in the same direction as for targets).

Whole-head PLS Results

In Experiment 1, the nonrotated task-PLS analysis found significant effects of both Distractor picture old/new status (p = .011, accounting for 30% of cross-block covariance) and Target word old/new status (p < .001, 58% of cross-block covariance), but no interaction (p = .218, 12% of cross-block covariance). The ratios of electrode salience to bootstrapped standard error for significant contrasts are shown in Figure 3, where it can be seen that the whole-head exploratory PLS analysis revealed similar findings to the targeted analysis. In Experiment 1, target old words elicited reliably more positive ERPs than new words, and this effect peaked between about 400 and 700 msec onward with a central and left parietal distribution. Distractor old pictures elicited reliably more positive ERPs than new pictures across frontal and central sites between around 300 and 500 msec, which was followed by a sustained negativity (maximal ∼500–1000 msec poststimulus) for old compared with new pictures across parietal and occipital sites.

Figure 3. 

Topographic distributions of the ratios of electrode salience to bootstrapped standard error for significant contrasts in the whole-head PLS analysis. The bootstrap ratios are approximately equivalent to z scores; values > 1.96 indicate electrodes and time points that reliably show more positive ERP amplitude for old than new items, and values < −1.96 indicate electrodes and time points that reliably show more negative ERP amplitude for old than new items.

Figure 3. 

Topographic distributions of the ratios of electrode salience to bootstrapped standard error for significant contrasts in the whole-head PLS analysis. The bootstrap ratios are approximately equivalent to z scores; values > 1.96 indicate electrodes and time points that reliably show more positive ERP amplitude for old than new items, and values < −1.96 indicate electrodes and time points that reliably show more negative ERP amplitude for old than new items.

In Experiment 2, only the effect of Target picture old/new status was significant (p < .001, 89% of cross-block covariance), and there was no effect of Distractor word old/new status (p = .255, 8% of cross-block covariance), nor was the interaction significant (p = .961, 3% of cross-block covariance). The ratios of electrode salience to bootstrapped standard error for the significant picture main effect contrast are shown in Figure 3, again showing similar effects as the targeted statistical analysis. Target old pictures elicited more positive ERPs than new pictures, and this effect was highly reliable across central and parietal sites, peaking between around 300 and 700 msec. Toward the end of the epoch, the distribution of this positivity had a right frontal distribution. No such right frontal old > new effect was observed for either words or pictures in Experiment 1.

DISCUSSION

We investigated the neurocognitive underpinnings of distraction effects on recognition memory to determine whether intentional recognition of target stimuli and unintentional recognition of distracting stimuli in the same environment would be associated with similar or different underlying brain mechanisms. The results showed that unintentional recognition was associated with a distinct ERP old/new pattern that differed substantially from ERPs during intentional recognition. Unintentional distractor recognition was only associated with the FN400 ERP correlate of familiarity and not the left parietal ERP correlate of recollection, as the latter was uniquely elicited by intentional target recognition. The results thus revealed a clear dissociation between these two well-established ERP markers of recognition processes (e.g., Wilding & Ranganath, 2012; Rugg & Curran, 2007; Rugg et al., 1998), in line with dual-process models that consider familiarity and recollection as functionally and neurally independent retrieval processes (Yonelinas & Jacoby, 2012). The findings are consistent with the view that unintentional recognition is driven by relatively automatic familiarity rather than recollection (Anderson et al., 2011) and with previous evidence that recollection can be voluntarily suppressed when it is unwanted (Hu et al., 2015; Bergström, Anderson, et al., 2013; Bergström et al., 2007, 2009a, 2009b; Hanslmayr et al., 2009; Mecklinger et al., 2009).

As predicted, unintentional recognition of distracting pictures had a biasing effect on the accuracy of participants' word recognition judgments, so that accuracy was highest when the memory status of the picture was congruent rather than incongruent with the word memory status. Word distractors however did not affect the accuracy of picture recognition judgments, in line with previous research that found less consistent effects of word distractors on intentional picture recognition, than vice versa (Anderson et al., 2011). Likewise, ERP evidence of distractor recognition was only found for pictures and not words, suggesting that the difference in bias between word and picture distractors was related to their likelihood of eliciting unintentional recognition (i.e., the actual recovery of stored memory information), rather than the extent to which people engaged postretrieval processing to discount unintentional memory signals for words versus pictures (see Rugg & Wilding, 2000). Consistent with this account, previous research has found that irrelevant old contexts can sometimes elicit familiarity-related ERP FN400 effects while participants make recognition judgments about superimposed objects (Tsivilis, Otten, & Rugg, 2001) but that cueing participants to selectively attend to the objects reduces the context effect on ERPs (Ecker, Zimmer, Groh-Bordin, & Mecklinger, 2007). Our findings extend on this prior research by linking the FN400 effect for irrelevant distractors with biased recognition judgments to targets, providing evidence that distractor-induced behavioral biases are driven by familiarity-related brain processes.

Pictures may have been more likely to elicit unintentional recognition than words in our study because they were perceptually more salient and thus more likely to attract attention. However, other lines of research suggest that word processing is often highly automatic and can interfere greatly with participants' ability to make accurate judgments, for example, in the classic Stroop task where word reading interferes with color naming (Stroop, 1935). One possibility for this discrepancy may be that, in our experiments, participants conducted a verbal WM task that may have interfered more with their word processing than their picture processing (cf. Fernandes & Moscovitch, 2000), thus making unintentional recognition of word distractors less likely. However, previous research using a very similar paradigm to ours (Anderson et al., 2011) also found larger, more consistent effects of picture than word distractors even without a simultaneous divided attention task, suggesting that the nature of our WM task cannot be the sole reason for the difference between material types. Future research should clarify the factors that determine whether a stimulus elicits unintentional recognition, which likely includes perceptual salience as well as other factors such as memorability and distinctiveness (see Anderson et al., 2011, and Ste-Marie & Jacoby, 1993, for further discussion). It is also important to assess whether unintentional recognition is sensitive to domain overlap with concurrent, task-related processing (Fernandes & Moscovitch, 2000).

When old pictures were intentionally recognized as targets, they showed a typical pattern with both early fronto-central and late parietal positive ERP amplitudes compared with new target pictures. However, when the same pictures were unintentionally recognized as distractors, they still elicited a large initial early positivity, but later ERPs across parietal and occipital electrode sites were reversed so that old distractor pictures showed more negative ERP amplitudes than new distractor pictures. Left parietal amplitudes in memory tests are typically positively correlated with the amount of information that is recollected (Vilberg, Moosavi, & Rugg, 2006; Wilding, 2000), but it is unlikely that new distractors would elicit more recollection than old distractors. Furthermore, the PLS analysis showed that the topography of the distractor old < new effect was more posterior than the target old > new effect. Instead, the enhanced negativity for old distractors likely corresponds to the late posterior negativity that originates in the precuneus (Bergström, Henson, Taylor, & Simons, 2013). The late posterior negativity is thought to index evaluation of retrieved information or monitoring of responses (Johansson & Mecklinger, 2003). In the current paradigm, such retrieval monitoring processes may be recruited when old items trigger automatic familiarity to determine the source of the memory signal. Thus, early unintentional recognition may be followed by later, intentionally engaged monitoring that enables people to counteract the biasing influence of distracting recognition (cf. Hu et al., 2015).

Together with previous research, our findings have implications for eyewitness memory tests where recognition of a suspect is tested in the context of multiple distractors that may elicit feelings of familiarity and thereby bias recognition responses to the suspect. For example, standard lineups typically present the suspect simultaneously with other “filler” nonsuspects (cf. Steblay, Dysart, & Wells, 2011), and some types of facial composite creation systems involve asking the eyewitness to recognize the face that most resembles the suspect among several alternatives (Solomon, Gibson, & Mist, 2013; Frowd, 2012). Although our study did not investigate face recognition biases because of distractor familiarity, other studies have shown that presenting unfamiliar faces on a familiar background can cause people to falsely attribute memory signals from the background to the face (Deffler, Brown, & Marsh, 2014; Gruppuso, Lindsay, & Masson, 2007). Similar misattributions of familiarity may also occur between multiple faces that are simultaneously presented (Bower & Karlin, 1974), and such biases may be more likely in populations that are particularly susceptible to distraction, such as those with impaired attentional control (Engle, 2002), which likely includes older adults (Anderson et al., 2011; Campbell, Hasher, & Thomas, 2010; De Fockert, Ramchurn, Van Velzen, Bergström, & Bunce, 2009). Future research should determine the extent of recognition biases in eyewitness memory tests.

In conclusion, our findings show that unintentional and intentional recognition are dissociated by the well-established ERP correlates of familiarity and recollection. Unintentional recognition is driven by a rapid automatic familiarity process, whereas intentional recognition also involves a slower recollection process that is under voluntary control. These results are relevant to how recognition memory works in real-world environments where we are surrounded with multiple stimuli that range in familiarity. Distraction-induced biases may render recognition memory in the real world less accurate than implied by typical experiments in the scientific literature.

Reprint requests should be sent to Zara M. Bergström, School of Psychology, Keynes College, University of Kent, Canterbury, Kent CT2 7NP, United Kingdom, or via e-mail: z.m.bergstrom@kent.ac.uk.

REFERENCES

REFERENCES
Anderson
,
B. A.
,
Jacoby
,
L. L.
,
Thomas
,
R. C.
, &
Balota
,
D. A.
(
2011
).
The effects of age and divided attention on spontaneous recognition
.
Memory & Cognition
,
39
,
725
735
.
Bergström
,
Z. M.
,
Anderson
,
M. C.
,
Buda
,
M.
,
Simons
,
J. S.
, &
Richardson-Klavehn
,
A.
(
2013
).
Intentional retrieval suppression can conceal guilty knowledge in ERP memory detection tests
.
Biological Psychology
,
94
,
1
11
.
Bergström
,
Z. M.
,
De Fockert
,
J.
, &
Richardson-Klavehn
,
A.
(
2009a
).
Event-related potential evidence that automatic recollection can be voluntarily avoided
.
Journal of Cognitive Neuroscience
,
21
,
1280
1301
.
Bergström
,
Z. M.
,
De Fockert
,
J.
, &
Richardson-Klavehn
,
A.
(
2009b
).
ERP and behavioural evidence for direct suppression of unwanted memories
.
Neuroimage
,
21
,
1280
1301
.
Bergström
,
Z. M.
,
Henson
,
R. N.
,
Taylor
,
J. T.
, &
Simons
,
J. S.
(
2013
).
Multimodal imaging reveals the spatiotemporal dynamics of recollection
.
Neuroimage
,
68
,
141
153
.
Bergström
,
Z. M.
,
Velmans
,
M.
,
De Fockert
,
J.
, &
Richardson-Klavehn
,
A.
(
2007
).
ERP evidence for successful voluntary avoidance of conscious recollection
.
Brain Research
,
1151
,
19
133
.
Bower
,
G. H.
, &
Karlin
,
M. B.
(
1974
).
Depth of processing pictures of faces and recognition memory
.
Journal of Experimental Psychology
,
103
,
751
757
.
Bradley
,
M. M.
, &
Lang
,
P. J.
(
1999
).
Affective norms for English words (ANEW): Stimuli, instruction manual and affective ratings (Technical report C-1)
.
Gainesville, FL
:
The Center for Research in Psychophysiology, University of Florida
.
Campbell
,
K. L.
,
Hasher
,
L.
, &
Thomas
,
R. C.
(
2010
).
Hyperbinding: A unique age effect
.
Psychological Science
,
21
,
399
405
.
Curran
,
T.
(
2000
).
Brain potentials of recollection and familiarity
.
Memory & Cognition
,
28
,
923
938
.
Dan-Glauser
,
E. S.
, &
Scherer
,
K. R.
(
2011
).
The Geneva affective picture database (GAPED): A new 730-picture database focusing on valence and normative significance
.
Behavior Research Methods
,
43
,
468
477
.
De Fockert
,
J. W.
,
Ramchurn
,
A.
,
Van Velzen
,
J.
,
Bergström
,
Z. M.
, &
Bunce
,
D.
(
2009
).
Behavioral and ERP evidence of greater distractor processing in old age
.
Brain Research
,
1282
,
67
73
.
De Fockert
,
J. W.
,
Rees
,
G.
,
Frith
,
C.
, &
Lavie
,
N.
(
2001
).
The role of working memory in visual selective attention
.
Science
,
291
,
1803
1806
.
Deffler
,
S. A.
,
Brown
,
A. S.
, &
Marsh
,
E. J.
(
2014
).
Judging the familiarity of strangers: Does the context matter?
Psychonomic Bulletin and Review
,
22
,
1041
1047
.
Delorme
,
A.
, &
Makeig
,
S.
(
2004
).
EEGLAB: An open source toolbox for analysis of singletrial EEG dynamics including independent component analysis
.
Journal of Neuroscience Methods
,
134
,
9
21
.
Dunlap
,
W. P.
,
Cortina
,
J. M.
,
Vaslow
,
J. B.
, &
Burke
,
M. J.
(
1996
).
Meta-analysis of experiments with matched groups or repeated measures designs
.
Psychological Methods
,
1
,
170
177
.
Ecker
,
U. K. H.
,
Zimmer
,
H. D.
,
Groh-Bordin
,
C.
, &
Mecklinger
,
A.
(
2007
).
Context effects on familiarity are familiarity effects of context—An electrophysiological study
.
International Journal of Psychophysiology
,
64
,
146
156
.
Engle
,
R. W.
(
2002
).
Working memory capacity as executive attention
.
Current Directions in Psychological Science
,
11
,
19
23
.
Fernandes
,
M. A.
, &
Moscovitch
,
M.
(
2000
).
Divided attention and memory: Evidence of substantial interference effects at retrieval and encoding
.
Journal of Experimental Psychology: General
,
129
,
155
176
.
Frowd
,
C. D.
(
2012
).
Facial recall and computer composites
. In
C.
Wilkinson
&
C.
Rynn
(Eds.),
Craniofacial identification
.
Cambridge
:
Cambridge University Press
.
Gazzaley
,
A.
, &
Nobre
,
A. C.
(
2012
).
Top-down modulation: Bridging selective attention and working memory
.
Trends in Cognitive Sciences
,
16
,
129
135
.
Gruppuso
,
V.
,
Lindsay
,
D. S.
, &
Masson
,
M. E.
(
2007
).
I'd know that face anywhere!
Psychonomic Bulletin and Review
,
14
,
1085
1089
.
Hanslmayr
,
S.
,
Leipold
,
P.
,
Pastötter
,
B.
, &
Bäuml
,
K.-H.
(
2009
).
Anticipatory signatures of voluntary memory suppression
.
Journal of Neuroscience
,
29
,
2742
2747
.
Hayama
,
H. R.
,
Johnson
,
J. D.
, &
Rugg
,
M. D.
(
2008
).
The relationship between the right frontal old/new ERP effect and post-retrieval monitoring: Specific or non-specific?
Neuropsychologia
,
46
,
1211
1223
.
Hockley
,
W. E.
,
Bancroft
,
T. D.
, &
Bryant
,
E.
(
2012
).
Associative and familiarity-based effects of environmental context on memory
.
Canadian Journal of Experimental Psychology
,
66
,
81
89
.
Hu
,
X.
,
Bergström
,
Z. M.
,
Bodenhausen
,
G. V.
, &
Rosenfeld
,
J. P.
(
2015
).
Suppressing unwanted autobiographical memories reduces their automatic influences: Evidence from electrophysiology and an implicit autobiographical memory test
.
Psychological Science
,
26
,
1098
1106
.
Johansson
,
M.
, &
Mecklinger
,
A.
(
2003
).
The late posterior negativity in ERP studies of episodic memory: Action monitoring and retrieval of attribute conjunctions
.
Biological Psychology
,
64
,
91
117
.
Lang
,
P. J.
,
Bradley
,
M. M.
, &
Cuthbert
,
B. N.
(
2008
).
International affective picture system (IAPS): Affective ratings of pictures and instruction manual
.
Technical Report A-8
.
Gainesville, FL
:
University of Florida
.
Mandler
,
G.
(
1980
).
Recognizing: The judgment of previous occurrence
.
Psychological Review
,
87
,
252
271
.
McIntosh
,
A. R.
, &
Lobaugh
,
N. J.
(
2004
).
Partial least squares analysis of neuroimaging data: Applications and advances
.
Neuroimage
,
23(Suppl. 1)
,
S250
S263
.
Mecklinger
,
A.
,
Parra
,
M.
, &
Waldhauser
,
G. T.
(
2009
).
ERP correlates of intentional forgetting
.
Brain Research
,
1255
,
132
147
.
Murnane
,
K.
,
Phelps
,
M. P.
, &
Malmberg
,
K.
(
1999
).
Context-dependent recognition memory: The ICE theory
.
Journal of Experimental Psychology: General
,
128
,
403
415
.
Paller
,
K. A.
,
Voss
,
J. L.
, &
Boehm
,
S. G.
(
2007
).
Validating neural correlates of familiarity
.
Trends in Cognitive Sciences
,
11
,
243
250
.
Rugg
,
M. D.
, &
Curran
,
T.
(
2007
).
Event-related potentials and recognition memory
.
Trends in Cognitive Sciences
,
11
,
251
257
.
Rugg
,
M. D.
,
Mark
,
R. E.
,
Walla
,
P.
,
Schloerscheidt
,
A. M.
,
Birch
,
C. S.
, &
Allen
,
K
(
1998
).
Dissociation of the neural correlates of implicit and explicit memory
.
Nature
,
392
,
595
598
.
Rugg
,
M. D.
, &
Wilding
,
E. L.
(
2000
).
Retrieval processing and episodic memory
.
Trends in Cognitive Sciences
,
4
,
108
115
.
Snodgrass
,
J. G.
, &
Corwin
,
J.
(
1988
).
Pragmatics of measuring recognition memory: Applications to dementia and amnesia
.
Journal of Experimental Psychology: General
,
117
,
34
.
Solomon
,
C. J.
,
Gibson
,
S. J.
, &
Mist
,
J. J.
(
2013
).
Interactive evolutionary generation of facial composites for locating suspects in criminal investigations
.
Applied Soft Computing
,
13
,
3298
3306
.
Steblay
,
N. K.
,
Dysart
,
J. E.
, &
Wells
,
G. L.
(
2011
).
Seventy-two tests of the sequential lineup superiority effect: A meta-analysis and policy discussion
.
Psychology, Public Policy, and Law
,
17
,
99
139
.
Ste-Marie
,
D. M.
, &
Jacoby
,
L. L.
(
1993
).
Spontaneous versus directed recognition: The relativity of automaticity
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
19
,
777
788
.
Stroop
,
J. R.
(
1935
).
Studies of interference in serial verbal reactions
.
Journal of Experimental Psychology
,
18
,
643
662
.
Tsivilis
,
D.
,
Otten
,
L. J.
, &
Rugg
,
M. D.
(
2001
).
Context effects on the neural correlates of recognition memory: An electrophysiological study
.
Neuron
,
31
,
497
505
.
Vilberg
,
K. L.
,
Moosavi
,
R. F.
, &
Rugg
,
M. D.
(
2006
).
The relationship between electrophysiological correlates of recollection and amount of information retrieved
.
Brain Research
,
1122
,
161
170
.
Wilding
,
E. L.
(
2000
).
In what way does the parietal ERP old/new effect index recollection?
International Journal of Psychophysiology
,
35
,
81
87
.
Wilding
,
E. L.
, &
Ranganath
,
C.
(
2012
).
Electrophysiological correlates of episodic memory processes
. In
S.
Luck
& E. Kappenman
(Eds.),
The Oxford handbook of event-related potential components
(pp.
373
396
).
Oxford, UK
:
Oxford University Press
.
Yonelinas
,
A. P.
, &
Jacoby
,
L. J.
(
2012
).
The process-dissociation approach two decades later: Convergence, boundary conditions, and new directions
.
Memory & Cognition
,
40
,
663
680
.